class GenePopulation:
def __init__(self,
file_name_preq,
file_name_opt,
heteroatom_list,
epochs,
percents,
population_size=4,
code_mode='binary'):
self.population_size = population_size
self.code = Code(file_name_preq, file_name_opt, heteroatom_list,
code_mode)
self.epochs = epochs
self.percents = percents
self.percents = self.distribute()
self.population = self.code.generate_population(self.population_size)
self.current_epoch = 0
self.heteroatom_list = heteroatom_list
self.chemistry = Chemistry(self.population_size)
self._write_file()
self._write_head_lines()
def _write_file(self):
os.makedirs('generations/generation{}'.format(self.current_epoch))
for i, v in enumerate(self.population):
self.code.generate_file(self.current_epoch, i, v)
time.sleep(0.1)
def _write_head_lines(self):
with open('log.txt', 'w') as f:
f.write('***********************************************\n')
f.write('Genetic Algorithm in search of optimal material\n')
f.write('***********************************************\n')
f.write('initial population size: {}, maximum epoch: {}\n'.format(
self.population_size, self.epochs))
f.write('***********************************************\n')
f.write('heteroatom_list: {}\n'.format(self.heteroatom_list))
f.write('***********************************************\n')
f.write(
'population distribution percent:\n fit: {}; switch: {}; mutate: {}; random: {}\n'
.format(self.percents[0], self.percents[1], self.percents[2],
self.percents[3]))
f.write('***********************************************\n')
time.sleep(0.1)
def _write_fitness(self, cal_results):
with open('log.txt', 'a') as f:
f.write('***********************************************\n')
f.writelines(['\n', 'epoch: {}\n'.format(self.current_epoch)])
for i, result in enumerate(cal_results):
f.write('No.{} individual {}, fitness {}\n'.format(
i, result[0], result[1]))
f.write('***********************************************\n')
time.sleep(0.1)
def cal_fitness(self, percent):
self.write_message('starting to cal fitness')
cal_results = self.chemistry.cal_fitness(self.current_epoch)
self._write_fitness(cal_results)
new_population = []
for i in range(int(percent), 0, -1):
idx = list(cal_results.items())[0]
new_population.append(self.population[idx])
self.population = new_population
def switch(self, percent):
self.write_message('starting to switch')
new_population_list = []
for i in range(int(percent)):
random_idx = np.random.randint(0, len(self.population), 2)
rand_point = np.random.randint(0, self.code.get_sequence_len(), 1)
new_individual = self.population[random_idx[0]]
new_individual[rand_point[0]:] = self.population[
random_idx[1]][rand_point[0]:]
new_population_list.append(new_individual)
self.population.extend(new_population_list)
def mutate(self, percent):
self.write_message('starting to mutate')
new_population_list = []
for i in range(int(percent)):
new_individual = self.population[np.random.randint(
0, len(self.population), 1)[0]]
mutate_idx = np.random.randint(0, self.code.get_sequence_len(), 1)
new_individual[mutate_idx[0]] = self.code.random_mutation(
new_individual[mutate_idx[0]])
new_population_list.append(new_individual)
self.population.extend(new_population_list)
def random(self, percent):
self.write_message('starting to random')
new_population = self.code.generate_population(int(percent))
self.population.extend(new_population)
def iterate(self):
while self.current_epoch <= self.epochs:
self.cal_fitness(self.percents[0])
self.switch(self.percents[1])
self.mutate(self.percents[2])
self.random(self.percents[3])
self.current_epoch += 1
self._write_file()
@staticmethod
def write_message(message):
with open('log.txt', 'a') as f:
f.write('***********************************************\n')
f.write(message + '\n')
time.sleep(0.1)
def distribute(self):
leftover = 0.0
distributed_total = []
for weight in self.percents:
weight = float(weight)
leftover, weighted_value = modf(weight * self.population_size +
leftover)
distributed_total.append(weighted_value)
distributed_total[-1] = round(distributed_total[-1] +
leftover) # mitigate round off errors
distributed_total[0], distributed_total[-1] = distributed_total[
-1], distributed_total[0]
return distributed_total
