def torneio(self):
candidato_1 = self.populacao[randint(0, self.n_pop)]
candidato_2 = self.populacao[randint(0, self.n_pop)]
if candidato_1.fitness <= candidato_2.fitness:
self.pais.append(deep_copy(candidato_1))
else:
self.pais.append(deep_copy(candidato_2))
def cruzamento(self):
for i in range(0, self.n_pop, 2):
if randfloat() <= self.pc:
filho_1 = deep_copy(self.pais[i])
filho_2 = deep_copy(self.pais[i+1])
filho_1.cruzamento(filho_2)
self.filhos.append(filho_1)
self.filhos.append(filho_2)
else:
self.filhos.append(deep_copy(self.pais[i]))
self.filhos.append(deep_copy(self.pais[i+1]))
def remain_pop(self):
# sort everybody for the rank and crowd distance of individuals
population = []
for front in self.frontiers:
if len(front) + len(population) > self.pop_size:
population = self.crowd_distance(population=population, front=front)
break
for i in front:
population.append(deep_copy(self.everybody[i]))
if len(population) == self.pop_size:
break
self.population = deep_copy(population)
def selection(self):
self.fathers = []
for i in range(self.pop_size):
u = self.population[random.randint(0, self.pop_size - 1)]
v = self.population[random.randint(0, self.pop_size - 1)]
if u.dominates(other=v):
self.fathers.append(deep_copy(u))
elif v.dominates(other=u):
self.fathers.append(deep_copy(v))
else:
r = random.random()
if r >= 0.5:
self.fathers.append(deep_copy(u))
else:
self.fathers.append(deep_copy(v))
def cross_over(self):
self.sons = []
for i in range(self.pop_size): # create empty sons
ind = Individual(nvar=self.n_var,
nbits=self.bits_per_variable,
ncal=self.n_cal)
self.sons.append(deep_copy(ind))
for i in range(0, self.pop_size,
2): # cross over and fill their binary code
for j in range(self.n_var):
r = random.randint(1, self.bits_per_variable)
gray_code_father_1 = gc.bin_to_gray(
self.fathers[i].binary_code[j])
gray_code_father_2 = gc.bin_to_gray(
self.fathers[i + 1].binary_code[j])
gray_code_son_1 = [
gray_code_father_1[0:r], gray_code_father_2[r:]
]
gray_code_son_1 = ''.join(gray_code_son_1)
gray_code_son_2 = [
gray_code_father_2[0:r], gray_code_father_1[r:]
]
gray_code_son_2 = ''.join(gray_code_son_2)
self.sons[i].binary_code.append(
gc.gray_to_bin(gray_code_son_1))
self.sons[i + 1].binary_code.append(
gc.gray_to_bin(gray_code_son_2))
def add_relation(self, source, target, kind, label=None, metadata=None):
"""Adds a relation to the context graph."""
assert utilities.valid_string(source) and utilities.valid_string(target)
assert utilities.valid_string(kind)
assert utilities.valid_optional_string(label)
assert (metadata is None) or isinstance(metadata, types.DictType)
with self._lock:
# The timestamp of the relation should be inherited from the previous
# context graph.
key = (source, target, kind)
timestamp = self._previous_relations_to_timestamps.get(key)
if not utilities.valid_string(timestamp):
timestamp = utilities.now()
# Add the relation to the context graph data structure.
relation = {
'source': source,
'target': target,
'type': kind,
'timestamp': timestamp
}
self._current_relations_to_timestamps[key] = timestamp
# Add annotations as needed.
relation['annotations'] = {}
if metadata is not None:
relation['annotations']['metadata'] = copy.deep_copy(metadata)
relation['annotations']['label'] = label if label is not None else kind
if self._version is not None:
relation['annotations']['createdBy'] = self._version
self._context_relations.append(relation)
def initialize_population(self):
for i in range(self.pop_size):
ind = Individual(nvar=self.n_var, nbits=self.bits_per_variable, ncal=self.n_cal)
self.population.append(deep_copy(ind)) # define it as object
self.population[i].create_individual() # create binary string
self.population[i].binary_to_real(self.population[i].binary_code) # decript
self.population[i].calc_fitness() # set fitness
def _authenticate(cloud):
"""Authenticate to a cloud
Ideally we would use shade.openstack_cloud or
os_client_config.make_shade (equivalent), but they don't seem to
support passing in an arbitrary dict of configuration parameters
(the docs are not very clear ATM).
To work around this, we write a temporary file with the
appropriate configuration and tell os-client-config to load it.
"""
CLOUD_NAME = 'default'
tmp = tempfile.NamedTemporaryFile()
try:
cfg = {'clouds': {CLOUD_NAME: cloud}}
y = yaml.safe_dump(cfg)
tmp.write(y)
tmp.seek(0)
old_env = copy.deep_copy(os.environ)
try:
os.environ['OS_CLIENT_CONFIG_FILE'] = tmp.name
auth = os_client_config.make_shade()
return auth
finally:
os.environ = old_env
finally:
tmp.close()
def add_relation(self, source, target, kind, label=None, metadata=None):
"""Adds a relation to the context graph."""
assert utilities.valid_string(source) and utilities.valid_string(target)
assert utilities.valid_string(kind)
assert utilities.valid_optional_string(label)
assert (metadata is None) or isinstance(metadata, dict)
with self._lock:
# The timestamp of the relation should be inherited from the previous
# context graph.
key = (source, target, kind)
timestamp = self._previous_relations_to_timestamps.get(key)
if not utilities.valid_string(timestamp):
timestamp = utilities.now()
# Add the relation to the context graph data structure.
relation = {
'source': source,
'target': target,
'type': kind,
'timestamp': timestamp
}
self._current_relations_to_timestamps[key] = timestamp
# Add annotations as needed.
relation['annotations'] = {}
if metadata is not None:
relation['annotations']['metadata'] = copy.deep_copy(metadata)
relation['annotations']['label'] = label if label is not None else kind
self._context_relations.append(relation)
def deposits(self) -> Iterable[Deposit]:
"""Create a dict representation of this deposit suitable for being converted to JSON."""
if self.template is None:
raise TypeError(
"cannot generate JSON for a gradient without a template deposit"
)
for step_index in range(0, self.step_count + 1):
percent = step_index / self.step_count
deposit = deep_copy(self.template)
deposit.deposit_name += f"-{self.name}_step{step_index}"
for sweep in self.sweeps:
sweep.percent = percent
if sweep.name == "center_height":
sweep.assign(deposit.distribution)
elif sweep.name == "cluster_size":
sweep.assign(deposit)
elif sweep.name == "max_height":
sweep.assign(deposit.distribution)
elif sweep.name == "min_height":
sweep.assign(deposit.distribution)
elif sweep.name == "purity":
sweep.assign(deposit.vein, do_round=False)
else:
raise NotImplementedError(
f"no known sweep of {sweep.name}")
yield deposit
def _lookup_efs(self, ms_items):
for ms_item in ms_items:
for s in ms_item.sents:
s.pred_entity_fillers.clear()
for one_ef in s.entity_fillers:
copied_ef = deep_copy(one_ef)
copied_ef.links.clear() # clean links
s.pred_entity_fillers.append(copied_ef)
def roleta(self):
giro_da_roleta = randfloat(0, self.soma_roleta)
aux = 0
i = -1
while aux < giro_da_roleta:
aux += 1/self.populacao[i].fitness
i += 1
self.pais.append(deep_copy(self.populacao[i]))
def update_best(self, g_best_value, g_best_position):
if self.type_PSO == 'Global':
for i in range(self.swarm_size):
if self.fitness[i] < g_best_value:
g_best_value = deep_copy(self.fitness[i])
g_best_position = deep_copy(self.swarm[i])
elif self.type_PSO == 'Local':
aux_g_best_value = deep_copy(g_best_value)
aux_1 = -1
aux_2 = 1
for i in range(self.swarm_size):
if aux_2 >= self.swarm_size:
aux_2 = 0
vector = [
aux_g_best_value[aux_1], self.fitness[i],
aux_g_best_value[aux_2]
]
g_best_value[i] = deep_copy(np.min(vector))
if np.argmin(vector) == 0:
g_best_position[i] = deep_copy(g_best_position[aux_1])
elif np.argmin(vector) == 1:
g_best_position[i] = deep_copy(self.swarm[i])
elif np.argmin(vector) == 2:
g_best_position[i] = deep_copy(g_best_position[aux_2])
aux_1 += 1
aux_2 += 1
return g_best_value, g_best_position
def mutacao(y, fitness_aux, prob, max_v, min_v, bic_score, nodes, nao_dag):
for val in range(len(y)):
r = random.random()
if r <= prob:
valor_mut = deep_copy(y[val])
valor_mut_antigo = deep_copy(y[val])
while (valor_mut == y[val]):
valor_mut = min_v + random.randint(min_v, max_v)
y[val] = valor_mut
if y not in nao_dag:
G = vetor_Rede(y, nodes)
if G:
fitness_aux = abs(bic_score.score(G))
else:
nao_dag.append(y)
y[val] = valor_mut_antigo
else:
y[val] = valor_mut_antigo
return y, fitness_aux
def fromPipeline(cls, pipeline: pipelineIR.PipelineIR) -> Pipeline:
"""Create a new pipeline by copying an already existing `Pipeline`.
Parameters
----------
pipeline: `Pipeline`
An already created pipeline intermediate representation object
Returns
-------
pipeline: `Pipeline`
"""
return cls.fromIR(copy.deep_copy(pipeline._pipelineIR))
def move_rectangle2(rect, dx, dy):
"""A version of move_rectangle that creates and returns a new Rectangle
instead of modifying the old one.
"""
# copy both rect and the Point object representing its corner
new_rect = copy.deep_copy(rect)
# modify the Point coords
new_rect.corner.x += dx
new_rect.corner.y += dy
return new_rect
def move_rectangle2(rect, dx, dy):
"""A version of move_rectangle that creates and returns a new Rectangle
instead of modifying the old one.
"""
# copy both rect and the Point object representing its corner
new_rect = copy.deep_copy(rect)
# modify the Point coords
new_rect.corner.x += dx
new_rect.corner.y += dy
return new_rect
def init_fisher(self, Fishers, par_prior_variance=None, pars={}):
self.Npars = {}
self.pars = {}
for i in Fishers.keys():
self.Npars[i] = len(Fishers[i])
self.pars[i] = list(pars[i])
if par_prior_variance is None:
self.Fishers = Fishers
else:
self.Fishers_0prior = copy.deep_copy(Fishers)
self.Fisher_prior = np.diag(1. / np.array(par_prior_variance))
for i in Fishers.keys():
self.Fishers[i] = self.Fisher_0prior[i] + self.Fisher_prior
def convert_definition(self, conversion_fun):
"""
Convert a trajectory definition line-by-line using a conversion function.
Trajectory definition of type 'direct' and 'interp' can be converted, but
waveform trajectory definitions cannot.
Parameters
----------
conversion_fun : function
Conversion function taking in one waypoint (list) and returning waypoint (list)
Raises
------
RuntimeError
If the trajectory definition is not set
RuntimeError
If the trajectory type is incompatible (not direct or interp)
"""
if self.definition is None:
raise RuntimeError(
"Trajectory definition has not been set or loaded.")
if not (self.traj_type in ["direct", "interp"]):
raise RuntimeError("Incompatible trajectory type: %s" %
(self.traj_type))
def_new = copy.deepcopy(self.definition)
setpoints = copy.deep_copy(self.definition['config']['setpoints'])
setpoints_new = {}
for traj_segment_key in setpoints:
traj_segment = setpoints[traj_segment_key]
# If the trajectory segment is empty, pass that along
if traj_segment is None:
setpoints_new[traj_segment_key] = None
continue
# If the trajectory has lines, convert them
setpoints_new[traj_segment_key] = []
for line in traj_segment:
setpoints_new[traj_segment_key].append(conversion_fun(line))
setpoints_new['meta'] = {'converted': True}
def_new['config']['setpoints'] = setpoints_new
self.definition = def_new
self.build_traj()
def _copy_sent(sent, cur_event):
ret = shallow_copy(sent)
ret.events = []
ret.pred_events = []
if cur_event is not None:
# only one event for center
ret.events.append(
cur_event) # for events, use the original one
copied_event = deep_copy(
cur_event) # for pred, use the copied one
copied_event.links.clear(
) # and clear links(args) for prediction
ret.pred_events.append(copied_event)
ret.entity_fillers = shallow_copy(
ret.entity_fillers) # for training
ret.pred_entity_fillers = [] # to predict
ret.orig_sent = sent # used in prediction to append back to the original instances
return ret
def __init__(self, nvar, ncal):
# Atributos
self.n_var = nvar
self.n_cal = ncal
self.pc = 0.7
self.pm = 0.005
self.n_pop = int(np.sqrt(self.n_cal))
if self.n_pop > 200:
self.n_pop = 200
if self.n_pop % 2 == 1:
self.n_pop = self.n_pop - 1
self.n_iter = self.n_cal//self.n_pop
self.populacao = []
self.pais = []
self.filhos = []
self.problema = Rastringin(self.n_var)
self.fitness = []
self.variaveis = [] # gambs para plotar
# Auxiliares
self.media_fitness_pop = []
self.fitness_plot = []
self.geracoes = []
# Inicialização a população
for i in range(self.n_pop):
ind = self.problema.tipo_individuo()
ind.inicializar_parametros(n_var=self.n_var, tipo_funcao=self.problema)
ind.criar_individuo()
ind.calc_fitness()
self.variaveis.append(ind.variaveis)
self.fitness.append(ind.fitness)
self.populacao.append(ind)
# Definir Best em geração 0
self.best = deep_copy(self.populacao[0])
self.salvar_elite()
# salvar valores para plotar
self.media_fitness_pop.append(np.mean(self.fitness))
self.fitness_plot.append(self.best.fitness)
self.geracoes.append(0)
# Evolução
self.evoluir()
def batch_train(self, train, train_targets, test, test_targets, batch_size, partition_sizes = [0.7, 0.2, 0.1], epochs = 150, plot_training_results = True, shuffle=True, verbose = False):
#Batch training method. Picks a random sized batch. Calculates the average value of n vectors with average targets.
# Trains the MLP on the batch.
#We begin to randomize the vectors.
training_accuracy = np.zeros(epochs)
test_accuracy = np.zeros(epochs)
num = len(train)//batch_size
rest = len(train)%batch_size
flag = (rest != 0)
for k in range(epochs):
if(shuffle):
p = np.random.permutation(len(train))
train = train[p]
train_targets = train_targets[p]
for i in range(num):
inp = train[batch_size*i:batch_size*(i+1)]
target = train_targets[batch_size*i:batch_size*(i+1)]
# batch_inputs = np.mean(inp, axis= 0)
# batch_targets = np.mean(target, axis= 0)
self.backpropagation_multiple_layers_batch(inp, target, batch_size)
#Finally the last batch.
if(flag):
inp = train[batch_size*(i+1):batch_size*(i+2)]
target = train_targets[batch_size*(i+1):batch_size*(i+2)]
# batch_inputs = np.average(inp, axis= 0)
# batch_targets = np.average(target, axis= 0)
self.backpropagation_multiple_layers_batch(inp, target, batch_size)
training_accuracy[k] = self.loss_function(train, train_targets)
test_accuracy[k] = self.loss_function(test, test_targets)
if(test_accuracy[k] > self.best_accuracy):
#Keep a copy of the best performing network
self.best_accuracy = test_accuracy[k]
self.best_network = copy.deep_copy(self.network)
self.best_epoch = k
if(verbose):
print("Iteration %i, loss = %f" % (k, test_accuracy[k]))
if(plot_training_results):
self.plot_training_progress(training_accuracy, test_accuracy)
return(training_accuracy, test_accuracy)
def map_merge(subject, other, match_fn, *args):
'''Given two lists of dictionaries, merge the dictionaries.
Dictionaries are merged if the `match_fn` applied to the single elements,
returns true.
Additional arguments are passed to the match function.
'''
try:
match_fn = MATCHERS[match_fn]
except KeyError:
raise AnsibleFilterError('Unknown match function')
subject = list(subject)
other = list(other)
matched = []
result = []
for subject_item in subject:
for other_item in other:
if match_fn(subject_item, other_item, *args):
# Keep track of matched items.
matched.append(subject_item)
matched.append(other_item)
# Compute the resulting item
# (merging subject item with other item).
result_item = deep_copy(subject_item)
result_item.update(other_item)
result.append(result_item)
# Add items that never matched, without modifying them.
for item in subject + other:
if len([m for m in matched if item is m]) == 0:
# This item never matched.
result.append(item)
return result
def map_merge(subject, other, match_fn, *args):
'''Given two lists of dictionaries, merge the dictionaries.
Dictionaries are merged if the `match_fn` applied to the single elements,
returns true.
Additional arguments are passed to the match function.
'''
try:
match_fn = MATCHERS[match_fn]
except KeyError:
raise AnsibleFilterError('Unknown match function')
subject = list(subject)
other = list(other)
matched = []
result = []
for subject_item in subject:
for other_item in other:
if match_fn(subject_item, other_item, *args):
# Keep track of matched items.
matched.append(subject_item)
matched.append(other_item)
# Compute the resulting item
# (merging subject item with other item).
result_item = deep_copy(subject_item)
result_item.update(other_item)
result.append(result_item)
# Add items that never matched, without modifying them.
for item in subject + other:
if len([m for m in matched if item is m]) == 0:
# This item never matched.
result.append(item)
return result
def initialize_population(self): # ok!
'''# ________________________________________________________________________________________________________
# gambs to add old run as the new population
old_results = pd.read_csv('newpop.csv', header=None)
# end gambs =============================================================================================='''
for i in range(self.pop_size):
ind = Individual(ncal=self.n_cal)
self.population.append(deep_copy(ind)) # define it as object
'''# ____________________________________________________________________________________________________
# gambs to add old run as the new population
self.population[i].variables = np.array(old_results.iloc[i])
self.population[i].calc_fitness()
# end gambs =========================================================================================='''
self.population[i].create_individual(
) # create string of variables
self.population[i].calc_fitness() # set fitness
#
# inset the strong individuals
self.population[-3].variables = np.zeros([500]) + 1
self.population[-3].calc_fitness()
self.population[-2].variables = np.zeros([500]) + 2
self.population[-2].calc_fitness()
self.population[-1].variables = np.zeros([500])
self.population[-1].calc_fitness()
def create_new_population(self): # Creates the union of fathers and sons
self.everybody = []
for i in range(0, self.pop_size):
self.everybody.append(deep_copy(self.population[i]))
for i in range(0, self.pop_size):
self.everybody.append(deep_copy(self.sons[i]))
def salvar_elite(self):
for i in range(self.n_pop):
if self.populacao[i].fitness < self.best.fitness:
self.best = deep_copy(self.populacao[i])
def return_cards_to_dealer(self):
old_hand = copy.deep_copy(self.hand)
self.hand.clear()
return old_hand
def criar_nova_pop(self):
self.populacao = deep_copy(self.filhos)
self.populacao[np.argmax(self.fitness)] = self.best
def get_parameter_from(self, high_score_worker): # for exploit
self.trainer.hyper_parameter = deep_copy(
high_score_worker.trainer.hyper_parameter)
self.trainer.net = deep_copy(high_score_worker.trainer.net)
def annealing(maxsteps=1000, debug=True):
""" Optimize the black-box function 'cost_function' with the simulated annealing algorithm."""
#Ler data
with open('Asia.csv') as csv_file:
csv_reader = csv.reader(csv_file, delimiter=',')
aux = 0
data = []
data1 = [[] for i in range(8)]
for row in csv_reader:
data.append(row)
for i in range(len(row)):
data1[i].append(row[i])
aux = aux + 1
if aux == 50001:
break
data = {}
for i in range(len(data1)):
data[data1[i][0]] = [data1[i][j] for j in range(1, len(data1[i]))]
data = pd.DataFrame(data)
print("Data: ")
print(data) #Dados Retirandos do arquivo
prob = 0.5
min_valor = 0
max_valor = 2
nao_dag = []
nodes = ['Pollution', 'Smoker', 'Cancer', 'Xray', 'Dyspnoea']
nodes = ['asia', 'tub', 'smoke', 'lung', 'bronc', 'either', 'xray', 'dysp']
ind_size = round((len(nodes) * len(nodes) - len(nodes)) / 2)
ind = False
while ind == False:
aux = [random.randint(min_valor, max_valor) for i in range(ind_size)]
if aux not in nao_dag:
G = vetor_Rede(aux, nodes)
if G:
state = deep_copy(aux)
ind = True
else:
nao_dag.append(aux)
print('state')
print(state)
bic_score = BicScore(data)
print(vetor_Rede(state, nodes))
cost = cost_function(state, bic_score, nodes)
states, costs = [state], [cost]
for step in range(maxsteps):
print(step)
fraction = step / float(maxsteps)
T = temperature(fraction)
#[new_state,new_cost]=pertubacao(deep_copy(state),deep_copy(cost),prob,max_valor,min_valor,bic_score,nodes,nao_dag)
[new_state,
new_cost] = mutacao(deep_copy(state), deep_copy(cost), prob,
max_valor, min_valor, bic_score, nodes, nao_dag)
#new_cost = cost_function(new_state,bic_score,nodes)
#if debug: print("Step #{:>2}/{:>2} : T = {:>4.3g}, state = {:>4.3g}, cost = {:>4.3g}, new_state = {:>4.3g}, new_cost = {:>4.3g} ...".format(step, maxsteps, T, state, cost, new_state, new_cost))
if acceptance_probability(cost, new_cost, T) > random.random():
state1 = new_state.copy()
cost = deep_copy(new_cost)
states.append(state1)
costs.append(cost)
state = deep_copy(state1)
# print(" ==> Accept it!")
# else:
# print(" ==> Reject it...")
return state, cost_function(state, bic_score, nodes), states, costs
def clone(self) -> object:
return copy.deep_copy(self)
'passwd'.center(50, '-')
dir(str)
dir()
import utils
help(super)
l = [['peter', 'tim', 'kimberly'], ['la', 'ma', 'ca']]
b_l = l
del(l[1])
b_l
#shallow copy
id(l)
id(b_l)
l
b = l
import copy
dl = copy.deep_copy(l)
dl = copy.deepcopy(l)
l
import copy
dl = copy.deepcopy(l)
id(l)
id(dl)
dl
l
l.append('peter')
l
dl
import filecmp
help(filecmp.cmp)
import shutils
import shutil
def insertions(self):
return copy.deep_copy(self.text)