Skip to content

Commit

Permalink
refactored ga initialisation
Browse files Browse the repository at this point in the history
  • Loading branch information
@tkchafin
tkchafin committed May 8, 2024
1 parent 0ed6af8 commit 34272ef
Show file tree
Hide file tree
Showing 2 changed files with 146 additions and 56 deletions.
200 changes: 145 additions & 55 deletions src/resistnet/model_optimisation.py
View file
Original file line number Diff line number Diff line change
Expand Up @@ -74,6 +74,7 @@ def __init__(self, resistance_network, seed=1234, verbose=True):
self.only_keep = None
self.report_all = None
self.use_full = False
self.fixed_parameters = None

def initialize_ga(self):
"""
Expand All @@ -98,7 +99,6 @@ def initialize_ga(self):

if self.verbose:
print("Establishing a population of size:", str(popsize))
print(popsize)
self.population = self.toolbox.population(n=popsize)

def set_ga_parameters(self, mutpb, cxpb, indpb, popsize, maxpopsize,
Expand Down Expand Up @@ -578,57 +578,141 @@ def initialise_population(self):
def init_ga_attributes(self):
"""
Initializes the genetic algorithm's attributes and registers necessary
functions with the toolbox.
functions with the toolbox, allowing dynamic specification of fixed
parameters per variable and attribute.
"""
# Register attributes
self.toolbox.register("feature_sel", random.randint, 0, 1)

# weight
if self.posWeight:
self.toolbox.register(
"feature_weight", random.uniform, self.min_weight, 1.0
)
if self.fixWeight:
self.toolbox.register("feature_weight", random.uniform, 1.0, 1.0)
if not self.fixWeight and not self.posWeight:
self.toolbox.register("feature_weight", random.uniform, -1.0, 1.0)

# asymmetry parameter
if not self.fixAsym:
self.toolbox.register("feature_asym", random.randint, 0, 1)
else:
self.toolbox.register("feature_asym", random.randint, 0, 0)
# Helper functions to generate attributes
def feature_generator(
var_name, feature_type, lower, upper, is_int=False):
if (self.fixed_parameters and
var_name in self.fixed_parameters and
feature_type in self.fixed_parameters[var_name]):
# Return a lambda that always returns the fixed value
return lambda: self.fixed_parameters[var_name][feature_type]
elif is_int:
return lambda: random.randint(lower, upper)
else:
return lambda: random.uniform(lower, upper)

# shape
if not self.fixShape:
self.toolbox.register("feature_transform", random.randint, 0, 8)
# Registering attribute generators for each variable and each attribute
for var_name in self.resistance_network.variables:
self.toolbox.register(
"feature_shape", random.randint, 1, int(self.max_shape)
f"{var_name}_sel", feature_generator(
var_name, "sel", 0, 1, is_int=True))
if self.posWeight:
self.toolbox.register(
f"{var_name}_weight", feature_generator(
var_name, "weight", self.min_weight, 1.0))
elif self.fixWeight:
self.toolbox.register(
f"{var_name}_weight", feature_generator(
var_name, "weight", 1.0, 1.0))
else:
self.toolbox.register(
f"{var_name}_weight", feature_generator(
var_name, "weight", -1.0, 1.0))

if not self.fixAsym:
self.toolbox.register(
f"{var_name}_asym", feature_generator(
var_name, "asym", 0, 1, is_int=True))
else:
self.toolbox.register(
f"{var_name}_asym", feature_generator(
var_name, "asym", 0, 0, is_int=True))

if not self.fixShape:
self.toolbox.register(
f"{var_name}_transform", feature_generator(
var_name, "transform", 0, 8, is_int=True))
self.toolbox.register(
f"{var_name}_shape", feature_generator(
var_name, "shape", 1, self.max_shape, is_int=True))
else:
self.toolbox.register(
f"{var_name}_transform", feature_generator(
var_name, "transform", 0, 0, is_int=True))
self.toolbox.register(
f"{var_name}_shape", feature_generator(
var_name, "shape", 1, 1, is_int=True))

# Define how to create an individual
def create_individual():
return creator.Individual(
[getattr(self.toolbox, f"{var_name}_{attr}")()
for var_name in self.resistance_network.variables
for attr in [
'sel', 'weight', 'transform', 'shape', 'asym']]
)
else:
self.toolbox.register("feature_transform", random.randint, 0, 0)
self.toolbox.register("feature_shape", random.randint, 1, 1)

# Registering individual and population creation methods
self.toolbox.register("individual", create_individual)
self.toolbox.register(
"individual",
tools.initCycle,
creator.Individual,
(self.toolbox.feature_sel,
self.toolbox.feature_weight,
self.toolbox.feature_transform,
self.toolbox.feature_shape,
self.toolbox.feature_asym),
n=len(self.resistance_network.variables)
)
self.toolbox.register(
"population", tools.initRepeat, list, self.toolbox.individual
)
self.toolbox.register("evaluate", lambda ind: ind)
"population", tools.initRepeat, list, self.toolbox.individual)

# Evaluation, mating, mutation, and selection methods remain unchanged
self.toolbox.register("evaluate", lambda ind: ind) # spoof
self.toolbox.register("mate", tools.cxTwoPoint)
self.toolbox.register("mutate", self.mutate)
self.toolbox.register(
"select", tools.selTournament, tournsize=self.tournsize
)
"select", tools.selTournament, tournsize=self.tournsize)

# def init_ga_attributes(self):
# """
# Initializes the genetic algorithm's attributes and registers
# necessary functions with the toolbox.
# """
# # Register attributes
# self.toolbox.register("feature_sel", random.randint, 0, 1)

# # weight
# if self.posWeight:
# self.toolbox.register(
# "feature_weight", random.uniform, self.min_weight, 1.0
# )
# if self.fixWeight:
# self.toolbox.register("feature_weight", random.uniform, 1.0, 1.0)
# if not self.fixWeight and not self.posWeight:
# self.toolbox.register(
# "feature_weight", random.uniform, -1.0, 1.0)

# # asymmetry parameter
# if not self.fixAsym:
# self.toolbox.register("feature_asym", random.randint, 0, 1)
# else:
# self.toolbox.register("feature_asym", random.randint, 0, 0)

# # shape
# if not self.fixShape:
# self.toolbox.register("feature_transform", random.randint, 0, 8)
# self.toolbox.register(
# "feature_shape", random.randint, 1, int(self.max_shape)
# )
# else:
# self.toolbox.register("feature_transform", random.randint, 0, 0)
# self.toolbox.register("feature_shape", random.randint, 1, 1)

# self.toolbox.register(
# "individual",
# tools.initCycle,
# creator.Individual,
# (self.toolbox.feature_sel,
# self.toolbox.feature_weight,
# self.toolbox.feature_transform,
# self.toolbox.feature_shape,
# self.toolbox.feature_asym),
# n=len(self.resistance_network.variables)
# )
# self.toolbox.register(
# "population", tools.initRepeat, list, self.toolbox.individual
# )
# self.toolbox.register("evaluate", lambda ind: ind)
# self.toolbox.register("mate", tools.cxTwoPoint)
# self.toolbox.register("mutate", self.mutate)
# self.toolbox.register(
# "select", tools.selTournament, tournsize=self.tournsize
# )

def terminate_workers(self):
"""
Expand Down Expand Up @@ -766,24 +850,30 @@ def worker_task(task_queue, result_queue, worker_args, seed):

def mutate(self, individual):
"""
Custom mutation function for an individual.
Custom mutation function for an individual that aligns with the dynamic
toolbox registration.
"""
num_attributes = 5
num_vars = len(individual) // num_attributes

Args:
individual: An individual to be mutated.
for var_index in range(num_vars):
var_name = self.resistance_network.variables[var_index]
idx_base = var_index * num_attributes

Returns:
A tuple containing the mutated individual.
"""
length = len(individual) // 5 # number of covariates
for i in range(length):
if random.random() < self.mutpb:
individual[i * 5] = self.toolbox.feature_sel()
individual[idx_base] = getattr(
self.toolbox, f"{var_name}_sel")()
if random.random() < self.mutpb:
individual[i * 5 + 1] = self.toolbox.feature_weight()
individual[idx_base + 1] = getattr(
self.toolbox, f"{var_name}_weight")()
if random.random() < self.mutpb:
individual[i * 5 + 2] = self.toolbox.feature_transform()
individual[idx_base + 2] = getattr(
self.toolbox, f"{var_name}_transform")()
if random.random() < self.mutpb:
individual[i * 5 + 3] = self.toolbox.feature_shape()
individual[idx_base + 3] = getattr(
self.toolbox, f"{var_name}_shape")()
if random.random() < self.mutpb:
individual[i * 5 + 4] = self.toolbox.feature_asym()
individual[idx_base + 4] = getattr(
self.toolbox, f"{var_name}_asym")()

return individual,
2 changes: 1 addition & 1 deletion src/resistnet/params.py
View file
Original file line number Diff line number Diff line change
Expand Up @@ -90,7 +90,7 @@ def __init__(self):
self.max_shape = 100
self.min_weight = 0.0

# new model averaging options
# new model averaging options
self.use_full = False

# SAMC options
Expand Down

0 comments on commit 34272ef

Please sign in to comment.