def crossoverAndMutation(problem, individuals):
# Format a population data structure usable by DEAP's package
dIndividuals = deap_format(problem, individuals)
# Crossover
for ind1, ind2 in zip(dIndividuals[::2], dIndividuals[1::2]):
if random.random() <= 0.9: #crossover rate
tools.cxUniform(ind1, ind2, indpb=1.0 / len(problem.decisions))
# Mutation
for ind in dIndividuals:
tools.mutPolynomialBounded(ind,
eta=1.0,
low=[dec.low for dec in problem.decisions],
up=[dec.up for dec in problem.decisions],
indpb=0.1)
del ind.fitness.values
# Update beginning population data structure
for individual, dIndividual in zip(individuals, dIndividuals):
for i in range(len(individual.decisionValues)):
individual.decisionValues[i] = dIndividual[i]
individual.fitness = None
return individuals, 0
def mutate(self, indiv):
""" Mutate the individual design variables with different strategies according to the variables types :
- interval : Polynomial Bounded mutation or user defined
- set : Resampling the variable according to its initialization function
- pyleecan : Resampling the variable according to its initialization function
Parameters
----------
solver : Solver
solver to perform the genetic algorithm with DEAP
indiv : individual (e.g. OptiGenAlgIndivDeap)
individual to mutate
Returns
-------
is_mutation : boolean
True if at least one mutation occured
"""
is_mutation = False
for k in range(len(indiv.keys)):
if self.p_mutate < random.random(): # Perform mutation
is_mutation = True
if self.mutator == None:
if (
indiv.design_var[indiv.keys[k]].type_var == "interval"
): # Interval variable
# Using polynomial bounded mutation as in Deb and al., "A Fast and Elitist Multiobjective Genetic Algorithm: NSGA-II"
if isinstance(indiv[k], Iterable):
indiv[k] = mutPolynomialBounded(
indiv[k], 20, *indiv.design_var[indiv.keys[k]].space, 1
)[0]
else:
# Function takes list in argument and returns list
indiv[k] = mutPolynomialBounded(
[indiv[k]], 20, *indiv.design_var[indiv.keys[k]].space, 1
)[0][0]
else:
indiv[k] = indiv.design_var[indiv.keys[k]].function(
indiv.design_var[indiv.keys[k]].space
)
else: # User defined mutator
if (
indiv.design_var[indiv.keys[k]].type_var == "interval"
): # Interval variable
indiv[k] = self.mutator(indiv[k])
else:
# TODO Allow to redefine mutators for pyleecan types or set
indiv[k] = indiv.design_var[indiv.keys[k]].function(
indiv.design_var[indiv.keys[k]].space
)
return is_mutation
def mutate(self, indpb):
allele, = mutPolynomialBounded(self.allele,
indpb=self.indpb,
eta=self.mteta,
low=-0.5 * self.flexibility,
up=0.5 * self.flexibility)
self.allele[:] = [round(n, self.precision) for n in allele]
def mutate(self, individual, probability):
if self.method == 'Polynomial':
return tools.mutPolynomialBounded(individual,
eta=20.0,
low=0.0,
up=1.0,
indpb=1 / len(individual))[0]
elif self.method == 'Shuffle':
return tools.mutShuffleIndexes(individual, probability)[0]
def mutate(self, indpb):
if random.random() < 0.5:
allele, = mutPolynomialBounded(
self.allele,
indpb=self.indpb,
eta=self.mteta,
low=-0.5 * self.flexibility,
up=0.5 * self.flexibility,
)
self.allele[:] = [round(n, self.precision) for n in allele]
else:
self.allele = [self.random_angle() for i in xrange(self.max_bonds)]
def my_mutation(self,LL,UU,individual):
t=self.t
typeOfInput=len(t)
individual_ = []
for i in range(typeOfInput):
if t[i] =='float':
ind1=tools.mutPolynomialBounded([individual[i]], low=LL[i], up=UU[i], eta=20.0,indpb=1.0/len(self.t))
individual_.append(ind1[0][0])
elif t[i] =='int' or t[i]== 'bool':
ind2=tools.mutUniformInt([individual[i]], low=LL[i], up=UU[i], indpb= 0.9)
individual_.append(ind2[0][0])
return individual_
def crossoverAndMutation(problem, individuals):
# Format a population data structure usable by DEAP's package
dIndividuals = deap_format(problem, individuals)
# Crossover
for ind1, ind2 in zip(dIndividuals[::2], dIndividuals[1::2]):
if random.random() <= 0.9: #crossover rate
tools.cxUniform(ind1, ind2, indpb=1.0/len(problem.decisions))
# Mutation
for ind in dIndividuals:
tools.mutPolynomialBounded(ind, eta = 1.0, low=[dec.low for dec in problem.decisions], up=[dec.up for dec in problem.decisions], indpb=0.1 )
del ind.fitness.values
# Update beginning population data structure
for individual,dIndividual in zip(individuals, dIndividuals):
for i in range(len(individual.decisionValues)):
individual.decisionValues[i] = dIndividual[i]
individual.fitness = None
return individuals,0
def mutation(individual, indpb, eta):
"""Mutates an individual using polynomial bounded mutation.
:individual: the individual to mutate
:indpb: the probability for a mutation to occur
:returns: the (altered) individual
"""
gen = tools.mutPolynomialBounded(
individual.get_values(),
eta, # eta_m 1.0 == high spread; 4.0 low spread
0.0, # lower bound
1.0, # upper bound
indpb)
individual.set_values(gen[0])
return (individual, )
def init_algorithm(self, params):
toolbox = base.Toolbox()
ind_size = self.problem.ind_size
ngen = params['generations']
nind = params['num_individuals']
cxpb, mutpb, lb, ub, mu, lam = 0.7, 0.2, -1.0, 1.0, nind, nind
if hasattr(creator, 'FitnessMin') is False:
creator.create('FitnessMin', base.Fitness, weights=(-1.0, ))
if hasattr(creator, 'Individual') is False:
kw0 = {'typecode': 'd', 'fitness': creator.FitnessMin}
creator.create('Individual', array.array, **kw0)
atr = lambda: [random.uniform(lb, ub) for _ in range(ind_size)]
ind = lambda: tools.initIterate(creator.Individual, atr)
population = [ind() for _ in range(nind)]
kw1 = {'low': lb, 'up': ub, 'eta': 20.0, 'indpb': 1.0 / ind_size}
mut = lambda xs: tools.mutPolynomialBounded(xs, **kw1)
kw2 = {'low': lb, 'up': ub, 'eta': 20.0}
crs = lambda i1, i2: tools.cxSimulatedBinaryBounded(i1, i2, **kw2)
sel = lambda p, n: tools.selTournament(p, n, tournsize=3)
toolbox.register('evaluate', self.problem.objective_function)
toolbox.register('mate', crs)
toolbox.register('mutate', mut)
toolbox.register('select', sel)
stats = tools.Statistics(lambda ind: ind.fitness.values)
stats.register('min', np.min)
self.hof = tools.HallOfFame(5)
args = (population, toolbox, mu, lam, cxpb, mutpb, ngen)
kw3 = {'stats': stats, 'halloffame': self.hof, 'verbose': True}
self.algorithm = lambda: algorithms.eaMuPlusLambda(*args, **kw3)
def mutation_main(individual, indpb, column_names, heating_unit_names_share,
cooling_unit_names_share, column_names_buildings_heating,
column_names_buildings_cooling, district_heating_network,
district_cooling_network):
# create dict of individual with his/her name
individual_with_name_dict = dict(zip(column_names, individual))
if district_heating_network:
# MUTATE BUILDINGS CONNECTED
buildings_heating = [
individual_with_name_dict[column]
for column in column_names_buildings_heating
]
# apply mutations
buildings_heating_mutated = tools.mutFlipBit(buildings_heating,
indpb)[0]
# take back to the individual
for column, mutated_value in zip(column_names_buildings_heating,
buildings_heating_mutated):
individual_with_name_dict[column] = mutated_value
# MUTATE SUPPLY SYSTEM UNITS SHARE
heating_units_share = [
individual_with_name_dict[column]
for column in heating_unit_names_share
]
NDIM = len(heating_unit_names_share)
# apply mutations
heating_units_share_mutated = tools.mutPolynomialBounded(
heating_units_share, eta=20.0, low=0.0, up=1.0,
indpb=1.0 / NDIM)[0]
# takeback to teh individual
for column, mutated_value in zip(heating_unit_names_share,
heating_units_share_mutated):
individual_with_name_dict[column] = mutated_value
if district_cooling_network:
# MUTATE BUILDINGS CONNECTED
buildings_cooling = [
individual_with_name_dict[column]
for column in column_names_buildings_cooling
]
# apply mutations
buildings_cooling_mutated = tools.mutFlipBit(buildings_cooling,
indpb)[0]
# take back to teh individual
for column, mutated_value in zip(column_names_buildings_cooling,
buildings_cooling_mutated):
individual_with_name_dict[column] = mutated_value
# MUTATE SUPPLY SYSTEM UNITS SHARE
cooling_units_share = [
individual_with_name_dict[column]
for column in cooling_unit_names_share
]
NDIM = len(cooling_units_share)
# apply mutations
cooling_units_share_mutated = tools.mutPolynomialBounded(
cooling_units_share, eta=20.0, low=0.0, up=1.0,
indpb=1.0 / NDIM)[0]
# takeback to teh individual
for column, mutated_value in zip(cooling_unit_names_share,
cooling_units_share_mutated):
individual_with_name_dict[column] = mutated_value
# now validate individual
individual_with_name_dict = validation_main(
individual_with_name_dict, column_names_buildings_heating,
column_names_buildings_cooling, district_heating_network,
district_cooling_network)
# now pass all the values mutated to the original individual
for i, column in enumerate(column_names):
individual[i] = individual_with_name_dict[column]
return individual, # add the, because deap needs this
def nsga2_1iter(self, current_pop):
"""
Run one iteration of the NSGA-II optimizer.
Based on the crossover and mutation probabilities, the
offsprings are created and evaluated. Based on the fitness
of these offsprings and of the current population,
a new population is created.
Parameters
----------
current_pop : list
The initial population.
Returns
-------
new_pop : list
The updated population.
"""
# create offspring, clone of current population
offspring = [deepcopy(ind) for ind in current_pop]
# perform crossover
for ind1, ind2 in zip(offspring[::2], offspring[1::2]):
if random.random() < self.run_dict['cx prob']:
# apply crossover operator
# in place editing of individuals
tools.cxSimulatedBinaryBounded(ind1, ind2,
self.run_dict['eta'],
self.space_obj.l_b,
self.space_obj.u_b)
# set the fitness to an empty tuple of modified individuals
del ind1.fitness.values
del ind2.fitness.values
# perform mutation
for mutant in offspring:
if random.random() < self.run_dict['mut prob']:
# apply mutation operator
# in place editing of individuals
tools.mutPolynomialBounded(mutant, self.run_dict['eta'],
self.space_obj.l_b,
self.space_obj.u_b,
self.run_dict['mut prob'])
# set fitness to empty tuple of modified individuals
del mutant.fitness.values
# evaluate the modified individuals
n_eval = 0
invalid_indices = []
invalid_fit_individuals = []
for i, ind in enumerate(offspring):
if not ind.fitness.valid:
invalid_indices.append(i)
invalid_fit_individuals.append(ind)
if len(invalid_fit_individuals) > 0:
# create samples to evaluate
individuals_to_eval, unc_samples = self.define_samples_to_eval(
invalid_fit_individuals)
# evaluate samples
fitnesses = self.evaluate_samples(individuals_to_eval)
n_eval += len(individuals_to_eval)
# assign fitness to the orginal samples list
individuals_to_assign = self.assign_fitness_to_population(
invalid_fit_individuals, fitnesses, unc_samples)
# construct offspring list
for i, ind in zip(invalid_indices, individuals_to_assign):
offspring[i] = deepcopy(ind)
# select next population using the NSGA-II operator
new_pop = tools.selNSGA2(current_pop + offspring, len(current_pop))
# update the population and fitness files
self.append_points_to_file(new_pop, 'population')
fitness_values = [x.fitness.values for x in new_pop]
self.append_points_to_file(fitness_values, 'fitness')
# update the STATUS file
ite, evals = self.parse_status()
self.write_status('%8i%8i' % (ite + 1, evals + n_eval))
return new_pop
def init_algorithm(self, params):
if params['algorithm_class'] not in ['simple', 'multiobjective']:
raise ValueError('Non-existent algorithm class.')
toolbox = base.Toolbox()
ngen = params['generations']
nind = params['num_individuals']
cxpb = 0.5 if params['algorithm_class'] == 'simple' else 0.9
lb, ub = -1.0, 1.0
ind_size = self.problem.ind_size
if nind % 4 != 0:
raise ValueError('Number of individuals must be multiple of four')
if hasattr(creator, 'FitnessMin') is False:
creator.create('FitnessMin', base.Fitness, weights=(-1.0, -1.0))
if hasattr(creator, 'Individual') is False:
creator.create('Individual',
array.array,
typecode='d',
fitness=creator.FitnessMin)
atr = lambda: [random.uniform(lb, ub) for _ in range(ind_size)]
ind = lambda: tools.initIterate(creator.Individual, atr)
population = [ind() for _ in range(nind)]
if params['algorithm_class'] == 'simple':
self.hof = tools.HallOfFame(1)
mut = lambda xs: tools.mutGaussian(xs, mu=0, sigma=1, indpb=0.1)
crs = tools.cxTwoPoint
sel = lambda p, n: tools.selTournament(p, n, tournsize=3)
else:
self.hof = tools.ParetoFront()
mut = lambda xs: tools.mutPolynomialBounded(
xs, low=lb, up=ub, eta=20.0, indpb=1.0 / ind_size)
crs = lambda ind1, ind2: tools.cxSimulatedBinaryBounded(
ind1, ind2, low=lb, up=ub, eta=20.0)
sel = tools.selNSGA2
toolbox.register('evaluate', self.problem.objective_function)
toolbox.register('mate', crs)
toolbox.register('mutate', mut)
toolbox.register('select', sel)
stats = tools.Statistics(lambda ind: ind.fitness.values)
stats.register('avg', np.mean, axis=0)
stats.register('std', np.std, axis=0)
stats.register('min', np.min, axis=0)
stats.register('max', np.max, axis=0)
args = (population, toolbox)
kwargs = {'cxpb': cxpb, 'ngen': ngen, 'stats': stats}
kwargs['halloffame'] = self.hof
if params['algorithm_class'] == 'simple':
kwargs['mutpb'] = 0.2
kwargs['verbose'] = True
self.algorithm = lambda: algorithms.eaSimple(*args, **kwargs)
else:
self.algorithm = lambda: self.multiobjective(*args, **kwargs)
def __init__(self, params: AlgorithmParameters):
"""Initialize the genetic algorithm that will solve the control synthesis problem.
The params dictionary has the following items
num_control_params The number of control parameters used in a solution.
generations The number of iterations of the genetic algorithm to run.
"""
assert len(params.lower_bounds) == len(params.upper_bounds)
self.ind_size = len(params.lower_bounds)
ngen = params.generations
nind = params.num_individuals
cxpb, mutpb, mu, lam = params.crossover_prob, params.mutation_prob, nind, nind
# Here we are creating two new types. The FitnessMin type and the Individual type.
# FitnessMin inherits from deap.base.Fitness and has a new weights attribute that is
# a tuple. For single objective optimization, the weights attribute has one element
# for multi-objective optimization it will have multiple elements. A negative weight
# indicates this is a minimization problem.
if hasattr(creator, 'FitnessMin') is False:
creator.create('FitnessMin', base.Fitness, weights=(-1.0, ))
# The individual type inherits from array.array (could use list) and it needs a
# fitness attribute to know how to calculate the fitness of the individual.
# Arrays store only one type of data. This is set by the tyepcode paramter. 'd' is for double.
if hasattr(creator, 'Individual') is False:
kw0 = {'typecode': 'd', 'fitness': creator.FitnessMin}
creator.create('Individual', array.array, **kw0)
# Creates the initial lists of individuals.
atr = lambda: [
random.uniform(lb, ub)
for lb, ub in zip(params.lower_bounds, params.upper_bounds)
]
ind = lambda: creator.Individual(atr())
population = [ind() for _ in range(nind)]
# The toolbox is a container to store partial functions. In particular we want the evaluate, mate, mutate,
# and select genetic algorithm functions defined here.
toolbox = base.Toolbox()
kw1 = {
'low': params.lower_bounds,
'up': params.upper_bounds,
'eta': 20.0,
'indpb': 1.0 / self.ind_size
}
mut = lambda xs: tools.mutPolynomialBounded(xs, **kw1)
kw2 = {
'low': params.lower_bounds,
'up': params.upper_bounds,
'eta': 20.0
}
crs = lambda i1, i2: tools.cxSimulatedBinaryBounded(i1, i2, **kw2)
sel = lambda p, n: tools.selTournament(p, n, tournsize=3)
toolbox.register('evaluate', params.cost_function)
toolbox.register('mate', crs)
toolbox.register('mutate', mut)
toolbox.register('select', sel)
# DEAP provides objects taht will record algorithm statisitics and keep trak of the best solutions.
stats = tools.Statistics(lambda ind: ind.fitness.values)
stats.register('min', np.min)
self.hof = tools.HallOfFame(5)
# Setting up the DEAP library for the genetic algorithm comes down to creating and
# executing an algorithm function. There are a number of pre-build algorithms that
# can be used rather than coding your own. The eaMuPlusLambda algorithm requires
# the parameters that have been generated above.
args = (population, toolbox, mu, lam, cxpb, mutpb, ngen)
kw3 = {'stats': stats, 'halloffame': self.hof, 'verbose': True}
self.algorithm = lambda: algorithms.eaMuPlusLambda(*args, **kw3)
def decoy_mutPoly(individual):
individual, = tools.mutPolynomialBounded(individual, 0, 0.4, 0.95, 0.4)
individual[0] = clamp(individual[0], MIN_SUPPORT_THRESHOLD, MAX_SUPPORT_THRESHOLD)
individual[1] = clamp(individual[1], MIN_CONF_THRESHOLD, MAX_CONF_THRESHOLD)
return individual,
def deepmutation(self):
self.chaneOrder()
returned = mutPolynomialBounded(self.gen[1], 20, 0, 1, 0.5)
self.gen[1] = returned[0]
def crossoverAndMutation(problem, individuals):
from copy import copy
new_individuals = individuals
if CULLING is True:
count = 0
new_crop = []
# Format a population data structure usable by DEAP's package
dIndividuals = deap_format(problem, individuals)
while True:
count += 1
if count > 10 or len(dIndividuals) == 1 or len(new_crop) == len(individuals):
break
# Crossover
for ind1, ind2 in zip(dIndividuals[::2], dIndividuals[1::2]):
if random.random() <= 0.9: #crossover rate
tools.cxUniform(ind1, ind2, indpb=1.0/len(problem.decisions))
# Mutation
for ind in dIndividuals:
temp = []
tools.mutPolynomialBounded(ind, eta = 1.0, low=[dec.low for dec in problem.decisions], up=[dec.up for dec in problem.decisions], indpb=0.1 )
del ind.fitness.values
temp = problem.evaluate(ind)
if temp[0] > jmoo_properties.CULLING_PD and temp[1] < jmoo_properties.CULLING_PF:
import numpy as np
test = np.array(ind).tolist()
# print "IND: ", ind
# print "TEST: ", test
if len(new_crop) != len(individuals):
new_crop = helper_list(new_crop, test)
# print ">> ", problem.evaluate(ind)
print "NEW CROP: ", len(new_crop)
for x in new_crop:
if x in dIndividuals:
dIndividuals.remove(x)
else:
dIndividuals.remove(random.choice(dIndividuals))
for _ in xrange(1):
print ">> ", problem.evaluate(x), x
dIndividuals.extend(new_crop)
print "Length : ", len(dIndividuals)
# Update beginning population data structure
for individual,dIndividual in zip(individuals, dIndividuals):
for i in range(len(individual.decisionValues)):
individual.decisionValues[i] = dIndividual[i]
individual.fitness = None
return individuals, len(individuals) * (count - 1)
else:
# Format a population data structure usable by DEAP's package
dIndividuals = deap_format(problem, individuals)
new_dIndividuals = []
# print "Number of Individuals: ", len(dIndividuals)
# Crossover
for ind1, ind2 in zip(dIndividuals[::2], dIndividuals[1::2]):
if random.random() <= 1: #crossover rate
# print ".",
ind1, ind2 = tools.cxUniform(ind1, ind2, indpb=1.0/len(problem.decisions))
new_dIndividuals.append(ind1)
new_dIndividuals.append(ind2)
else:
new_dIndividuals.append(ind1)
new_dIndividuals.append(ind2)
# Mutation
for ind in new_dIndividuals:
# print "+",
tools.mutPolynomialBounded(ind, eta = 1.0, low=[dec.low for dec in problem.decisions], up=[dec.up for dec in problem.decisions], indpb=0.1 )
del ind.fitness.values
# Update beginning population data structure
for individual, dIndividual in zip(new_individuals, new_dIndividuals):
for i in range(len(individual.decisionValues)):
individual.decisionValues[i] = dIndividual[i]
individual.fitness = None
return new_individuals,0
