def move_gene_in_chromosome(self,
original_fitness: float) -> TSP_Chromosome:
"""
Move a random gene to the point in the chromosome that will produce the best result.
"""
(best_new_chrom, best_new_fitness) = (None, None)
len_chrom = len(self)
for i in sample(range(len_chrom), min(3, len_chrom)):
gene_before = self[i - 1]
removed_gene = self[i]
i_p_1 = (i + 1) % len_chrom
gene_after = self[i_p_1]
fitness_after_removal = original_fitness - gene_before.distance_to(removed_gene) \
- removed_gene.distance_to(gene_after) \
+ gene_before.distance_to(gene_after)
# noinspection PyArgumentList
partial_chromosome: TSP_Chromosome = GA_World.chromosome_class(
self[:i] + self[i + 1:])
(new_chrom,
new_fitness) = partial_chromosome.add_gene_to_chromosome(
fitness_after_removal, removed_gene)
if best_new_fitness is None or new_fitness < best_new_fitness:
(best_new_chrom, best_new_fitness) = (new_chrom, new_fitness)
return GA_World.chromosome_class(best_new_chrom)
def gen_individual():
chromosome_list: List = sample(GA_World.gene_pool,
Cycle_World.cycle_length)
Cycle_World.individuals += 1
individual = GA_World.individual_class(
GA_World.chromosome_class(chromosome_list))
return individual
def gen_individual(self):
# Use ceil to ensure we have enough genes.
zeros = [0]*ceil(self.chromosome_length/2)
ones = [1]*ceil(self.chromosome_length/2)
mixture = sample(zeros + ones, self.chromosome_length)
chromosome_tuple: Tuple[Gene] = GA_World.chromosome_class(Gene(id, val) for (id, val) in zip(count(), mixture))
individual = GA_World.individual_class(chromosome_tuple)
return individual
def gen_individual():
path_methods = [TSP_Chromosome.random_path]
if gui_get('Greedy'):
path_methods.append(TSP_Chromosome.greedy_path)
if gui_get('Min spanning tree'):
path_methods.append(TSP_Chromosome.spanning_tree_path)
gen_path_method = choice(path_methods)
chromosome_list: List = gen_path_method()
chromo = GA_World.chromosome_class(chromosome_list)
individual = GA_World.individual_class(chromo, generator=gen_path_method, original_sequence=chromosome_list)
return individual
def mutate(self) -> Individual:
chromosome = self.chromosome
if randint(0, 100) <= SimEngine.gui_get('replace_gene'):
(chromosome, self.fitness,
_) = chromosome.replace_gene_in_chromosome(self.fitness)
elif randint(0, 100) <= SimEngine.gui_get('reverse_subseq'):
chromosome = chromosome.reverse_subseq()
self.fitness = self.compute_fitness()
self.chromosome: Chromosome = GA_World.chromosome_class(chromosome)
return self
def create_node(self):
new_point = self.create_random_agent(color=Color('white'), shape_name='node', scale=1)
new_point.set_velocity(TSP_World.random_velocity())
# If the same individual is in the population
# multiple times, don't process it more than once.
updated_chromos = set()
for ind in self.population:
old_chromo = ind.chromosome
if old_chromo not in updated_chromos:
# noinspection PyUnresolvedReferences
(new_chromo, ind.fitness) = old_chromo.add_gene_to_chromosome(ind.fitness, new_point)
# noinspection PyArgumentList
ind.chromosome = GA_World.chromosome_class(new_chromo)
updated_chromos.add(new_chromo)
def delete_node(self):
# Can't use choice with a set.
node = sample(World.agents, 1)[0]
# If the same individual is in the population
# multiple times, don't process it more than once.
updated_chromos = set()
for ind in self.population:
old_chromo = ind.chromosome
if old_chromo not in updated_chromos:
new_chromo_list = list(old_chromo)
new_chromo_list.remove(node)
new_chromo = GA_World.chromosome_class(new_chromo_list)
ind.chromosome = new_chromo
updated_chromos.add(new_chromo)
node.delete()
def update_cycle_lengths(self, cycle_length):
for ind in self.population:
# To keep PyCharm's type checker happy
assert isinstance(ind.chromosome, Loop_Chromosome)
chromosome: Loop_Chromosome = ind.chromosome
if cycle_length < len(chromosome):
new_chromosome = chromosome[:cycle_length]
ind.fitness = ind.compute_fitness()
else:
available_genes = GA_World.agents - set(chromosome)
new_genes = sample(available_genes,
cycle_length - len(chromosome))
new_chromosome = chromosome
for gene in new_genes:
(new_chromosome, ind.fitness, _) = \
chromosome.add_gene_to_chromosome(ind.fitness, gene)
ind.chromosome = GA_World.chromosome_class(new_chromosome)
def mutate(self) -> Individual:
chromosome = self.chromosome
if randint(0, 100) <= SimEngine.gui_get('exchange_genes'):
assert isinstance(self.chromosome, Segregation_Chromosome)
chromosome = chromosome.exchange_genes(self.satisfactions)
elif randint(0, 100) <= SimEngine.gui_get('move_unhappy_gene'):
candidate_indices = Segregation_Individual.unhappy_gene_value_indices(self.satisfactions)
if not candidate_indices:
return self
chromosome = chromosome.move_unhappy_gene(candidate_indices)
elif randint(0, 100) <= SimEngine.gui_get('move_gene'):
chromosome = chromosome.move_gene()
elif randint(0, 100) <= SimEngine.gui_get('reverse_subseq'):
chromosome = chromosome.reverse_subseq()
self.chromosome = GA_World.chromosome_class(chromosome)
(self.satisfactions, self.fitness) = self.chromosome.compute_chromosome_fitness()
return self
def gen_individual():
chromosome_tuple: Tuple[Gene] = GA_World.chromosome_class(
Gene(id, val) for (id, val) in zip(count(), GA_World.gene_pool))
individual = GA_World.individual_class(chromosome_tuple)
return individual
def gen_individual(self):
chromosome_list: List = sample(World.agents, self.cycle_length)
individual = GA_World.individual_class(
GA_World.chromosome_class(chromosome_list))
return individual
