Added weighted crossover and then some
- Added weighted methods. (default setting does not change). - Removed unnecessary exception catching. Letting exceptions flow through instead, and let the user decide how to handle them. - Removed list conversion. - Removed parent sorting for extrapolate. - Simplified variable names.
This commit is contained in:
SimpleArt
@ -7,51 +7,10 @@ def append_to_next_population(population_method):
|
||||
|
||||
return lambda ga:\
|
||||
ga.population.append_children(
|
||||
list(population_method(ga, ga.population.mating_pool))
|
||||
population_method(ga, ga.population.mating_pool)
|
||||
)
|
||||
|
||||
|
||||
def weak_check_exceptions(population_method):
|
||||
"""Checks if the first and last chromosomes can be crossed."""
|
||||
|
||||
def new_method(ga, mating_pool):
|
||||
|
||||
# check if any genes are an Exception from
|
||||
# crossing just the first and last parents.
|
||||
for gene in ga.crossover_individual_impl(ga, mating_pool[0], mating_pool[-1]):
|
||||
if isinstance(gene.value, Exception):
|
||||
raise gene.value
|
||||
|
||||
# continue if no Exceptions found.
|
||||
else:
|
||||
return population_method(ga, ga.population.mating_pool)
|
||||
|
||||
return new_method
|
||||
|
||||
|
||||
def strong_check_exceptions(population_method):
|
||||
"""Checks if every pair of selected chromosomes can be crossed.
|
||||
Warning: Very slow, consider comparing the types of genes
|
||||
allowed to the method used beforehand instead.
|
||||
"""
|
||||
|
||||
def new_method(ga, mating_pool):
|
||||
|
||||
next_population = list(population_method(ga, ga.population.mating_pool))
|
||||
|
||||
# check if any genes are an Exception.
|
||||
for chromosome in next_population:
|
||||
for gene in chromosome:
|
||||
if isinstance(gene.value, Exception):
|
||||
raise gene.value
|
||||
|
||||
# continue if no Exceptions found.
|
||||
else:
|
||||
return next_population
|
||||
|
||||
return new_method
|
||||
|
||||
|
||||
def genes_to_chromosome(individual_method):
|
||||
"""Converts a collection of genes into a chromosome.
|
||||
Note: Will recreate the gene list if given gene list.
|
||||
@ -59,9 +18,9 @@ def genes_to_chromosome(individual_method):
|
||||
and use yield for efficiency.
|
||||
"""
|
||||
|
||||
return lambda ga, parent_1, parent_2:\
|
||||
return lambda ga, parent_1, parent_2, weight:\
|
||||
ga.make_chromosome(
|
||||
list(individual_method(ga, parent_1, parent_2))
|
||||
individual_method(ga, parent_1, parent_2, weight)
|
||||
)
|
||||
|
||||
|
||||
@ -70,11 +29,11 @@ def values_to_genes(individual_method):
|
||||
Returns a generator of genes to avoid storing a new list.
|
||||
"""
|
||||
|
||||
return lambda ga, parent_1, parent_2:\
|
||||
return lambda ga, parent_1, parent_2, weight:\
|
||||
(
|
||||
ga.make_gene(value)
|
||||
for value
|
||||
in individual_method(ga, parent_1, parent_2)
|
||||
in individual_method(ga, parent_1, parent_2, weight)
|
||||
)
|
||||
|
||||
|
||||
@ -82,8 +41,6 @@ class Crossover_Methods:
|
||||
|
||||
# Private method decorators, see above.
|
||||
_append_to_next_population = append_to_next_population
|
||||
_weak_check_exceptions = weak_check_exceptions
|
||||
_strong_check_exceptions = strong_check_exceptions
|
||||
_genes_to_chromosome = genes_to_chromosome
|
||||
_values_to_genes = values_to_genes
|
||||
|
||||
@ -93,7 +50,6 @@ class Crossover_Methods:
|
||||
|
||||
|
||||
@append_to_next_population
|
||||
@weak_check_exceptions
|
||||
def sequential_selection(ga, mating_pool):
|
||||
"""Select sequential pairs from the mating pool.
|
||||
Every parent is paired with the previous parent.
|
||||
@ -104,12 +60,12 @@ class Crossover_Methods:
|
||||
yield ga.crossover_individual_impl( # apply crossover to
|
||||
ga, #
|
||||
mating_pool[index], # the parent and
|
||||
mating_pool[index-1] # the previous parent
|
||||
mating_pool[index-1], # the previous parent
|
||||
0.5 # with equal weight
|
||||
)
|
||||
|
||||
|
||||
@append_to_next_population
|
||||
@weak_check_exceptions
|
||||
def random_selection(ga, mating_pool):
|
||||
"""Select random pairs from the mating pool.
|
||||
Every parent is paired with a random parent.
|
||||
@ -119,7 +75,8 @@ class Crossover_Methods:
|
||||
yield ga.crossover_individual_impl( # apply crossover to
|
||||
ga, #
|
||||
parent, # the parent and
|
||||
random.choice(mating_pool) # a random parent
|
||||
random.choice(mating_pool), # a random parent
|
||||
0.5 # with equal weight
|
||||
)
|
||||
|
||||
|
||||
@ -128,25 +85,36 @@ class Crossover_Methods:
|
||||
|
||||
|
||||
@genes_to_chromosome
|
||||
def single_point(ga, parent_1, parent_2):
|
||||
def single_point(ga, parent_1, parent_2, weight = 0.5):
|
||||
"""Cross two parents by swapping genes at one random point."""
|
||||
|
||||
swap_index = random.randrange(len(parent_1))
|
||||
N = min(len(parent_1), len(parent_2))
|
||||
|
||||
if weight == 0.5:
|
||||
swap_index = random.randrange(N)
|
||||
else:
|
||||
weights = [
|
||||
weight*n + (1-weight)*(N-n)
|
||||
for n
|
||||
in range(N)
|
||||
]
|
||||
swap_index = random.choices(range(N), weights)[0]
|
||||
|
||||
return parent_1[:swap_index] + parent_2[swap_index:]
|
||||
|
||||
|
||||
@genes_to_chromosome
|
||||
def multi_point(ga, parent_1, parent_2):
|
||||
def multi_point(ga, parent_1, parent_2, weight = 0.5):
|
||||
"""Cross two parents by swapping genes at multiple points."""
|
||||
pass
|
||||
|
||||
|
||||
@genes_to_chromosome
|
||||
def uniform(ga, parent_1, parent_2):
|
||||
def uniform(ga, parent_1, parent_2, weight = 0.5):
|
||||
"""Cross two parents by swapping all genes randomly."""
|
||||
|
||||
for gene_pair in zip(parent_1, parent_2):
|
||||
yield random.choice(gene_pair)
|
||||
yield random.choice(gene_pair, [weight, 1-weight])
|
||||
|
||||
|
||||
class Arithmetic:
|
||||
@ -154,61 +122,56 @@ class Crossover_Methods:
|
||||
|
||||
@genes_to_chromosome
|
||||
@values_to_genes
|
||||
def random(ga, parent_1, parent_2):
|
||||
def random(ga, parent_1, parent_2, weight = 0.5):
|
||||
"""Cross two parents by taking a random integer or float value between each of the genes."""
|
||||
|
||||
value_iter_1 = parent_1.gene_value_iter
|
||||
value_iter_2 = parent_2.gene_value_iter
|
||||
values_1 = parent_1.gene_value_iter
|
||||
values_2 = parent_2.gene_value_iter
|
||||
|
||||
for value_1, value_2 in zip(values_1, values_2):
|
||||
|
||||
value = weight*values_1 + (1-weight)*random.uniform(value_1, value_2)
|
||||
|
||||
for value_1, value_2 in zip(value_iter_1, value_iter_2):
|
||||
if type(value_1) == type(value_2) == int:
|
||||
yield random.randint(*sorted([value_1, value_2]))
|
||||
else:
|
||||
try:
|
||||
yield random.uniform(value_1, value_2)
|
||||
except:
|
||||
yield ValueError("Unhandled gene type found. Use integer or float genes.")
|
||||
value = round(value)
|
||||
|
||||
yield value
|
||||
|
||||
|
||||
@genes_to_chromosome
|
||||
@values_to_genes
|
||||
def average(ga, parent_1, parent_2):
|
||||
def average(ga, parent_1, parent_2, weight = 0.5):
|
||||
"""Cross two parents by taking the average of the genes."""
|
||||
|
||||
value_iter_1 = parent_1.gene_value_iter
|
||||
value_iter_2 = parent_2.gene_value_iter
|
||||
values_1 = parent_1.gene_value_iter
|
||||
values_2 = parent_2.gene_value_iter
|
||||
|
||||
for value_1, value_2 in zip(values_1, values_2):
|
||||
|
||||
value = weight*value_1 + (1-weight)*value_2
|
||||
|
||||
for value_1, value_2 in zip(value_iter_1, value_iter_2):
|
||||
if type(value_1) == type(value_2) == int:
|
||||
yield (value_1+value_2)//2
|
||||
else:
|
||||
try:
|
||||
yield (value_1+value_2)/2
|
||||
except:
|
||||
raise ValueError("Could not take the average of the gene values. Use integer or float genes.")
|
||||
value = round(value)
|
||||
|
||||
yield value
|
||||
|
||||
|
||||
@genes_to_chromosome
|
||||
@values_to_genes
|
||||
def extrapolate(ga, parent_1, parent_2):
|
||||
"""Cross two parents by extrapolating towards the better parent.
|
||||
def extrapolate(ga, parent_1, parent_2, weight = 0.5):
|
||||
|
||||
"""Cross two parents by extrapolating towards the first parent.
|
||||
May result in gene values outside the expected domain.
|
||||
"""
|
||||
|
||||
# Swap so that parent 1 is the better parent.
|
||||
if ga.target_fitness_type == 'min' and parent_1.fitness > parent_2.fitness:
|
||||
parent_1, parent_2 = parent_2, parent_1
|
||||
if ga.target_fitness_type == 'max' and parent_1.fitness < parent_2.fitness:
|
||||
parent_1, parent_2 = parent_2, parent_1
|
||||
values_1 = parent_1.gene_value_iter
|
||||
values_2 = parent_2.gene_value_iter
|
||||
|
||||
value_iter_1 = parent_1.gene_value_iter
|
||||
value_iter_2 = parent_2.gene_value_iter
|
||||
for value_1, value_2 in zip(values_1, values_2):
|
||||
|
||||
value = (2-weight)*value_1 + (weight-1)*value_2
|
||||
|
||||
for value_1, value_2 in zip(value_iter_1, value_iter_2):
|
||||
if type(value_1) == type(value_2) == int:
|
||||
yield value_1 + (value_1-value_2)//4
|
||||
else:
|
||||
try:
|
||||
yield value_1 + (value_1-value_2)/4
|
||||
except:
|
||||
raise ValueError("Could not take the average of the gene values. Use integer or float genes.")
|
||||
value = round(value)
|
||||
|
||||
yield value
|
||||
|
||||
Reference in New Issue
Block a user