Struct BayesianOptimizer 

pub struct BayesianOptimizer {
    pub bounds: Vec<(f64, f64)>,
    pub gp: GaussianProcess,
    pub x_observed: Vec<Vec<f64>>,
    pub y_observed: Vec<f64>,
    pub best_y: f64,
    pub best_x: Vec<f64>,
    pub opts: BayesOpts,
}

Bayesian optimization over a bounded box using a GP surrogate.

The optimizer maintains a GP fitted to all observations so far, and at each step proposes the point that maximises the acquisition function.

Fields

bounds: Vec<(f64, f64)>

Search space bounds: one (lo, hi) per dimension.

gp: GaussianProcess

GP surrogate.

x_observed: Vec<Vec<f64>>

All observed inputs.

y_observed: Vec<f64>

All observed outputs.

best_y: f64

Best output value observed so far.

best_x: Vec<f64>

Input that yielded best_y.

opts: BayesOpts

Optimizer configuration.

Implementations

impl BayesianOptimizer

pub fn new( bounds: Vec<(f64, f64)>, kernel: KernelType, params: KernelParams, opts: BayesOpts, ) -> Self

Construct a new optimizer.

• bounds — axis-aligned box, one (lo, hi) per input dimension.
• kernel — kernel type for the GP surrogate.
• params — kernel hyper-parameters.
• opts — algorithm options (number of iterations, acquisition, …).

pub fn update(&mut self, x: Vec<f64>, y: f64) -> Result<(), String>

Incorporate a new observation (x, y) into the optimizer state.

The GP surrogate is re-fitted after each call.

pub fn suggest_next(&self) -> Vec<f64>

Suggest the next candidate point to evaluate.

Uses Latin-hypercube random candidates and picks the one with the highest acquisition value. Falls back to a random LHS point if the GP has not been fitted yet.

pub fn optimize<F>(&mut self, f: F) -> (Vec<f64>, f64)

where F: Fn(&[f64]) -> f64,

Run the full optimization loop, evaluating the black-box f.

First draws n_initial LHS points, then iterates for max_iter steps. Returns the best (x, y) pair found.

pub fn n_observations(&self) -> usize

Number of observations collected so far.

Trait Implementations

impl Clone for BayesianOptimizer

fn clone(&self) -> BayesianOptimizer

Returns a duplicate of the value.

fn clone_from(&mut self, source: &Self)

Performs copy-assignment from source.

impl Debug for BayesianOptimizer

fn fmt(&self, f: &mut Formatter<'_>) -> Result

Formats the value using the given formatter.