Struct BayesianOptimizer 

pub struct BayesianOptimizer<F>
where
    F: Fn(&[f64]) -> f64,
{ /* private fields */ }

Sequential Bayesian optimization over a box.

1. Sample n_initial random points, evaluate, take the best.
2. Fit a GP to all evaluations.
3. For each iteration: score n_candidates random points by Expected Improvement, evaluate the argmax, refit, repeat.

Implementations

impl<F> BayesianOptimizer<F>

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

pub const fn new(fitness: F, seed: u64) -> Self

Create an optimizer over the fitness function with a PRNG seed.

pub fn optimize( &mut self, bounds: &[(f64, f64)], n_initial: usize, n_iterations: usize, n_candidates: usize, exploration: f64, ) -> Result<OptimizeResult, String>

Run the optimization.

bounds: [(min, max), ...] per dimension (all required). Returns the full trace and the best (params, fitness).