Module bayesian_opt 

Bayesian Optimization with Gaussian Process Surrogate

Implements Bayesian Optimization for black-box function minimization using a GP surrogate model with RBF or Matern 5/2 kernels. Supports EI, PI, UCB, and Thompson Sampling acquisition functions with hyperparameter optimization via marginal log-likelihood.

Example

use numrs2::optimize::bayesian_opt::*;

let f = |x: &[f64]| -> f64 { x.iter().map(|xi| (xi - 0.5).powi(2)).sum() };
let config = BayesOptConfig {
    bounds: vec![(0.0, 1.0), (0.0, 1.0)],
    n_initial: 5, n_iterations: 20,
    ..Default::default()
};
let result = bayesian_optimize(f, config).expect("optimization should succeed");
assert!(result.f_best < 0.1);

Structs

BayesOptConfig

Configuration for Bayesian Optimization.

BayesOptResult

Result of Bayesian Optimization.

GaussianProcess

Gaussian Process (GP) surrogate model for function approximation.

Enums

AcquisitionType

Acquisition function type for guiding optimization.

KernelType

Kernel function type for the GP surrogate model.

Functions

bayesian_optimize

Run Bayesian Optimization to minimize the given objective function.