def steady_state_crossover(self):
self.offspring = list(map(self.toolbox.clone, self.pop))
lista1 = list(self.offspring)
children = []
parents = tools.selRandom(self.offspring, self.ss_gap)
for parent1, parent2 in zip(parents[::2], parents[1::2]):
child1, child2 = self.toolbox.clone(parent1), self.toolbox.clone(
parent2)
self.toolbox.mate(child1, child2)
if random.random() < self.mutation_rate:
self.toolbox.mutate(child1)
if random.random() < self.mutation_rate:
self.toolbox.mutate(child2)
del child1.fitness.values
del child2.fitness.values
children.append(child1)
children.append(child2)
for child in children:
self.offspring.remove(tools.selRandom(self.offspring, 1)[0])
self.offspring.append(child)
def selection(individuals, k, tournsize, alpha, fit_attr="fitness"):
chosen = []
N = int(k*alpha)
for i in range(N):
aspirants = tools.selRandom(individuals, tournsize)
chosen.append(max(aspirants, key=attrgetter(fit_attr)))
new = tools.selRandom(individuals,k-N)
return chosen+new
def Tournament(individuals, finalselect, tournsize):
chosen = []
for i in range(finalselect):
aspirants = tools.selRandom(individuals, tournsize)
chosen.append(max(individuals,
key=selectCriteria))
return chosen
def _selMetaTournament(individuals,
k,
tournsize,
meta_model,
primitives_to_hash_dic,
gptree,
pset,
df,
le,
max_pipeline_size,
fit_attr="fitness"):
"""Select the best individual among *tournsize* randomly chosen
individuals, *k* times. The list returned contains
references to the input *individuals*.
:param individuals: A list of individuals to select from.
:param k: The number of individuals to select.
:param tournsize: The number of individuals participating in each tournament.
:param fit_attr: The attribute of individuals to use as selection criterion
:returns: A list of selected individuals.
This function uses the :func:`~random.choice` function from the python base
:mod:`random` module.
"""
chosen = []
for i in range(k):
aspirants = selRandom(individuals, tournsize)
test_df = create_ranking_df_from_pop(aspirants, primitives_to_hash_dic,
gptree, pset, df,
max_pipeline_size)
top_offspring_index = rank_pop(meta_model, test_df, 1, le)
top_offspring = [aspirants[i] for i in top_offspring_index]
chosen.append(top_offspring[0])
return chosen
def mutate_nodes(self, pop = None):
"""
Mutates the nodes - fnParameters, including bias and bias weight
"""
#TODO: consider mutating the prodCosntant [EXPERIMENT]
if verbose:
print ">> Mutation -> Nodes";
if pop == None:
pop = self.pop['node_pop'];
if random.random() < self.params['prob_mutation']:
ind = tools.selRandom(pop,1).pop();
mutant = self.tbox.clone(ind);
self.tbox.mutate_fnParams(mutant);
mutant.id = None;
del mutant.fitness.values;
if verbose:
print "parent", ind;
print "child", mutant;
return mutant;
else:
return None;
def mutate_models(self, pop = None):
"""
Mutates the models - {lateral connection, gaussian mutation of inputs, context layer}
"""
if verbose:
print ">> Mutation -> Models";
if pop == None:
pop = self.pop['model_pop'];
if random.random() < self.params['prob_mutation']:
ind = tools.selRandom(pop,1).pop();
mutant = self.tbox.clone(ind);
self.tbox.mutate_conn(mutant);
mutant.id = None;
del mutant.fitness.values;
if verbose:
print "parent", ind;
print "child", mutant;
return mutant;
else:
return None;
def mutate_connHO(self, pop = None):
"""
Mutates the connection components (Hidden to Output layers weights)
"""
if verbose:
print ">> Mutation -> connHO";
if pop == None:
pop = self.pop['connActive_HO_pop'];
if random.random() < self.params['prob_mutation']:
ind = tools.selRandom(pop,1).pop();
mutant = self.tbox.clone(ind);
for i in xrange(len(mutant)):
self.tbox.mutate_conn(mutant[i]);
mutant.id = None;
del mutant.fitness.values;
if verbose:
print "parent", ind;
print "child", mutant;
return mutant;
else:
return None;
def mutate_weightsHO(self, pop = None):
"""
Mutates the weight components (Hidden to Output layers weights)
"""
if verbose:
print ">> Mutation -> WeightHO";
if pop == None:
pop = self.pop['connWeights_HO_pop'];
if random.random() < self.params['prob_mutation']:
ind = tools.selRandom(pop,1).pop();
mutant = self.tbox.clone(ind);
self.tbox.mutate_weights(mutant);
mutant.id = None;
del mutant.fitness.values;
if verbose:
print "parent", ind;
print "child", mutant;
return mutant;
else:
return None;
def batch_tournament_selection(individuals,
k,
tournsize,
batch_size,
fit_attr="fitness"):
fitness_cases_num = len(individuals[0].case_values)
idx_cases_batch = np.arange(0, fitness_cases_num)
np.random.shuffle(idx_cases_batch)
_batches = np.array_split(idx_cases_batch,
max(fitness_cases_num // batch_size, 1))
batch_ids = np.arange(0, len(_batches))
assert len(_batches[0]) >= batch_size or fitness_cases_num < batch_size
chosen = []
while len(chosen) < k:
batches: list = copy.deepcopy(batch_ids.tolist())
while len(batches) > 0 and len(chosen) < k:
idx_candidates = selRandom(individuals, tournsize)
cand_fitness_for_this_batch = []
for idx in idx_candidates:
cand_fitness_for_this_batch.append(
np.mean(idx.case_values[_batches[batches[0]]]))
idx_winner = np.argmin(cand_fitness_for_this_batch)
winner = idx_candidates[idx_winner]
chosen.append(winner)
batches.pop(0)
return chosen
def get_subset(self):
if self.population:
# random selection
#sample = tools.selRandom(self.population, self.subset_size)
# best selection
#sample = tools.selBest(self.population, self.subset_size)
# worst selection
#sample = tools.selWorst(self.population, self.subset_size)
# bestish selection
sample = tools.selTournament(self.population, self.subset_size, 3)
if self.best:
sample[0] = self.best
# if not unique, sample one more time
sample_strings = [ind.__str__() for ind in sample]
if len(set(sample_strings)) < len(sample_strings):
for i, ind_i in enumerate(sample):
for j, ind_j in enumerate(sample[i + 1:]):
if ind_i == ind_j:
sample[i + 1 + j] = tools.selRandom(
self.population, 1)[0]
print 'sample', sample
self.subset = sample
sample = self.pre_process(sample)
else:
sample = self.get_default()
return sample
def selTournament(self, individuals, k, tournsize, fit_attr="fitness"): chosen = [] for i in xrange(k): aspirants = tools.selRandom(individuals, tournsize) best = self.utils.getBest(aspirants) chosen.append(best) return chosen
def _evolve(toolbox, optimizer, seed, gen=0, mu=1, lambda_=1, cxpb=1, mutp=1):
random.seed(seed)
np.random.seed(seed)
pop = remove_twins(toolbox.population(n=mu))
pop = list(toolbox.map(optimizer, pop))
invalid_ind = [ind for ind in pop if not ind.fitness.valid]
fitness = list(toolbox.map(toolbox.evaluate, invalid_ind))
for ind, fit in zip(invalid_ind, fitness):
ind.fitness.values = fit
pop = toolbox.select(pop, mu)
stats = tools.Statistics(lambda ind: ind.fitness.values)
stats.register("min", np.nanmin, axis=0)
stats.register("max", np.nanmax, axis=0)
stats.register("diversity", lambda pop: len(set(map(str, pop))))
logbook = tools.Logbook()
logbook.header = "gen", 'evals', 'min', 'max', 'diversity'
record = stats.compile(pop)
logbook.record(gen=0, evals=(len(invalid_ind)), **record)
print(logbook.stream)
if record['min'][0] == 0.0:
return pop, logbook
for g in range(1, gen):
offspring = tools.selRandom(pop, mu)
offspring = [toolbox.clone(ind) for ind in offspring]
# mate
for child1, child2 in zip(offspring[::2], offspring[1::2]):
if random.random() <= cxpb:
toolbox.mate(child1, child2)
del child1.fitness.values
del child2.fitness.values
# mutate
for mutant in offspring:
if random.random() <= mutp:
toolbox.mutate(mutant)
del mutant.fitness.values
# re-evaluate
invalid_ind = [ind for ind in offspring if not ind.fitness.valid]
fitness = list(toolbox.map(toolbox.evaluate, invalid_ind))
for ind, fit in zip(invalid_ind, fitness):
ind.fitness.values = fit
# select
pop = toolbox.select(remove_twins(pop + offspring), mu)
record = stats.compile(pop)
logbook.record(gen=g, evals=len(invalid_ind), **record)
print(logbook.stream)
if record['min'][0] < 1E-4:
break
return pop, logbook
def _evolve(toolbox, optimizer, seed, gen=0, mu=1, lambda_=1, cxpb=1, mutp=1):
random.seed(seed)
np.random.seed(seed)
pop = remove_twins(toolbox.population(n=mu))
pop = list(toolbox.map(optimizer, pop))
invalid_ind = [ind for ind in pop if not ind.fitness.valid]
fitness = list(toolbox.map(toolbox.evaluate, invalid_ind))
for ind, fit in zip(invalid_ind, fitness):
ind.fitness.values = fit
pop = toolbox.select(pop, mu)
stats = tools.Statistics(lambda ind: ind.fitness.values)
stats.register("min", np.nanmin, axis=0)
stats.register("max", np.nanmax, axis=0)
stats.register("diversity", lambda pop: len(set(map(str, pop))))
logbook = tools.Logbook()
logbook.header = "gen", 'evals', 'min', 'max', 'diversity'
record = stats.compile(pop)
logbook.record(gen=0, evals=(len(invalid_ind)), **record)
print(logbook.stream)
if record['min'][0] == 0.0:
return pop, logbook
for g in range(1, gen):
offspring = tools.selRandom(pop, mu)
offspring = [toolbox.clone(ind) for ind in offspring]
# mate
for child1, child2 in zip(offspring[::2], offspring[1::2]):
if random.random() <= cxpb:
toolbox.mate(child1, child2)
del child1.fitness.values
del child2.fitness.values
# mutate
for mutant in offspring:
if random.random() <= mutp:
toolbox.mutate(mutant)
del mutant.fitness.values
# re-evaluate
invalid_ind = [ind for ind in offspring if not ind.fitness.valid]
fitness = list(toolbox.map(toolbox.evaluate, invalid_ind))
for ind, fit in zip(invalid_ind, fitness):
ind.fitness.values = fit
# select
pop = toolbox.select(remove_twins(pop + offspring), mu)
record = stats.compile(pop)
logbook.record(gen=g, evals=len(invalid_ind), **record)
print(logbook.stream)
if record['min'][0] < 1E-4:
break
return pop, logbook
def selTournamentPlus(individuals, k, tournsize):
"""
Select individuals based on the sum of case values
:param individuals: population
:param k: number of offspring
:param tournsize: tournament size
:return:
"""
chosen = []
for i in range(k):
aspirants = selRandom(individuals, tournsize)
chosen.append(min(aspirants, key=lambda x: np.sum(x.case_values)))
return chosen
def selCustom(individuals, k):
#using DEAP's sortNondominated, we get a list of fronts,
#where each front 'i' dominates front 'i+1'.
#we use this to create a new attribute for each individual, called "rank"
#the rank is then used as the fitness value for the tournament selection,
#as specified by Ombuki et. al
pareto_fronts = tools.sortNondominated(individuals, k)
for front_rank in range(1, len(pareto_fronts) + 1):
front = pareto_fronts[front_rank - 1]
for ind in front:
setattr(ind, 'rank', front_rank)
#the first rank is the "elite" (Pareto-optimal) set of solutions
#to which we want to guarantee a spot in the next generation
#therefore, we extract them before the tournament selection takes place
elite = pareto_fronts.pop(0)
individuals_excluding_elite = [i for i in individuals if i not in elite]
#we update k, the number of individuals to be chosen by the tournament selection
k -= len(elite)
#as specified by the paper
tournsize = 4
r_thresh = 0.8
chosen = []
for i in range(k):
aspirants = tools.selRandom(individuals_excluding_elite, tournsize)
if random.random() < r_thresh:
chosen_individual = min(aspirants, key=attrgetter("rank"))
else:
chosen_individual = tools.selRandom(aspirants, 1)[0]
chosen.append(chosen_individual)
#add in the elite solutions
chosen += elite
return chosen
def selTournament(individuals, k, tournsize, noise_factor = 0):
# add noise if necessary
if noise_factor > 0:
# add some noise to the fitness values
pop_std = np.std([ind.fitness.values[0] for ind in individuals])
if (pop_std == 0):
pop_std = 1
noise = np.random.normal(0, np.sqrt(noise_factor) * pop_std,
len(individuals))
noisy_individuals = []
for index, ind in enumerate(individuals):
noisy_individuals.append((ind,
ind.fitness.values[0] + noise[index]))
chosen = []
for i in range(k):
aspirants = tools.selRandom(individuals, tournsize)
indices = [individuals.index(j) for j in aspirants]
best = min([noisy_individuals[j] for j in indices],
key = lambda x:x[1])
chosen.append(best[0])
return chosen
else:
# normal tournament selection
chosen = []
for i in range(k):
aspirants = tools.selRandom(individuals, tournsize)
chosen.append(max(aspirants, key=attrgetter("fitness")))
return chosen
def improveTournament(individuals, k, tournsize, fit_attr="fitness"):
for i in xrange(k):
chosen = []
chosenTemp=[]
totalCountersTemp=9999999
aspirants = tools.selRandom(individuals, tournsize)
for aspira in aspirants:
totalCounters = checkCounters(aspira)
#os._exit(1)
if totalCounters <= totalCountersTemp:
chosenTemp = aspira
totalCountersTemp=totalCounters
chosen.append(chosenTemp)
return chosen
def steady_state_crossover(self): self.offspring = list(map(self.toolbox.clone, self.pop)) lista1 = list(self.offspring) children = [] parents = tools.selRandom(self.offspring, self.ss_gap) for parent1, parent2 in zip(parents[::2], parents[1::2]): child1, child2 = self.toolbox.clone(parent1), self.toolbox.clone(parent2) self.toolbox.mate(child1, child2) if random.random() < self.mutation_rate: self.toolbox.mutate(child1) if random.random() < self.mutation_rate: self.toolbox.mutate(child2) del child1.fitness.values del child2.fitness.values children.append(child1) children.append(child2) for child in children: self.offspring.remove(tools.selRandom(self.offspring, 1)[0]) self.offspring.append(child)
def selTournamentRemove(individuals, k, tournsize):
"""Select *k* individuals from the input *individuals* using *k*
tournaments of *tournsize* individuals and remove them from the initial list.
The list returned contains references to the input *individuals*.
:param individuals: A list of individuals to select from.
:param k: The number of individuals to select.
:param tournsize: The number of individuals participating in each tournament.
:returns: A list of selected individuals.
This function uses the :func:`~random.choice` function from the python base
:mod:`random` module.
"""
chosen = []
for i in xrange(k):
aspirants = tools.selRandom(individuals, tournsize)
chosen.append(max(aspirants, key=attrgetter("fitness")))
individuals.remove(max(aspirants, key=attrgetter("fitness")))
return chosen
def replace_rand_by_new_inds(self, pop, num):
if MPI_RANK != 0: return None
r_pop = tools.selRandom(pop, num)
nd = self.get_nondominated_inds(pop)
nd_num = len(nd)
n_pop_f = int(self.n_pop * 0.3)
for ppp in r_pop:
new_ind = self.toolbox.individual()
if nd_num < n_pop_f and ppp in nd:
continue
for j in range(len(ppp)):
ppp[j] = new_ind[j]
del ppp.fitness.values
return pop
def selTournament(individuals, k, tournsize, fit_attr="fitness"):
# reproducing selTournament here because I can't find a way to get
# algorithms.selTournament to take more than one fitness value
# into account
chosen = []
for i in xrange(k):
aspirants = tools.selRandom(individuals, tournsize)
# get DEAP's best
best = max(aspirants, key=attrgetter(fit_attr))
bestFitness = calculateFitness(best.fitness.getValues())
# check for individuals with a better fitness once depth penalty has been applied
for individual in aspirants:
thisFitness = calculateFitness(individual.fitness.getValues())
if (thisFitness > bestFitness):
best = individual
bestFitness = thisFitness
chosen.append(best)
return chosen
#cc1 cc2 is totally new
cc1, cc2 = toolbox.mateG(c1, c2)
cxOut.append(cc1)
cxOut.append(cc2)
# mutation
mutantOut = []
for mutant in offspring:
if random.random() < GMUTPB:
mut = toolbox.mutateG(mutant, dest)
if mut != None:
mutantOut.append(mut)
#compose a big group of population
#bigPop = pop + mutantOut + cxOut
bigPop = mutantOut + cxOut + lPop
gPop = tools.selRandom(bigPop, GPOPU)
""" L part !!!"""
# select
#offspring = toolbox.selectL(lPop, 2)
offspring = tools.selBest(lPop, LPOPU)
# deep copy out
# offspring = list(map(toolbox.clone, offspring))
# cross over
cxOut = []
for c1 in offspring:
if random.random() < LCXPB:
#cc1 cc2 is totally new
#random select c2 from gPop
c2 = tools.selRandom(gPop, 1)[0]
cc1, cc2 = toolbox.mateL(c1, c2)
def GAEEII_IVFm(data, dimensions, number_endmembers):
start_time = time.time()
number_rows = int(dimensions[0])
number_columns = int(dimensions[1])
number_bands = int(dimensions[2])
number_pixels = number_rows * number_columns
sigma_MAX = max(number_rows, number_generations)
data_proj = numpy.asarray(affine_projection(data, number_endmembers))
creator.create("max_fitness", base.Fitness, weights=(-1.0, -1.0))
creator.create("individual", list, fitness=creator.max_fitness)
toolbox = base.Toolbox()
toolbox.register("create_individual", generate_individual, creator,
number_endmembers, number_pixels, number_rows,
number_columns)
toolbox.register("initialize_population", tools.initRepeat, list,
toolbox.create_individual)
toolbox.register("evaluate_individual",
multi_fitness,
data=data,
data_proj=data_proj,
number_endmembers=number_endmembers)
toolbox.register("cross_twopoints", tools.cxTwoPoint)
toolbox.register("selNSGA2", tools.selNSGA2)
toolbox.register("gaussian_mutation_op",
tools.mutGaussian,
mu=0,
sigma=0,
indpb=mutation_probability)
toolbox.register("gaussian_mutation",
gaussian_mutation,
toolbox=toolbox,
number_rows=number_rows,
number_columns=number_columns)
toolbox.register("random_mutation",
random_mutation,
number_pixels=number_pixels,
number_endmembers=number_endmembers,
mutation_probability=mutation_probability)
population = toolbox.initialize_population(n=population_size)
population_fitnesses = [
toolbox.evaluate_individual(individual) for individual in population
]
for individual, fitness in zip(population, population_fitnesses):
individual.fitness.values = fitness
hof = tools.HallOfFame(3)
hof.update(population)
current_generation = 0
current_sigma = sigma_MAX
generations_fitness_1 = []
generations_fitness_2 = []
generations_population = []
stop_criteria = deque(maxlen=5)
stop_criteria.extend([1, 2, 3, 4, 5])
stats = tools.Statistics(lambda ind: ind.fitness.values)
stats.register("avg", numpy.mean, axis=0)
stats.register("std", numpy.std, axis=0)
stats.register("min", numpy.min, axis=0)
stats.register("max", numpy.max, axis=0)
logbook = tools.Logbook()
logbook.header = "gen", "evals", "std", "min", "avg", "max"
record = stats.compile(population)
logbook.record(gen=0, evals=len(population), **record)
while current_generation < number_generations and numpy.var(
numpy.array(stop_criteria)) > 0.000001:
toolbox.unregister("gaussian_mutation_op")
toolbox.register("gaussian_mutation_op",
tools.mutGaussian,
mu=0,
sigma=current_sigma,
indpb=mutation_probability)
offspring = tools.selRandom(population, k=int(population_size / 2))
offspring = list(map(toolbox.clone, offspring))
# Crossing
for child_1, child_2 in zip(offspring[::2], offspring[1::2]):
if random.random() < crossing_probability:
toolbox.cross_twopoints(child_1, child_2)
del child_1.fitness.values
del child_2.fitness.values
# Mutation
for mutant in offspring:
if random.random() < mutation_probability:
toolbox.gaussian_mutation(mutant)
del mutant.fitness.values
# Fitness
offspring_fitnesses = [
toolbox.evaluate_individual(individual) for individual in offspring
]
for individual, fitness in zip(offspring, offspring_fitnesses):
individual.fitness.values = fitness
# In Vitro Fertilization Module
ivfoffspring = list(map(toolbox.clone, population))
ivffits = [ind.fitness.values[0] for ind in ivfoffspring]
fatheridx = numpy.argmax(ivffits)
# fatherfit = numpy.max(ivffits)
father = creator.individual(ivfoffspring[fatheridx].copy())
for ind in ivfoffspring[::2]:
toolbox.random_mutation(ind)
del ind.fitness.values
for child1 in ivfoffspring:
child2 = creator.individual(father.copy())
toolbox.cross_twopoints(child1, child2)
del child1.fitness.values
del child2.fitness.values
ivffitnesses = [
toolbox.evaluate_individual(ind) for ind in ivfoffspring
]
for ind, fit in zip(ivfoffspring, ivffitnesses):
ind.fitness.values = fit
popmax = max(offspring_fitnesses)
for ind in ivfoffspring:
if (ind.fitness.values >= popmax):
population.append(ind)
new_population = population + offspring
# Selection
selected = toolbox.selNSGA2(new_population, population_size)
selected = list(map(toolbox.clone, selected))
population[:] = selected
hof.update(population)
record = stats.compile(population)
logbook.record(gen=current_generation, evals=len(population), **record)
# print(logbook.stream)
# Statistics
fits_1 = [ind.fitness.values[0] for ind in population]
fits_2 = [ind.fitness.values[1] for ind in population]
mean_1_offspring = sum(fits_1) / len(population)
mean_2_offspring = sum(fits_2) / len(population)
generations_fitness_1.append(numpy.log10(mean_1_offspring))
generations_fitness_2.append(numpy.log(mean_2_offspring))
stop_criteria.append(numpy.log10(mean_1_offspring))
generations_population.append(population.copy())
current_generation += 1
current_sigma = sigma_MAX / ((current_generation + 1) / 1.5)
best_individual = tools.selNSGA2(population, 1)[0]
# print('Result:', multi_fitness(best_individual,data, data_proj, number_endmembers))
M = data[:, best_individual]
duration = time.time() - start_time
return M, duration, [
generations_fitness_1, generations_fitness_2, generations_population,
current_generation
]
def GAEEII(data, dimensions, number_endmembers):
start_time = time.time()
population_size = 10
number_generations = 100
crossing_probability = 1
mutation_probability = 0.5
stop_criteria_MAX = 20
random.seed(64)
number_endmembers = number_endmembers
number_rows = int(dimensions[0])
number_columns = int(dimensions[1])
number_bands = int(dimensions[2])
number_pixels = number_rows * number_columns
sigma_MAX = max(number_rows, number_generations)
data_proj = numpy.asarray(affine_tranform(data, number_endmembers))
# data = numpy.asarray(data)
# _coeff, score, _latent = princomp(data.T)
# data_proj = numpy.squeeze(score[0:number_endmembers,:])
creator.create("min_fitness", base.Fitness, weights=(1.0, ))
creator.create("individual", list, fitness=creator.min_fitness)
toolbox = base.Toolbox()
toolbox.register("create_individual", generate_individual, creator,
number_endmembers, number_pixels, number_rows,
number_columns)
toolbox.register("initialize_population", tools.initRepeat, list,
toolbox.create_individual)
toolbox.register("evaluate_individual",
multi_fitness,
data=data,
data_proj=data_proj,
number_endmembers=number_endmembers)
toolbox.register("cross_twopoints", tools.cxTwoPoint)
toolbox.register("selNSGA2", tools.selNSGA2)
toolbox.register("gaussian_mutation_op",
tools.mutGaussian,
mu=0,
sigma=0,
indpb=mutation_probability)
toolbox.register("gaussian_mutation",
gaussian_mutation,
toolbox=toolbox,
number_rows=number_rows,
number_columns=number_columns)
population = toolbox.initialize_population(n=population_size)
ensemble_pop = []
for i in range(0, 5):
[_, duration, other] = classical.VCA(data, dimensions,
number_endmembers)
# print(other[0])
# print('Time GAEE:',duration)
ensemble_pop.append(creator.individual(other[0]))
population[5:] = ensemble_pop
population_fitnesses = [
toolbox.evaluate_individual(individual) for individual in population
]
for individual, fitness in zip(population, population_fitnesses):
individual.fitness.values = fitness
hof = tools.HallOfFame(3)
hof.update(population)
current_generation = 0
current_sigma = sigma_MAX
generations_fitness_1 = []
generations_fitness_2 = []
generations_population = []
stop_criteria = deque(maxlen=stop_criteria_MAX)
stop_criteria.extend(list(range(1, stop_criteria_MAX)))
while current_generation < number_generations and numpy.var(
numpy.array(stop_criteria)) > 0.000001:
toolbox.unregister("gaussian_mutation_op")
toolbox.register("gaussian_mutation_op",
tools.mutGaussian,
mu=0,
sigma=current_sigma,
indpb=mutation_probability)
offspring = tools.selRandom(population, k=int(population_size / 2))
offspring = list(map(toolbox.clone, offspring))
# Crossing
for child_1, child_2 in zip(offspring[::2], offspring[1::2]):
if random.random() < crossing_probability:
toolbox.cross_twopoints(child_1, child_2)
del child_1.fitness.values
del child_2.fitness.values
# Mutation
for mutant in offspring:
if random.random() < mutation_probability:
toolbox.gaussian_mutation(mutant)
del mutant.fitness.values
# Fitness
offspring_fitnesses = [
toolbox.evaluate_individual(individual) for individual in offspring
]
for individual, fitness in zip(offspring, offspring_fitnesses):
individual.fitness.values = fitness
new_population = population + offspring
# Selection
selected = toolbox.selNSGA2(new_population, population_size)
selected = list(map(toolbox.clone, selected))
population[:] = selected
hof.update(population)
# Statistics
fits_1 = [ind.fitness.values[0] for ind in population]
# fits_2 = [ind.fitness.values[1] for ind in population]
mean_1_offspring = sum(fits_1) / len(population)
# mean_2_offspring = sum(fits_2) / len(population)
generations_fitness_1.append(numpy.log10(mean_1_offspring))
# generations_fitness_2.append(numpy.log(mean_2_offspring))
generations_population.append(population.copy())
stop_criteria.append(numpy.log10(mean_1_offspring))
current_generation += 1
current_sigma = sigma_MAX / ((current_generation + 1) / 4)
best_individual = tools.selBest(population, 1)[0]
# best_individual = hof[0]
# print('Result:', multi_fitness(best_individual,data, data_proj,number_endmembers))
M = data[:, best_individual]
duration = time.time() - start_time
return M, duration, [
generations_fitness_1, generations_fitness_2, generations_population,
current_generation
]
childs = []
for i in range(DIES_EVERY_DAY - NEW_EVERY_DAY):
a, b = tools.selTournament(survivors,
2,
BREED_TOURN_SIZE,
fit_attr="fitness")
child = breed_brains(a, b)
child.fitness = None
childs.append(child)
for i in range(NEW_EVERY_DAY):
new = vlf.neural_network.FFNeuralNetwork()
new.fitness = None
survivors.append(new)
mutants = tools.selRandom(survivors, MUTATED_EVERY_DAY)
for m in mutants:
mutate_brain(m)
m.fitness = None
population = survivors + childs
for p in population:
if not p.fitness:
p.fitness = get_fitness(p)
f = [p.fitness for p in population]
best = max(population, key=lambda x: x.fitness)
print(best.get_flattened_array())
show(best)
print("GENERATION", g)
def Tournament(individuals, finalselect, tournsize):
chosen = []
for i in range(finalselect):
aspirants = tools.selRandom(individuals, tournsize)
chosen.append(max(individuals, key=selectCriteria))
return chosen
def main(seed=None):
random.seed(seed)
NGEN = 250
MU = 100
CXPB = 0.9
stats = tools.Statistics(lambda ind: ind.fitness.values)
# stats.register("avg", numpy.mean, axis=0)
# stats.register("std", numpy.std, axis=0)
stats.register("min", numpy.min, axis=0)
stats.register("max", numpy.max, axis=0)
logbook = tools.Logbook()
logbook.header = "gen", "evals", "std", "min", "avg", "max"
pop = toolbox.population(n=MU)
# Evaluate the individuals with an invalid fitness
invalid_ind = [ind for ind in pop if not ind.fitness.valid]
fitnesses = toolbox.map(toolbox.evaluate, invalid_ind)
for ind, fit in zip(invalid_ind, fitnesses):
ind.fitness.values = fit
record = stats.compile(pop)
logbook.record(gen=0, evals=len(invalid_ind), **record)
print(logbook.stream)
# Begin the generational process
for gen in range(1, NGEN):
# Vary the population
offspring = tools.selRandom(pop, len(pop))
offspring = [toolbox.clone(ind) for ind in offspring]
for ind1, ind2 in zip(offspring[::2], offspring[1::2]):
if random.random() <= CXPB:
toolbox.mate(ind1, ind2)
toolbox.mutate(ind1)
toolbox.mutate(ind2)
del ind1.fitness.values, ind2.fitness.values
# Evaluate the individuals with an invalid fitness
invalid_ind = [ind for ind in offspring if not ind.fitness.valid]
fitnesses = toolbox.map(toolbox.evaluate, invalid_ind)
for ind, fit in zip(invalid_ind, fitnesses):
ind.fitness.values = fit
# Select the next generation population
pop = pop + offspring
fronts = toolbox.sort(pop, len(pop))
chosen = []
for i, front in enumerate(fronts):
# Move is front to chosen population til it is almost full
if len(chosen) + len(front) <= MU:
chosen.extend(front)
else:
# Assign hypervolume contribution to individuals of front that
# cannot be completely move over to chosen individuals
fitness_hv = hypervolume_contrib(front)
for ind, fit_hv in zip(front, fitness_hv):
ind.fitness_hv.values = (fit_hv, )
# Fill chosen with best indiviuals from inspect front
# (based on hypervolume contribution)
chosen.extend(toolbox.select(front, MU - len(chosen)))
break
pop = chosen
record = stats.compile(pop)
logbook.record(gen=gen, evals=len(invalid_ind), **record)
print(logbook.stream)
print("Final population hypervolume is %f" %
hypervolume(pop, [11.0, 11.0]))
return pop, logbook
def main(epsilon, delta):
global lambda_values
global population_size
global population
global population_obj_func_vals
global iterations
outputs = []
for lambda_index in range(1, number_of_lambdas+1):
_lambda = float(lambda_index-1)/(number_of_lambdas-1)
improved_solutions = [] #improved solutions for a particular lambda
v_lambda = float('Inf')
for i in range(1 , population_size+1):
sample = [[0,0] for x in range(number_of_assets)]
Q = random.sample([x for x in range(1, number_of_assets+1)],10)
for asset in Q:
sample[asset-1] = [1, myRandom(0,1)]
population[i-1] = list(sample)
for ind,S in enumerate(population):
[population_obj_func_vals[ind], v_lambda, improved] = evaluate(S, _lambda, 0, v_lambda, False, improved_solutions, UPDATE_H)
for itr in range(iterations):
#binary tournament for selecting parents S_star and S_double_star
chosen = []
for i in xrange(2):
aspirants = tools.selRandom(population, 40)
for i,aspirant in enumerate(aspirants):
h = []
f_aspirant = 0
v = float('Inf')
imp = False
[f_aspirant, v, imp] = evaluate(aspirant, _lambda, 0, v, False, h, DONT_UPDATE_H)
aspirants[i] = [aspirant, f_aspirant]
chosen.append(min(aspirants, key=lambda x: x[1])[0])
S_star = list(chosen[0])
S_double_star = list(chosen[1])
#Uniform crossover to find child C
C = []
[C1, C2] = tools.cxUniform(S_star, S_double_star, ind_prob)
f1 = f2 = 0
v = float('Inf')
imp = False
h = []
[f1, v, imp] = evaluate(C1, _lambda, 0, v, False, h, DONT_UPDATE_H)
[f2, v, imp] = evaluate(C2, _lambda, 0, v, False, h, DONT_UPDATE_H)
if f1<f2:
C = C1
else:
C = C2
#find assets in parents but not in child
A_star = []
for i in range(number_of_assets):
if S_star[i][0]==1 and S_double_star[i][0]==0 and C[i][0]==0:
A_star.append([i+1, S_star[i][1]])
elif S_star[i][0]==0 and S_double_star[i][0]==1 and C[i][0]==0:
A_star.append([i+1, S_double_star[i][1]])
#Mutation
mutation_index = random.randint(0, number_of_assets-1)
while C[mutation_index][0]==0:
mutation_index = random.randint(0, number_of_assets-1)
m = random.randint(0,1)
if m == 0:
# C_i = 0.9*(epsilon + C[mutation_index][1]) - epsilon
C_i = 0.9*(epsilon[mutation_index] + C[mutation_index][1]) - epsilon[mutation_index]
else:
# C_i = 1.1*(epsilon + C[mutation_index][1]) - epsilon
C_i = 1.1*(epsilon[mutation_index] + C[mutation_index][1]) - epsilon[mutation_index]
if C_i<0:
C[mutation_index][0] = 0
C[mutation_index][1] = 0
#check if child contains more or less than K assets and fix it
total_assets_in_child = 0
child_copy = list(C)
sorted_child_copy = sorted(child_copy, key=itemgetter(1))
for item in sorted_child_copy:
if item[0]==1:
total_assets_in_child += 1
if total_assets_in_child > K:
for item in sorted_child_copy[:-K]:
# C.index(item) = [0,0]
C[C.index(item)] = [0,0]
elif total_assets_in_child < K:
while total_assets_in_child < K:
if len(A_star) > 0:
random_index = random.randint(0,len(A_star)-1)
random_element = A_star.pop(random_index)
C[random_element[0]-1] = [1, random_element[1]]
else:
random_index = random.randint(0,30)
while C[random_index][0]==1:
random_index = random.randint(0,30)
C[random_index] = [1, 0]
total_assets_in_child += 1
#change the population by adding child to it
obj_func_val_child = 0
[obj_func_val_child, v_lambda, improved] = evaluate(C, _lambda, 0, v_lambda, False, improved_solutions, UPDATE_H)
if improved:
population[population_obj_func_vals.index(max(population_obj_func_vals))] = C
print lambda_index
outputs.append(evaluate(improved_solutions[-1], _lambda, 0, float('Inf'), False, None, FINAL_SAMPLE))
H.append(improved_solutions)
np_out_matrix = np.matrix(outputs)
np_out_matrix = np_out_matrix.T
outputs = np_out_matrix.tolist()
f = open("out2.csv", "w")
writer = csv.writer(f)
for row in outputs:
writer.writerow(tuple(row))
f.close()
def generate_aspirant_pairs(pop_size):
pop_idx = list(range(pop_size))
return [tools.selRandom(pop_idx, 2) for _ in range(pop_size)]
def select(pop, mu, lambda_):
# pop - 親集団インスタンス（creator.SRESindividualインスタンスのリスト）
offspring = tools.selRandom(tools.selBest(pop, mu), lambda_)
offspring.sort(cmp=lambda ind1, ind2: cmp(ind1.fitness.values[0], ind2.
fitness.values[0]))
return offspring
def main():
"""Complete generational algorithm
mapを使うとジェネレータが帰ってくる"""
n=100
pop = toolbox.population(n)
CXPB, MUTPB, NGEN = 0.6, 0.3, 100
elite = 0.16
random_elite = 0.04
#初期個体の評価
fitnesses = toolbox.map(toolbox.evaluate, pop) #[(6782,),(2342,)...]になってる
for ind, fit in zip(pop, fitnesses): #ここでfitnessesが遅延評価される
#print ind, fit
ind.fitness.values = fit
#Select the next generation individuals
for g in xrange(NGEN):
#g世代の個体についてstatisticsで設定した値を記録
logbook.add_dictionary(g)
#すべての個体の適応度をリストにまとめる
fits = [ind.fitness.values for ind in pop]
# recordに渡す値は実行する関数と対応
#関数の引数が複数の場合は(データ、引数)で渡す
logbook.record(fits, (pop, 1))
#上位16%の個体は残す
bestoffspring = tools.selBest(pop, int(n*elite))
top = tools.selBest(pop, 1)
print top
#下位84%の個体からランダムに4%の個体を選び、エリートに加える
worstoffspring = tools.selWorst(pop, int(n-(n*elite)))
random_ind = tools.selRandom(worstoffspring, int(n*random_elite))
save_offspring = bestoffspring + random_ind
#元のバージョンoffspring = [toolbox.clone(ind) for ind in pop]やめたほうがいい
#averageがおかしくなる
#エリート保存(20%)
save_offspring = list(toolbox.map(toolbox.clone, save_offspring))
#オペレータを適用する個体(80%)
#ランダムバージョン
operated_offspring = tools.selRandom(worstoffspring,
int(n - len(save_offspring)))
all_offspring = list(toolbox.map(toolbox.clone,
save_offspring + operated_offspring))
#print len(operated_offspring), len(save_offspring), len(all_offspring)
#Apply crossover and mutation on the offspring
# 偶数番目と奇数番目の個体を取り出して交差
for child1, child2 in zip(operated_offspring[::2],
operated_offspring[1::2]):
if random.random() < CXPB:
toolbox.mate(child1, child2)
del child1.fitness.values
del child2.fitness.values
for mutant in operated_offspring:
if random.random() < MUTPB:
toolbox.mutate(mutant)
del mutant.fitness.values
#Evaluate the individuals with an invalid fitness
invalid_ind = [ind for ind in operated_offspring if not ind.fitness.valid]
fitnesses = toolbox.map(toolbox.evaluate, invalid_ind)
#評価されていない個体を評価する.
for ind, fit in zip(invalid_ind, fitnesses):
ind.fitness.values = fit
#評価値に従って個体を選択
#The population is entirely replaced by the offspring
#pop = toolbox.select(pop + offspring, len(offspring))
pop[:] = save_offspring + operated_offspring
return logbook
def main():
MU, CXPB, MUTPB, NGEN = 200, 0.6, 0.1, 50
pop_long = toolbox.population(n=MU)
pop_short = toolbox.population(n=MU)
hof_l = tools.HallOfFame(1)
hof_s = tools.HallOfFame(1)
stats = tools.Statistics(lambda ind: ind.fitness.values)
stats.register("avg", np.mean)
stats.register("std", np.std)
stats.register("min", np.min)
stats.register("max", np.max)
logbook = tools.Logbook()
logbook.header = "gen", "type", "evals", "std", "min", "avg", "max"
best_long = tools.selRandom(pop_long, 1)[0]
best_short = tools.selRandom(pop_short, 1)[0]
for ind in pop_long:
ind.fitness.values = toolbox.evaluate(ind, best_short, points=train)
for ind in pop_short:
ind.fitness.values = toolbox.evaluate(best_long, ind, points=train)
hof_l.update(pop_long)
hof_s.update(pop_short)
record = stats.compile(pop_long)
logbook.record(gen=0, type='long', evals=len(pop_long), **record)
record = stats.compile(pop_short)
logbook.record(gen=0, type='short', evals=len(pop_short), **record)
print(logbook.stream)
# Begin the evolution
for g in range(1, NGEN):
# select and clone the offspring
off_long = toolbox.select(pop_long, MU)
off_short = toolbox.select(pop_short, MU)
off_long = [toolbox.clone(ind) for ind in off_long]
off_short = [toolbox.clone(ind) for ind in off_short]
# Apply crossover and mutation
for ind1, ind2 in zip(off_long[::2], off_long[1::2]):
if random.random() <= CXPB:
toolbox.mate(ind1, ind2)
del ind1.fitness.values
del ind2.fitness.values
elif random.random() <= MUTPB:
toolbox.mutate(ind)
del ind.fitness.values
for ind1, ind2 in zip(off_short[::2], off_short[1::2]):
if random.random() <= CXPB:
toolbox.mate(ind1, ind2)
del ind1.fitness.values
del ind2.fitness.values
elif random.random() <= MUTPB:
toolbox.mutate(ind)
del ind.fitness.values
# Evaluate the individuals
#long_representative = tools.selTournament(pop_long, 1, tournsize=3)[0]
#short_representative = tools.selTournament(pop_short, 1, tournsize=3)[0]
best_long = tools.selBest(pop_long+off_long, 1)[0]
best_short = tools.selBest(pop_short+off_short, 1)[0]
for ind in off_long:
ind.fitness.values = toolbox.evaluate(ind, best_short, points=train)
for ind in off_short:
ind.fitness.values = toolbox.evaluate(ind, best_long, points=train)
# Replace the old population by the offspring
pop_long = toolbox.select(pop_long+off_long, MU)
pop_short = toolbox.select(pop_short+off_short, MU)
record = stats.compile(pop_long)
logbook.record(gen=g, type='long', evals=len(pop_long), **record)
record = stats.compile(pop_short)
logbook.record(gen=g, type='short', evals=len(pop_short), **record)
print(logbook.stream)
hof_l.update(pop_long)
hof_s.update(pop_short)
print("Best Long individual is %s, %s" % (best_long, best_long.fitness.values))
print("Best Short individual is %s, %s" % (best_short, best_short.fitness.values))
return pop_long, pop_short, best_long, best_short, hof_l, hof_s, logbook
def main(seed=None):
random.seed(seed)
NGEN = 250
MU = 100
CXPB = 0.9
stats = tools.Statistics(lambda ind: ind.fitness.values)
# stats.register("avg", numpy.mean, axis=0)
# stats.register("std", numpy.std, axis=0)
stats.register("min", numpy.min, axis=0)
stats.register("max", numpy.max, axis=0)
logbook = tools.Logbook()
logbook.header = "gen", "evals", "std", "min", "avg", "max"
pop = toolbox.population(n=MU)
# Evaluate the individuals with an invalid fitness
invalid_ind = [ind for ind in pop if not ind.fitness.valid]
fitnesses = toolbox.map(toolbox.evaluate, invalid_ind)
for ind, fit in zip(invalid_ind, fitnesses):
ind.fitness.values = fit
record = stats.compile(pop)
logbook.record(gen=0, evals=len(invalid_ind), **record)
print(logbook.stream)
# Begin the generational process
for gen in range(1, NGEN):
# Vary the population
offspring = tools.selRandom(pop, len(pop))
offspring = [toolbox.clone(ind) for ind in offspring]
for ind1, ind2 in zip(offspring[::2], offspring[1::2]):
if random.random() <= CXPB:
toolbox.mate(ind1, ind2)
toolbox.mutate(ind1)
toolbox.mutate(ind2)
del ind1.fitness.values, ind2.fitness.values
# Evaluate the individuals with an invalid fitness
invalid_ind = [ind for ind in offspring if not ind.fitness.valid]
fitnesses = toolbox.map(toolbox.evaluate, invalid_ind)
for ind, fit in zip(invalid_ind, fitnesses):
ind.fitness.values = fit
# Select the next generation population
pop = pop + offspring
fronts = toolbox.sort(pop, len(pop))
chosen = []
for i, front in enumerate(fronts):
# Move is front to chosen population til it is almost full
if len(chosen) + len(front) <= MU:
chosen.extend(front)
else:
# Assign hypervolume contribution to individuals of front that
# cannot be completely move over to chosen individuals
fitness_hv = hypervolume_contrib(front)
for ind, fit_hv in zip(front, fitness_hv):
ind.fitness_hv.values = (fit_hv,)
# Fill chosen with best indiviuals from inspect front
# (based on hypervolume contribution)
chosen.extend(toolbox.select(front, MU - len(chosen)))
break
pop = chosen
record = stats.compile(pop)
logbook.record(gen=gen, evals=len(invalid_ind), **record)
print(logbook.stream)
print("Final population hypervolume is %f" % hypervolume(pop, [11.0, 11.0]))
return pop, logbook
def main():
pop_ga = toolbox_ga.population(n=200)
pop_gp = toolbox_gp.population(n=200)
stats = tools.Statistics(lambda ind: ind.fitness.values)
stats.register("avg", numpy.mean)
stats.register("std", numpy.std)
stats.register("min", numpy.min)
stats.register("max", numpy.max)
logbook = tools.Logbook()
logbook.header = "gen", "type", "evals", "std", "min", "avg", "max"
best_ga = tools.selRandom(pop_ga, 1)[0]
best_gp = tools.selRandom(pop_gp, 1)[0]
for ind in pop_gp:
ind.fitness.values = toolbox_gp.evaluate(ind, points=best_ga)
for ind in pop_ga:
ind.fitness.values = toolbox_gp.evaluate(best_gp, points=ind)
record = stats.compile(pop_ga)
logbook.record(gen=0, type='ga', evals=len(pop_ga), **record)
record = stats.compile(pop_gp)
logbook.record(gen=0, type='gp', evals=len(pop_gp), **record)
print(logbook.stream)
CXPB, MUTPB, NGEN = 0.5, 0.2, 50
# Begin the evolution
for g in range(1, NGEN):
# Select and clone the offspring
off_ga = toolbox_ga.select(pop_ga, len(pop_ga))
off_gp = toolbox_gp.select(pop_gp, len(pop_gp))
off_ga = [toolbox_ga.clone(ind) for ind in off_ga]
off_gp = [toolbox_gp.clone(ind) for ind in off_gp]
# Apply crossover and mutation
for ind1, ind2 in zip(off_ga[::2], off_ga[1::2]):
if random.random() < CXPB:
toolbox_ga.mate(ind1, ind2)
del ind1.fitness.values
del ind2.fitness.values
for ind1, ind2 in zip(off_gp[::2], off_gp[1::2]):
if random.random() < CXPB:
toolbox_gp.mate(ind1, ind2)
del ind1.fitness.values
del ind2.fitness.values
for ind in off_ga:
if random.random() < MUTPB:
toolbox_ga.mutate(ind)
del ind.fitness.values
for ind in off_gp:
if random.random() < MUTPB:
toolbox_gp.mutate(ind)
del ind.fitness.values
# Evaluate the individuals with an invalid fitness
for ind in off_ga:
ind.fitness.values = toolbox_gp.evaluate(best_gp, points=ind)
for ind in off_gp:
ind.fitness.values = toolbox_gp.evaluate(ind, points=best_ga)
# Replace the old population by the offspring
pop_ga = off_ga
pop_gp = off_gp
record = stats.compile(pop_ga)
logbook.record(gen=g, type='ga', evals=len(pop_ga), **record)
record = stats.compile(pop_gp)
logbook.record(gen=g, type='gp', evals=len(pop_gp), **record)
print(logbook.stream)
best_ga = tools.selBest(pop_ga, 1)[0]
best_gp = tools.selBest(pop_gp, 1)[0]
print("Best individual GA is %s, %s" % (best_ga, best_ga.fitness.values))
print("Best individual GP is %s, %s" % (best_gp, best_gp.fitness.values))
return pop_ga, pop_gp, best_ga, best_gp, logbook
def optimize(self, pop):
mstats, log = support.statistics(), support.logbook(
) # DEAP statistics and logbook
front = tools.ParetoFront() # Initialize ParetoFront class
self.evals = len(pop)
gds = [] # Initialize generational distance list
gen = totalEvals = gd = 0
while True:
# Step 3: Update global Pareto front
prevFront = front.items[:]
front.update(pop)
# Record statistics
record = mstats.compile(front)
log.record(gen=gen, evals=self.evals, gd=gd, **record)
print('\r', end='')
print(log.stream)
totalEvals += self.evals
self.evals = 0
# Step 4: Local search
for i, c in enumerate(pop):
# Fill local archive with solutions from Pareto front that do not dominate c
archive = [
ind for ind in front if
not ind.fitness.dominates(c.fitness) and ind is not c
]
archive.append(c) # Copy candidate into H
pop[i] = self.paes(
c, archive, front
) # Replace c with improved version by local search
# Step 5: Recombination
popInter = [] # Initialize intermediate population
while len(popInter) < self.sizePop:
r = 0
while True:
# Randomly choose two parents from P + G
mom, dad = tools.selRandom(pop + front.items, 2)
mom2, dad2 = self.tb.clone(mom), self.tb.clone(dad)
# Recombine to form offspring, evaluate
child, _ = self.tb.mate(mom2, dad2)
child.fitness.values = self.evaluator.evaluate(child)
self.evals += 1
childInMoreCrowdedArea = childDominated = False
for ind in front:
if ind.fitness.dominates(child.fitness):
childDominated = True
break
if not childDominated:
crowding_distance(pop + front.items + [child])
# Check if c is in more crowded grid location than both parents
if child.fitness.crowding_dist < mom.fitness.crowding_dist and\
child.fitness.crowding_dist < dad.fitness.crowding_dist:
childInMoreCrowdedArea = True
# Update pareto front with c as necessary
front.update([child])
r += 1
if not ((childDominated or childInMoreCrowdedArea)
and r < self.cr_trials):
break
childAsList = tools.selTournament(
front.items, k=1,
tournsize=2) if childDominated else [child]
popInter.extend(childAsList)
# Step 4: Termination
gd = self.termination(prevFront, front, gen, gds, totalEvals)
if not isinstance(gd, float):
return front, log, totalEvals
pop = popInter
gen += 1
def selBestOrRandom(individuals, k):
chosen = []
if random.random() <= 0.1:
return tools.selRandom(individuals, k)
else:
return tools.selBest(individuals, k)
def varDE(population,
toolbox,
cxpb=1.0,
mutpb=1.0,
jitter=0,
low=None,
up=None):
"""Part of an evolutionary algorithm applying the variation part
of the differential evolution algorithm. The modified individuals have their
fitness invalidated. The individuals are cloned so returned population is
independent of the input population.
:param population: A list of individuals to vary.
:param toolbox: A :class:`~deap.base.Toolbox` that contains the evolution
operators.
:param cxpb: The crossover probability CR: probability of a 'base' parameter
value being replaced with a differentially evolved parameter
value
:param mutpb: The mutation rate F: a scaling of the differential vector that
gets added to the base parameter
:param low: The low bound of each parameter
:param up: The upper bound of each parameter
:returns: A list of varied individuals that are independent of their
parents.
The variation goes as follows. For each member of the population, we will
generate a new offspring. For each offspring, we select 3 random members of
the parent population: a base, and 2 parents for the differential. Then we
select a random parameter 'index' that we will always change. Then, for each
parameter, if a random number [0:1] is < Cx, or if the parameter is 'index',
we change the parameter from base by adding the differential of that
parameter in each of the second two parents (multiplied by rate factor F),
(jittered by +/- jitter/2).
That's it!
"""
offspring = [toolbox.clone(ind) for ind in population]
# Differentially evolve each base offspring:
for individual in offspring:
# Keep mutating until all parameters are within all boundaries:
inside = False
while inside == False:
base, a, b = tools.selRandom(population, 3)
while (base == a or base == b or a == b):
base, a, b = tools.selRandom(population, 3)
index = random.randrange(len(individual))
for j, parameter in enumerate(individual):
if j == index or random.random() < cxpb:
diff = (a[j] - b[j])
individual[j] = base[j] + (mutpb * diff) + ((
(random.random() * jitter) - (jitter / 2)) * diff)
# perform 'bounce' away from boundaries to maintain parameter diversity
if individual[j] < low[j]:
individual[j] = low[j] + (low[j] - individual[j])
elif individual[j] > up[j]:
individual[j] = up[j] - (individual[j] - up[j])
# Check boundary
inside = True
for j, parameter in enumerate(individual):
if individual[j] < low[j]:
inside = False
elif individual[j] > up[j]:
inside = False
del individual.fitness.values
return offspring
HoF = tools.HallOfFame(36)
HoF.update(pop)
best = tools.selBest(pop, 1)
print('Best seeded outcome:')
print(best[0].fitness.values)
for i in range(N_GEN):
print('Start Gen: ', str(i), '...')
# Define new population
pop_new = toolbox.population(n=0)
winners = tools.selBest(pop, int(POP_SIZE *
0.25)) # Keep 25% best in population
rand = tools.selRandom(pop, int(POP_SIZE *
0.5)) # randomly select 50% to crossover
new = toolbox.population(n=int(
POP_SIZE * 0.25)) # introduce 25% of new randomness into population
# print(len(winners))
# print(len(rand))
# print(len(new))
# SELECT WINNERS
winners = toolbox.clone(winners)
for w in winners:
del w.fitness.values
mutant = toolbox.clone(w)
mutant = tools.mutGaussian(mutant, mu=0.0, sigma=0.1, indpb=0.2)
pop_new.append(mutant[0])
def optim(self):
"""
"""
##Create population of species to work with
self.createPopulations();
#select representatives
repr = dict();
repr['hidden_nodes'] = tools.selRandom(self.pop['node_pop'],
self.params['numHiddenNodes']);
repr['out_nodes'] = tools.selRandom(self.pop['node_pop'],
self.params['numOutputNodes']);
#select random connections and weights
repr['model'] = tools.selRandom(self.pop['model_pop'], 1);
repr['connActive_IH'] = tools.selRandom(self.pop['connActive_IH_pop'], 1);
repr['connActive_HH'] = tools.selRandom(self.pop['connActive_HH_pop'], 1);
repr['connActive_HO'] = tools.selRandom(self.pop['connActive_HO_pop'], 1);
repr['connWeights_IH'] = tools.selRandom(self.pop['connWeights_IH_pop'], 1);
repr['connWeights_HH'] = tools.selRandom(self.pop['connWeights_HH_pop'], 1);
repr['connWeights_HO'] = tools.selRandom(self.pop['connWeights_HO_pop'], 1);
# print "Representatives", repr;
#Co-op Coevolution
g = 0;
for g in xrange(self.params['NGEN']):
#Go through components population
for comp_key in self.pop.keys():
next_repr = self.tbox.clone(repr);
if comp_key == 'node_pop':
#Output units
for ind in self.pop['node_pop']:
#clone representative
components = self.tbox.clone(repr);
#evaluate the components population
components['out_nodes'] = [ind];
#assign fitness
ind.fitness.values = self.evaluate(components),-1;
print "n", ind.fitness.values[0];
del components;
self.pop['node_pop'] = tools.selTournament(self.pop['node_pop'], len(self.pop['node_pop']), 3);
next_repr['out_nodes'] = tools.selBest(self.pop['node_pop'],1);
# #clone
# components = self.tbox.clone(repr);
# #Hidden units
# for nodes in self.pop['node_pop']:
# #get representatives
# node_repr = self.tbox.clone(repr['hidden_nodes']);
# for i, n in enumerate(nodes):
# r1 = (node_repr[:i]);
# r2 = (node_repr[i+1:]);
# r1.extend(r2);
# print "Representative:", r1;
# #evaluate the components population
# components['hidden_nodes'] = r1.extend([n]);
# #assign fitness
# ind.fitness.values = self.evaluate(components),-1;
# # print "fitness:", ind.fitness.values[0];
if comp_key == 'connActive_IH_pop':
#CONN IH
for ind in self.pop['connActive_IH_pop']:
#clone representative
components = self.tbox.clone(repr);
#evaluate the components population
components['connActive_IH'] = ind;
#assign fitness
ind.fitness.values = self.evaluate(components),-1;
print "fitness:", ind.fitness.values[0];
del components;
self.pop['connActive_IH_pop'] = tools.selTournament(self.pop['connActive_IH_pop'],
len(self.pop['connActive_IH_pop']), 3);
next_repr['connActive_IH'] = tools.selBest(self.pop['connActive_IH_pop'],1);
if comp_key == 'connActive_HH_pop':
#CONN IH
for ind in self.pop['connWeights_HH_pop']:
#clone representative
components = self.tbox.clone(repr);
#evaluate the components population
components['connActive_HH'] = ind;
#assign fitness
ind.fitness.values = self.evaluate(components),-1;
print "fitness:", ind.fitness.values[0];
del components;
self.pop['connActive_HH_pop'] = tools.selTournament(self.pop['connActive_HH_pop'],
len(self.pop['connActive_HH_pop']), 3);
next_repr['connActive_HH'] = tools.selBest(self.pop['connActive_HH_pop'],1);
if comp_key == 'connActive_HO_pop':
#CONN IH
for ind in self.pop['connWeights_HO_pop']:
#clone representative
components = self.tbox.clone(repr);
#evaluate the components population
components['connActive_HO'] = ind;
#assign fitness
ind.fitness.values = self.evaluate(components),-1;
print "fitness:", ind.fitness.values[0];
del components;
self.pop['connActive_HO_pop'] = tools.selTournament(self.pop['connActive_HO_pop'],
len(self.pop['connActive_HO_pop']), 3);
next_repr['connActive_HO'] = tools.selBest(self.pop['connActive_HO_pop'],1);
if comp_key == 'connWeights_IH_pop':
#CONN IH
for ind in self.pop['connWeights_IH_pop']:
#clone representative
components = self.tbox.clone(repr);
#evaluate the components population
components['connWeights_IH'] = ind;
#assign fitness
ind.fitness.values = self.evaluate(components),-1;
print "fitness:", ind.fitness.values[0];
del components;
self.pop['connWeights_IH_pop'] = tools.selTournament(self.pop['connWeights_IH_pop'],
len(self.pop['connWeights_IH_pop']), 3);
next_repr['connWeights_IH'] = tools.selBest(self.pop['connWeights_IH_pop'],1);
if comp_key == 'connWeights_HH_pop':
#clone representative
components = self.tbox.clone(repr);
#CONN IH
for ind in self.pop['connWeights_HH_pop']:
#evaluate the components population
components['connWeights_HH'] = ind;
#assign fitness
ind.fitness.values = self.evaluate(components),-1;
print "fitness:", ind.fitness.values[0];
self.pop['connWeights_HH_pop'] = tools.selTournament(self.pop['connWeights_HH_pop'],
len(self.pop['connWeights_HH_pop']), 3);
next_repr['connWeights_HH'] = tools.selBest(self.pop['connWeights_HH_pop'],1);
if comp_key == 'connWeights_HO_pop':
#clone representative
components = self.tbox.clone(repr);
#CONN IH
for ind in self.pop['connWeights_HO_pop']:
#evaluate the components population
components['connWeights_HO'] = ind;
#assign fitness
ind.fitness.values = self.evaluate(components),-1;
print "fitness:", ind.fitness.values[0];
self.pop['connWeights_HO_pop'] = tools.selTournament(self.pop['connWeights_HO_pop'],
len(self.pop['connWeights_HO_pop']), 3);
next_repr['connWeights_HO'] = tools.selBest(self.pop['connWeights_HO_pop'],1);
repr = next_repr;
##### TEST ########
# p = pyNDMOptim();
# p.optim();
# sys.exit();
#TODO: Coevolution of neural computation paths
#TODO:
#cc1 cc2 is totally new
cc1, cc2 = toolbox.mateG(c1, c2)
cxOut.append(cc1)
cxOut.append(cc2)
# mutation
mutantOut = []
for mutant in offspring:
if random.random() < GMUTPB:
mut = toolbox.mutateG(mutant, dest)
if mut != None:
mutantOut.append(mut)
#compose a big group of population
#bigPop = pop + mutantOut + cxOut
bigPop = mutantOut + cxOut + lPop
gPop = tools.selRandom(bigPop, GPOPU)
""" L part !!!"""
# select
#offspring = toolbox.selectL(lPop, 2)
offspring = tools.selBest(lPop, LPOPU)
# deep copy out
# offspring = list(map(toolbox.clone, offspring))
# cross over
cxOut = []
for c1 in offspring:
if random.random() < LCXPB:
#cc1 cc2 is totally new
#random select c2 from gPop
c2 = tools.selRandom(gPop, 1)[0]
cc1, cc2 = toolbox.mateL(c1, c2)
cxOut.append(cc1)
def run_pso(instance_name,
particle_size,
pop_size,
max_iteration,
cognitive_coef,
social_coef,
s_limit=3,
plot=False,
save=False,
logs=False):
instance = load_problem_instance(instance_name)
if instance is None:
return
if plot:
plot_instance(instance_name=instance_name,
customer_number=particle_size)
creator.create("FitnessMax", base.Fitness, weights=(1.0, ))
creator.create("Particle",
list,
fitness=creator.FitnessMax,
speed=list,
smin=None,
smax=None,
best=None)
toolbox = base.Toolbox()
toolbox.register("particle",
generate_particle,
size=particle_size,
val_min=1,
val_max=particle_size,
s_min=-s_limit,
s_max=s_limit)
toolbox.register("population", tools.initRepeat, list, toolbox.particle)
toolbox.register("update",
update_particle,
phi1=cognitive_coef,
phi2=social_coef)
toolbox.register('evaluate', calculate_fitness, data=instance)
pop = toolbox.population(n=pop_size)
stats = tools.Statistics(lambda ind: ind.fitness.values)
stats.register("avg", numpy.mean)
stats.register("std", numpy.std)
stats.register("min", numpy.min)
stats.register("max", numpy.max)
logbook = tools.Logbook()
logbook.header = ["gen", "evals"] + stats.fields
best = None
iter_num = 0
previous_best = 0
print('### EVOLUTION START ###')
start = time.time()
for g in range(max_iteration):
fit_count = 0
for part in pop:
part.fitness.values = toolbox.evaluate(part)
if part.fitness.values[0] > previous_best:
previous_best = part.fitness.values[0]
iter_num = g + 1
elif part.fitness.values[0] == previous_best:
fit_count += 1
if fit_count > int(numpy.ceil(pop_size * 0.15)):
rand_pop = toolbox.population(n=pop_size)
for part in rand_pop:
part.fitness.values = toolbox.evaluate(part)
some_inds = tools.selRandom(
rand_pop, int(numpy.ceil(pop_size * 0.1))) # random pop here
mod_pop = tools.selWorst(pop, int(numpy.ceil(pop_size * 0.9)))
else:
some_inds = tools.selBest(pop, int(numpy.ceil(
pop_size * 0.05))) # elite pop here
mod_pop = tools.selRandom(pop, int(numpy.ceil(pop_size * 0.95)))
mod_pop = list(map(toolbox.clone, mod_pop))
for part in mod_pop:
if not part.best or part.best.fitness < part.fitness:
part.best = creator.Particle(part)
part.best.fitness.values = part.fitness.values
if not best or best.fitness < part.fitness:
best = creator.Particle(part)
best.fitness.values = part.fitness.values
for part in mod_pop:
toolbox.update(part, best)
mod_pop.extend(some_inds)
pop[:] = mod_pop
# Gather all the stats in one list and print them
logbook.record(gen=g + 1, evals=len(pop), **stats.compile(pop))
print(logbook.stream)
end = time.time()
print('### EVOLUTION END ###')
best_ind = tools.selBest(pop, 1)[0]
print(f'Best individual: {best_ind}')
route = create_route_from_ind(best_ind, instance)
print_route(route)
print(f'Fitness: { round(best_ind.fitness.values[0],2) }')
print(f'Total cost: { round(calculate_fitness(best_ind, instance)[1],2) }')
print(f'Found in (iteration): { iter_num }')
print(f'Execution time (s): { round(end - start,2) }')
# print(f'{round(best_ind.fitness.values[0], 2)} & {round(calculate_fitness(best_ind, instance)[1],2)} & {iter_num} & {round(end - start, 2)}')
if plot:
plot_route(route=route, instance_name=instance_name)
return route
def main(epsilon, delta):
global lambda_values
global population_size
global population
global population_obj_func_vals
global iterations
outputs = []
for lambda_index in range(1, number_of_lambdas+1):
_lambda = float(lambda_index-1)/(number_of_lambdas-1)
improved_solutions = [] #improved solutions for a particular lambda
v_lambda = float('Inf')
for i in range(1 , population_size+1):
sample = [[0,0] for x in range(number_of_assets)]
Q = random.sample([x for x in range(1, number_of_assets+1)],10)
for asset in Q:
sample[asset-1] = [1, myRandom(0,1)]
population[i-1] = list(sample)
for ind,S in enumerate(population):
[population_obj_func_vals[ind], v_lambda, improved] = evaluate(S, _lambda, 0, v_lambda, False, improved_solutions, UPDATE_H)
for itr in range(iterations):
#binary tournament for selecting parents S_star and S_double_star
chosen = []
for i in xrange(2):
aspirants = tools.selRandom(population, 40)
for i,aspirant in enumerate(aspirants):
h = []
f_aspirant = 0
v = float('Inf')
imp = False
[f_aspirant, v, imp] = evaluate(aspirant, _lambda, 0, v, False, h, DONT_UPDATE_H)
aspirants[i] = [aspirant, f_aspirant]
chosen.append(min(aspirants, key=lambda x: x[1])[0])
S_star = list(chosen[0])
S_double_star = list(chosen[1])
#Uniform crossover to find child C
C = []
[C1, C2] = tools.cxUniform(S_star, S_double_star, ind_prob)
f1 = f2 = 0
v = float('Inf')
imp = False
h = []
[f1, v, imp] = evaluate(C1, _lambda, 0, v, False, h, DONT_UPDATE_H)
[f2, v, imp] = evaluate(C2, _lambda, 0, v, False, h, DONT_UPDATE_H)
if f1<f2:
C = C1
else:
C = C2
#find assets in parents but not in child
A_star = []
for i in range(number_of_assets):
if S_star[i][0]==1 and S_double_star[i][0]==0 and C[i][0]==0:
A_star.append([i+1, S_star[i][1]])
elif S_star[i][0]==0 and S_double_star[i][0]==1 and C[i][0]==0:
A_star.append([i+1, S_double_star[i][1]])
#Mutation
mutation_index = random.randint(0, number_of_assets-1)
while C[mutation_index][0]==0:
mutation_index = random.randint(0, number_of_assets-1)
m = random.randint(0,1)
if m == 0:
# C_i = 0.9*(epsilon + C[mutation_index][1]) - epsilon
C_i = 0.9*(epsilon[mutation_index] + C[mutation_index][1]) - epsilon[mutation_index]
else:
# C_i = 1.1*(epsilon + C[mutation_index][1]) - epsilon
C_i = 1.1*(epsilon[mutation_index] + C[mutation_index][1]) - epsilon[mutation_index]
if C_i<0:
C[mutation_index][0] = 0
C[mutation_index][1] = 0
#check if child contains more or less than K assets and fix it
total_assets_in_child = 0
child_copy = list(C)
sorted_child_copy = sorted(child_copy, key=itemgetter(1))
for item in sorted_child_copy:
if item[0]==1:
total_assets_in_child += 1
if total_assets_in_child > K:
for item in sorted_child_copy[:-K]:
# C.index(item) = [0,0]
C[C.index(item)] = [0,0]
elif total_assets_in_child < K:
while total_assets_in_child < K:
if len(A_star) > 0:
random_index = random.randint(0,len(A_star)-1)
random_element = A_star.pop(random_index)
C[random_element[0]-1] = [1, random_element[1]]
else:
random_index = random.randint(0,30)
while C[random_index][0]==1:
random_index = random.randint(0,30)
C[random_index] = [1, 0]
total_assets_in_child += 1
#change the population by adding child to it
obj_func_val_child = 0
[obj_func_val_child, v_lambda, improved] = evaluate(C, _lambda, 0, v_lambda, False, improved_solutions, UPDATE_H)
if improved:
population[population_obj_func_vals.index(max(population_obj_func_vals))] = C
print lambda_index
outputs.append(evaluate(improved_solutions[-1], _lambda, 0, float('Inf'), False, None, FINAL_SAMPLE))
H.append(improved_solutions)
np_out_matrix = np.matrix(outputs)
np_out_matrix = np_out_matrix.T
outputs = np_out_matrix.tolist()
f = open("out5.csv", "w")
writer = csv.writer(f)
for row in outputs:
writer.writerow(tuple(row))
f.close()
def random_subset(self):
#return random.sample(self.population, self.subset_size)
return tools.selRandom(self.population, self.subset_size)
def main():
pop_ga = toolbox_ga.population(n=200)
pop_gp = tools_gp.population(n=200)
stats_ga = tools.Statistics(lambda ind: ind.fitness.values)
stats_ga.register("avg", tools.mean)
stats_ga.register("std", tools.std)
stats_ga.register("min", min)
stats_ga.register("max", max)
stats_gp = tools.Statistics(lambda ind: ind.fitness.values)
stats_gp.register("avg", tools.mean)
stats_gp.register("std", tools.std)
stats_gp.register("min", min)
stats_gp.register("max", max)
column_names = ["gen", "evals"]
column_names.extend(stats_ga.functions.keys())
logger = tools.EvolutionLogger(column_names)
logger.logHeader()
best_ga = tools.selRandom(pop_ga, 1)[0]
best_gp = tools.selRandom(pop_gp, 1)[0]
for ind in pop_gp:
ind.fitness.values = evalSymbReg(ind, best_ga)
for ind in pop_ga:
ind.fitness.values = evalSymbReg(best_gp, ind)
stats_ga.update(pop_ga)
stats_gp.update(pop_gp)
logger.logGeneration(gen="0 (ga)", evals=len(pop_ga), stats=stats_ga)
logger.logGeneration(gen="0 (gp)", evals=len(pop_gp), stats=stats_gp)
CXPB, MUTPB, NGEN = 0.5, 0.2, 50
# Begin the evolution
for g in range(1, NGEN):
# Select and clone the offspring
off_ga = toolbox_ga.select(pop_ga, len(pop_ga))
off_gp = tools_gp.select(pop_gp, len(pop_gp))
off_ga = [toolbox_ga.clone(ind) for ind in off_ga]
off_gp = [tools_gp.clone(ind) for ind in off_gp]
# Apply crossover and mutation
for ind1, ind2 in zip(off_ga[::2], off_ga[1::2]):
if random.random() < CXPB:
toolbox_ga.mate(ind1, ind2)
del ind1.fitness.values
del ind2.fitness.values
for ind1, ind2 in zip(off_gp[::2], off_gp[1::2]):
if random.random() < CXPB:
tools_gp.mate(ind1, ind2)
del ind1.fitness.values
del ind2.fitness.values
for ind in off_ga:
if random.random() < MUTPB:
toolbox_ga.mutate(ind)
del ind.fitness.values
for ind in off_gp:
if random.random() < MUTPB:
tools_gp.mutate(ind)
del ind.fitness.values
# Evaluate the individuals with an invalid fitness
for ind in off_ga:
ind.fitness.values = evalSymbReg(best_gp, ind)
for ind in off_gp:
ind.fitness.values = evalSymbReg(ind, best_ga)
# Replace the old population by the offspring
pop_ga = off_ga
pop_gp = off_gp
stats_ga.update(pop_ga)
stats_gp.update(pop_gp)
best_ga = tools.selBest(pop_ga, 1)[0]
best_gp = tools.selBest(pop_gp, 1)[0]
logger.logGeneration(gen="%d (ga)" % g, evals=len(off_ga), stats=stats_ga)
logger.logGeneration(gen="%d (gp)" % g, evals=len(off_gp), stats=stats_gp)
print("Best individual GA is %s, %s" % (best_ga, best_ga.fitness.values))
print("Best individual GP is %s, %s" % (gp.stringify(best_gp), best_gp.fitness.values))
return pop_ga, pop_gp, stats_ga, stats_gp, best_ga, best_gp
def main():
pop_ga = toolbox_ga.population(n=200)
pop_gp = toolbox_gp.population(n=200)
stats = tools.Statistics(lambda ind: ind.fitness.values)
stats.register("avg", numpy.mean)
stats.register("std", numpy.std)
stats.register("min", numpy.min)
stats.register("max", numpy.max)
logbook = tools.Logbook()
logbook.header = "gen", "type", "evals", "std", "min", "avg", "max"
best_ga = tools.selRandom(pop_ga, 1)[0]
best_gp = tools.selRandom(pop_gp, 1)[0]
for ind in pop_gp:
ind.fitness.values = toolbox_gp.evaluate(ind, points=best_ga)
for ind in pop_ga:
ind.fitness.values = toolbox_gp.evaluate(best_gp, points=ind)
record = stats.compile(pop_ga)
logbook.record(gen=0, type='ga', evals=len(pop_ga), **record)
record = stats.compile(pop_gp)
logbook.record(gen=0, type='gp', evals=len(pop_gp), **record)
print(logbook.stream)
CXPB, MUTPB, NGEN = 0.5, 0.2, 50
# Begin the evolution
for g in range(1, NGEN):
# Select and clone the offspring
off_ga = toolbox_ga.select(pop_ga, len(pop_ga))
off_gp = toolbox_gp.select(pop_gp, len(pop_gp))
off_ga = [toolbox_ga.clone(ind) for ind in off_ga]
off_gp = [toolbox_gp.clone(ind) for ind in off_gp]
# Apply crossover and mutation
for ind1, ind2 in zip(off_ga[::2], off_ga[1::2]):
if random.random() < CXPB:
toolbox_ga.mate(ind1, ind2)
del ind1.fitness.values
del ind2.fitness.values
for ind1, ind2 in zip(off_gp[::2], off_gp[1::2]):
if random.random() < CXPB:
toolbox_gp.mate(ind1, ind2)
del ind1.fitness.values
del ind2.fitness.values
for ind in off_ga:
if random.random() < MUTPB:
toolbox_ga.mutate(ind)
del ind.fitness.values
for ind in off_gp:
if random.random() < MUTPB:
toolbox_gp.mutate(ind)
del ind.fitness.values
# Evaluate the individuals with an invalid fitness
for ind in off_ga:
ind.fitness.values = toolbox_gp.evaluate(best_gp, points=ind)
for ind in off_gp:
ind.fitness.values = toolbox_gp.evaluate(ind, points=best_ga)
# Replace the old population by the offspring
pop_ga = off_ga
pop_gp = off_gp
record = stats.compile(pop_ga)
logbook.record(gen=g, type='ga', evals=len(pop_ga), **record)
record = stats.compile(pop_gp)
logbook.record(gen=g, type='gp', evals=len(pop_gp), **record)
print(logbook.stream)
best_ga = tools.selBest(pop_ga, 1)[0]
best_gp = tools.selBest(pop_gp, 1)[0]
print("Best individual GA is %s, %s" % (best_ga, best_ga.fitness.values))
print("Best individual GP is %s, %s" % (best_gp, best_gp.fitness.values))
return pop_ga, pop_gp, best_ga, best_gp, logbook
def selBestOrRandom(individuals, k):
chosen = []
if random.random() <= 0.1:
return tools.selRandom(individuals, k)
else:
return tools.selBest(individuals, k)
