Struct ExperimentalDesign 

pub struct ExperimentalDesign;

Design of experiments helper

Implementations

impl ExperimentalDesign

pub fn factorial_design(factors: &[Vec<f64>]) -> Vec<Vec<f64>>

Generate a factorial design

pub fn fractional_factorial_design( factors: &[Vec<f64>], fraction: f64, seed: u64, ) -> Vec<Vec<f64>>

Generate a random fractional factorial design

pub fn central_composite_design(dimensions: usize, alpha: f64) -> Vec<Vec<f64>>

Generate a central composite design

pub fn latin_hypercube_design( factors: &[Vec<f64>], n_points: Option<usize>, seed: u64, ) -> Vec<Vec<f64>>

Generate a Latin Hypercube Sampling (LHS) design.

LHS is a stratified sampling method that produces samples that are well-distributed across each dimension’s range. For each of the D dimensions, the unit interval [0, 1) is split into N equal strata and exactly one sample is drawn from each stratum; the per-dimension samples are then independently shuffled to randomize cross-dimensional pairings. Each [0, 1) sample is finally mapped to a discrete factor level by index (u * K) as usize clamped to K - 1, where K is the number of levels of that factor.

Arguments

• factors - Per-factor level vectors (each factors[d] is the set of discrete levels for dimension d).
• n_points - Number of LHS points. When None, defaults to the maximum number of levels across all factors so that every level of every factor is visited at least once.
• seed - Deterministic RNG seed.

Returns

A vector of n_points design points; each point is a Vec<f64> of length factors.len() whose entries are drawn from the respective factor’s level set. Returns a single empty point when factors is empty (matching the behavior of factorial_design), and an empty vector when the resolved n is zero.