from random import randint
from typing import Tuple, Optional, List
import numpy as np
from evolutionary_optimization.genotype.genotype_model.abstract_genotype import AbstractGenotype
from evolutionary_optimization.genotype.genotype_model.genotype_utils import single_point_crossover
[docs]class BinaryListGenotype(AbstractGenotype):
[docs] def __init__(
self,
genotype: Optional[List[int]] = None,
mutation_probability: float = 0.5,
ratio_of_population_for_crossover: float = 0.5,
number_of_genes: int = 32,
value_range: Optional[Tuple[int, int]] = (0, 1),
):
"""Initialise instance of AbstractGenotype.
Args:
genotype: genotype used for mutation, crossover and to calculate phenotype_value in binary form.
mutation_probability: probability of a gene mutating.
ratio_of_population_for_crossover: ratio of population used for crossover when updating population.
number_of_genes: number of genes in the genotype.
value_range: minimum and maximum values of a gene.
.. warning::
This genotype can only be used for single value optimization e.g. for a ParabolaPhenotype,
and would not work for BoothPhenotype.
"""
self.binary_genotype = genotype
self._genotype = self.return_integer_form()
self.mutation_probability = mutation_probability
self.ratio_of_population_for_crossover = ratio_of_population_for_crossover
self.number_of_genes = number_of_genes
self.value_range = (0, 1)
@property
def genotype(self):
"""Genotype value used for evaluation of phenotype.
.. warning::
This is the integer form of the self.binary_genotype."""
return self._genotype
@genotype.setter
def genotype(self, value):
"""Genotype attribute setter."""
self._genotype = value
[docs] @classmethod
def build_random_genotype(
cls,
number_of_genes: int = 32,
value_range: Tuple[int, int] = (0, 1),
mutation_probability: Optional[float] = 0.1,
ratio_of_population_for_crossover: Optional[float] = 0.5
) -> "BinaryListGenotype":
"""Build random genotype attribute based on requirements.
Args:
number_of_genes: number of genes in the genotype.
value_range: minimum and maximum values of a gene.
mutation_probability: probability of a gene mutating.
ratio_of_population_for_crossover: ratio of population used for crossover when updating population.
Returns:
Genotype object with updated genotype attribute.
"""
genotype = []
for i in range(number_of_genes):
new_gene = randint(0, 1)
genotype.append(new_gene)
return cls(
genotype=genotype,
mutation_probability=mutation_probability,
ratio_of_population_for_crossover=ratio_of_population_for_crossover,
number_of_genes=number_of_genes,
value_range=value_range,
)
[docs] @classmethod
def from_genotype(cls, base_genotype: "BinaryListGenotype", new_genotype: List[int]) -> "BinaryListGenotype":
"""Create a new genotype using the parameters of an existing genotype."""
return cls(
genotype=new_genotype,
value_range=base_genotype.value_range,
mutation_probability=base_genotype.mutation_probability,
ratio_of_population_for_crossover=base_genotype.ratio_of_population_for_crossover
)
[docs] def mutate(self):
"""In place modification of the genotype by randomly changing genes based on mutation probability."""
mutation_mask = np.random.choice([True, False],
p=[self.mutation_probability, 1 - self.mutation_probability],
size=len(self.binary_genotype))
logical_binary_genotype = np.array(self.binary_genotype).astype(bool)
mutated_logical_binary_genotype = np.logical_xor(logical_binary_genotype, mutation_mask)
mutated_binary_genotype = mutated_logical_binary_genotype.astype(int)
self.binary_genotype = list(mutated_binary_genotype)
self.genotype = self.return_integer_form()
[docs] def crossover(
self,
parent_2_genotype: "BinaryListGenotype",
) -> Tuple["BinaryListGenotype", "BinaryListGenotype"]:
"""Performs single point crossover operation for 1 set of parents.
A random integer is generated to split the genotype of the two individuals -
this is the gene slice index. Then two child genotypes are generated with the complementary parts
of the parent genotypes. If the parent's genotype length is 1, crossover is impossible so the parent
instances are returned.
Example:
parent_1.genotype = [1, 2, 3, 4]
parent_2.genotype = [A, B, C, D]
gene_slice_index = 1
child_1.genotype = [1, B, C, D]
child_2.genotype = [A, 2, 3, 4]
Args:
parent_2_genotype: Individual which will be used to create an offspring.
Returns:
Tuple of AbstractGenotype, representing two children genotypes that are a combination of the parents.
"""
if len(self.binary_genotype) != len(parent_2_genotype.binary_genotype):
raise NameError("The Individuals have genotypes of different lengths - crossover is impossible")
if self.number_of_genes == 1:
return self, parent_2_genotype
else:
last_slice_index = self.number_of_genes - 1
gene_slice_index = randint(1, last_slice_index)
child_1_genotype = single_point_crossover(self.binary_genotype, parent_2_genotype.binary_genotype, gene_slice_index)
child_2_genotype = single_point_crossover(parent_2_genotype.binary_genotype, self.binary_genotype, gene_slice_index)
child_1 = self.from_genotype(parent_2_genotype, child_1_genotype)
child_2 = self.from_genotype(parent_2_genotype, child_2_genotype)
return child_1, child_2