Struct CmaEsSampler 

pub struct CmaEsSampler { /* private fields */ }

A Sampler driven by CMA-ES over the study’s numeric parameters.

Construct with CmaEsSampler::new (OS-seeded) or CmaEsSampler::seeded (reproducible). The population size λ defaults to the CMA-ES rule 4 + ⌊3·ln n⌋ and can be overridden with CmaEsSampler::population_size; the first generation is sampled at random as warmup (see CmaEsSampler::n_startup_trials).

Implementations

impl CmaEsSampler

pub fn new() -> CmaEsSampler

A CMA-ES sampler seeded from OS entropy (non-deterministic across runs).

pub fn seeded(seed: u64) -> CmaEsSampler

A CMA-ES sampler with a fixed seed — reproducible for tests/benchmarks.

pub fn population_size(self, lambda: usize) -> CmaEsSampler

Override the population size λ (default 4 + ⌊3·ln n⌋).

pub fn n_startup_trials(self, n: usize) -> CmaEsSampler

Number of initial trials drawn at random before CMA-ES takes over (default: one population, λ). A full generation of spread-out points makes the first covariance estimate meaningful.

pub fn sigma0(self, sigma0: f64) -> CmaEsSampler

Initial step size σ₀, as a fraction of the normalized [0, 1] range (default 0.2).

pub fn bound_handling(self, handling: BoundHandling) -> CmaEsSampler

How to repair out-of-box draws (default BoundHandling::Reflect).

Trait Implementations

impl Default for CmaEsSampler

fn default() -> CmaEsSampler

Returns the “default value” for a type.

impl Sampler for CmaEsSampler

fn suggest( &mut self, study_state: &StudyState, trial: &Trial, param_name: &str, distribution: &Distribution, ) -> Value