"""Continuous genetic-algorithm optimization for the acquisition function."""
import warnings
import numpy as np
from pymoo.algorithms.soo.nonconvex.ga import GA
from pymoo.core.population import Population
from pymoo.optimize import minimize
from pymoo.termination.default import DefaultSingleObjectiveTermination
from pymoo.core.problem import Problem
from pymoo.config import Config
from sklearn.utils import check_random_state
Config.warnings["not_compiled"] = False
[docs]
class PyMOORealProblem(Problem):
"""Pymoo continuous problem definition."""
def __init__(self, n_var, xl, xu, acq_func, constraint_fn, **kwargs):
super().__init__(
n_var=n_var,
n_obj=1,
xl=xl,
xu=xu,
n_eq_constr=int(constraint_fn is not None),
)
self.acq_func = acq_func
self.constraint_fn = constraint_fn
def _evaluate(self, x, out, *args, **kwargs):
y = self.acq_func(x).reshape(-1)
out["F"] = y
if self.constraint_fn is not None:
out["H"] = self.constraint_fn(x)
[docs]
class GAPymooAcqOptimizer:
"""Continuous GA optimizer using Pymoo."""
def __init__(
self,
space,
x_init,
y_init,
pop_size: int = 100,
random_state=None,
termination_kwargs=None,
constraint_fn=None,
):
self.space = space
# Bounds in the transformed/normalized space
transformed_bounds = self.space.transformed_bounds
xl, xu = list(zip(*transformed_bounds))
xl, xu = np.array(xl), np.array(xu)
self.xl = np.array(xl)
self.xu = np.array(xu)
self.n_var = len(self.xl)
if len(y_init) != len(x_init):
raise ValueError("The initial x_init and y_init values should have the same size.")
self.x_init = np.array(x_init)
self.y_init = np.array(y_init).reshape(-1)
if len(y_init) != pop_size:
raise ValueError("The initial x_init and y_init should have a size equal to pop_size.")
self.pop_size = pop_size
self.random_state = check_random_state(random_state)
if termination_kwargs is None:
termination_kwargs = {}
default_termination_kwargs = {
"xtol": 1e-8,
"ftol": 1e-6,
"period": 30,
"n_max_gen": 1_000,
}
default_termination_kwargs.update(termination_kwargs)
self.termination_kwargs = default_termination_kwargs
self.constraint_fn = constraint_fn
[docs]
def minimize(self, acq_func):
"""Minimize the acquisition function."""
problem = PyMOORealProblem(
n_var=self.n_var,
xl=self.xl,
xu=self.xu,
acq_func=acq_func,
constraint_fn=self.constraint_fn,
)
init_pop = Population.new(
"X",
self.x_init,
"F",
self.y_init.reshape(-1),
)
with warnings.catch_warnings():
warnings.simplefilter("ignore", category=DeprecationWarning)
res = minimize(
problem,
algorithm=GA(pop_size=self.pop_size, sampling=init_pop),
termination=DefaultSingleObjectiveTermination(**self.termination_kwargs),
seed=self.random_state.randint(0, np.iinfo(np.int32).max),
verbose=False,
)
return res.X
