class GaSolverASRS(object):
# problem and configuration information
problem = None
config = None
# GA basic parameters
population_size = 0
maximum_generation = 0
crossover_list = {}
mutation_list = {}
mutation_probability = 0.0
reproduction_num = 0
replacement_num = 0
roulette_constant = 0.0
# ga operators
evaluator = None
selector = None
offspring_maker = None
offspring_changer = None
offspinrg_enhancer = None
generation_maker = None
# ga objects
population = None
# ga variables
generation_num = 1
# for log
evolution_history = None
def __init__(self, problem=None, config=None,
population_size=10, maximum_generation=5, crossover_list=None, mutation_list=None,
mutation_probability=0.1, reproduction_num=1, replacement_num=1, replacement_policy='total_elitism',
roulette_constant=2.5, initial_population=None):
self.problem = problem
self.config = config
self.population_size = population_size
self.maximum_generation = maximum_generation
self.crossover_list = crossover_list
self.mutation_list = mutation_list
self.mutation_probability = mutation_probability
self.reproduction_num = reproduction_num
self.replacement_num = replacement_num
self.roulette_constant = roulette_constant
self.population = initial_population
self.evaluator = EvaluatorASRS(self.population, self.config)
self.selector = ParentSelectorASRS(self.population_size, self.reproduction_num, self.roulette_constant)
self.offspring_maker = OffspringMakerASRS(self.crossover_list, self.reproduction_num, self.config, self.problem)
self.offspring_changer = OffspringChangerASRS(self.mutation_list, self.config, self.problem)
self.offspinrg_enhancer = OffspringEnhancerASRS(self.config, self.evaluator)
self.generation_maker = GenerationMakerASRS(replacement_policy, self.replacement_num, self.reproduction_num)
def run_ga(self):
for i in range(0, self.population_size):
chromosome = self.population.chromosome_list[i]
optimized_chromosome = self.offspinrg_enhancer.cycle_optimize(chromosome)
self.population.chromosome_list[i] = optimized_chromosome
# print '3'
self.evaluator.evaluate_population(self.population)
while self.generation_num < self.maximum_generation:
previous_best = self.population.best_fitness
parents_pair_indices_dic = self.selector.get_parents(self.population)
offspring_list = self.offspring_maker.get_offsprings(self.population.chromosome_list,
parents_pair_indices_dic)
for i in range(0, len(offspring_list)):
offspring = offspring_list[i]
while np.random.rand() < self.mutation_probability:
self.offspring_changer.location_change_mutation(offspring)
mutant = self.offspring_maker.repair_chromosome(offspring)
offspring_list[i] = mutant
for i in range(0, len(offspring_list)):
offspring = offspring_list[i]
optimized_offspring = self.offspinrg_enhancer.cycle_optimize(offspring)
self.evaluator.evaluate_chromosome(optimized_offspring)
# self.evaluator.evaluate_chromosome(offspring)
offspring_list[i] = optimized_offspring
self.generation_maker.evolve_to_next_generation(self.population, offspring_list, parents_pair_indices_dic)
current_best = self.population.best_fitness
# print "%d th gen : " % self.generation_num + str(current_best) + ", " + str(self.population.worst_fitness)
if current_best < previous_best:
print "\t%d th generation: best = %f, worst = %f" % (self.generation_num, self.population.best_fitness,
self.population.worst_fitness)
self.generation_num += 1
# print self.problem.rack.cell_info.__str__()
# print self.problem.rack.cell_info.__str__()
print "\t" + self.population.chromosome_list[0].__str__()
print ""
# self.run_simulation(self.population.chromosome_list[0])
def run_simulation(self, chromosome):
rack = self.problem.rack
print "initial rack status"
print rack.cell_info.__str__()
print " "
for gene_idx in chromosome.gene_idx_list:
gene = chromosome.idx_gene_dic.get(gene_idx)
print gene.__str__()
print "location : " + rack.cell_info[gene.x, gene.y, gene.z]
if gene.task_type == "S":
rack.cell_info[gene.x, gene.y, gene.z] = gene.item_type
else:
rack.cell_info[gene.x, gene.y, gene.z] = "empty"
print "changed rack status"
print rack.cell_info.__str__()
print " "
class GaSolverASRSwithPandas(object):
# problem and configuration information
problem = None
config = None
# GA basic parameters
population_size = 0
maximum_generation = 0
crossover_list = {}
mutation_list = {}
mutation_probability = 0.0
reproduction_num = 0
replacement_num = 0
roulette_constant = 0.0
distance_prefer_constant = 0.0
# ga operators
evaluator = None
selector = None
offspring_maker = None
offspring_changer = None
offspinrg_enhancer = None
generation_maker = None
# ga objects
population = None
# ga variables
generation_num = 1
# for log
evolution_history = None
def __init__(self, problem=None, config=None,
population_size=10, maximum_generation=5, crossover_list=None, mutation_list=None,
mutation_probability=0.1, reproduction_num=1, replacement_num=1, replacement_policy='total_elitism',
roulette_constant=2.5, distance_prefer_constant=2.0, initial_population=None):
self.problem = problem
self.config = config
self.population_size = population_size
self.maximum_generation = maximum_generation
self.crossover_list = crossover_list
self.mutation_list = mutation_list
self.mutation_probability = mutation_probability
self.reproduction_num = reproduction_num
self.replacement_num = replacement_num
self.roulette_constant = roulette_constant
self.distance_prefer_constant = distance_prefer_constant
self.population = initial_population
self.evaluator = EvaluatorASRS(self.population, self.config)
self.selector = ParentSelectorASRS(self.population_size, self.reproduction_num, self.roulette_constant)
self.offspring_maker = OffspringMakerASRSwithPandas(self.crossover_list, self.reproduction_num,
self.distance_prefer_constant, self.config, self.problem)
self.offspring_changer = OffspringChangerASRS(self.mutation_list, self.distance_prefer_constant,
self.config, self.problem)
self.offspinrg_enhancer = OffspringEnhancerASRS(self.config, self.evaluator)
self.generation_maker = GenerationMakerASRS(replacement_policy, self.replacement_num, self.reproduction_num)
def run_ga(self):
for chromosome_index in range(0, self.population_size):
# print str(chromosome_index) + " th chromosome"
chromosome = self.population.chromosome_list[chromosome_index]
self.evaluator.evaluate_chromosome_with_pandas(chromosome)
self.offspinrg_enhancer.local_search_iteration(chromosome)
self.evaluator.evaluate_population_with_pandas(self.population)
print "\t\t\t\tinit gen : " + str(self.population.best_fitness)\
+ ", " + str(self.population.worst_fitness) + ""
while self.generation_num < self.maximum_generation:
previous_best = self.population.best_fitness
parents_pair_indices_dic = self.selector.get_parents(self.population)
# current_time = time.time()
# print ""
offspring_list = self.offspring_maker.get_offsprings(self.population.chromosome_list,
parents_pair_indices_dic)
# print "\toffspring is made : " + str(time.time() - current_time)
# current_time = time.time()
# print "\noffspring generate phase\n"
for offspring_index in range(0, len(offspring_list)):
offspring = offspring_list[offspring_index]
self.evaluator.evaluate_chromosome_with_pandas(offspring)
print "\t\t" + str(offspring_index) + " th offspring's initial fitness : " + str(offspring.fitness)
while np.random.rand() < self.mutation_probability:
# current_time = time.time()
self.offspring_changer.change_offspring(offspring)
# print "\tmutation : " + str(time.time() - current_time)
# current_time = time.time()
# self.evaluator.evaluate_chromosome_with_pandas(offspring)
# print "\t\t" + str(offspring_index) + " th offspring's mutation fitness : " + str(offspring.fitness)
# self.offspring_maker.repair_chromosome(offspring)
# print "\t\t" + str(offspring_index) + " th offspring's repaired fitness : " + str(offspring.fitness)
# self.evaluator.evaluate_chromosome_with_pandas(offspring)
# print "\t" + str(offspring_index) + " th offspring's repaired fitness : " + str(offspring.fitness)
self.offspinrg_enhancer.local_search_iteration(offspring)
# self.evaluator.evaluate_chromosome_with_pandas(offspring)
# print "\t" + str(offspring_index) + " th offspring's optimized fitness : " + str(offspring.fitness)
# self.evaluator.evaluate_chromosome_with_pandas(offspring)
# print str(offspring_index) + " th offspring's fitness : " + str(offspring.fitness) + "\n"
self.generation_maker.evolve_to_next_generation(self.population, offspring_list, parents_pair_indices_dic)
current_best = self.population.best_fitness
print "\t\t\t\t%d th gen : " % self.generation_num + str(current_best) + ", " + str(self.population.worst_fitness) + ""
if current_best < previous_best:
print "\t\t\t\t\t%d th generation: best = %f, worst = %f" % (self.generation_num,
self.population.best_fitness,
self.population.worst_fitness) + ""
self.generation_num += 1
# print self.problem.rack.cell_info.__str__()
# print self.problem.rack.cell_info.__str__()
print "\t" + self.population.chromosome_list[0].__str__()
print ""
# self.run_simulation(self.population.chromosome_list[0])
def run_simulation(self, chromosome):
rack = self.problem.rack
print "initial rack status"
print rack.cell_info.__str__()
print " "
for gene_idx in chromosome.gene_idx_list:
gene = chromosome.idx_gene_dic.get(gene_idx)
print gene.__str__()
print "location : " + rack.cell_info[gene.x, gene.y, gene.z]
if gene.task_type == "S":
rack.cell_info[gene.x, gene.y, gene.z] = gene.item_type
else:
rack.cell_info[gene.x, gene.y, gene.z] = "empty"
print "changed rack status"
print rack.cell_info.__str__()
print " "