Probabilistic Optimization API

API reference for probabilistic optimization algorithms in opt.probabilistic.

Module Overview

from opt.probabilistic import (
    EstimationOfDistribution,
    CrossEntropy,
    BayesianOptimization,
)

Common Interface

class ProbabilisticOptimizer(AbstractOptimizer):
    def __init__(
        self,
        func: Callable,
        lower_bound: float,
        upper_bound: float,
        dim: int,
        max_iter: int,
        population_size: int = 50,
        **kwargs
    ):
        pass

    def search(self) -> tuple[np.ndarray, float]:
        pass

Available Algorithms

• EstimationOfDistribution - Builds probability distribution of solutions
• CrossEntropy - Adaptive importance sampling
• BayesianOptimization - Gaussian process-based

Example Usage

from opt.probabilistic import EstimationOfDistribution
from opt.benchmark.functions import rastrigin

# Estimation of Distribution Algorithm
eda = EstimationOfDistribution(
    func=rastrigin,
    lower_bound=-5.12,
    upper_bound=5.12,
    dim=10,
    max_iter=100,
    population_size=100,
    selection_size=20  # Number of best individuals for distribution update
)
solution, fitness = eda.search()
print(f"Best fitness: {fitness:.6e}")

Bayesian Optimization Example

from opt.probabilistic import BayesianOptimization
from opt.benchmark.functions import ackley

# Bayesian Optimization (typically for expensive functions)
bo = BayesianOptimization(
    func=ackley,
    lower_bound=-32.768,
    upper_bound=32.768,
    dim=5,  # Lower dimensions recommended for BO
    max_iter=50,  # Fewer iterations than other methods
    n_initial_points=10,
    acquisition='ei'  # Expected Improvement
)
solution, fitness = bo.search()

See Also

• Probabilistic Optimization - Algorithm details