Struct SbxCrossover 

pub struct SbxCrossover {
    pub eta: f64,
    pub crossover_probability: f64,
    pub exchange_probability: f64,
}

Simulated Binary Crossover (SBX)

SBX generates offspring from parents using a spread factor that simulates single-point crossover for binary strings.

Two distinct probabilities

SBX involves two conceptually separate probabilities that used to be conflated into a single field (audit EV-72):

• crossover_probability — the per-pair probability that the operator is applied at all. This is the quantity reported by CrossoverOperator::crossover_probability and consumed by the driving algorithm (e.g. SimpleGA, NSGA-II) to decide, once per parent pair, whether to recombine. Canonical default: 0.9.
• exchange_probability — the per-gene probability, used inside SBX, that an individual variable is recombined via the spread factor (otherwise the children inherit the parents’ values unchanged for that variable). Deb’s canonical NSGA-II code uses 0.5.

Bounds handling

When bounds are supplied (via BoundedCrossoverOperator) the operator uses Deb & Agrawal’s bounds-aware spread factor: for each variable the spread is drawn from the polynomial SBX distribution truncated to the distance to each bound, so offspring land inside [min, max] by construction with no probability mass piled onto the boundaries (audit EV-71). The unbounded path uses the classic (untruncated) spread factor.

Reference: Deb, K., & Agrawal, R. B. (1995). Simulated Binary Crossover for Continuous Search Space. Bounds-aware variant: Deb, K. (2001), NSGA-II crossover.c.

Fields

eta: f64

Distribution index (typically 2-20) Higher values = offspring closer to parents

crossover_probability: f64

Per-pair probability that the operator is applied (reported by CrossoverOperator::crossover_probability). Default: 0.9.

exchange_probability: f64

Per-gene exchange probability used inside SBX. Default: 0.5 (canonical). This is a distinct quantity from crossover_probability.

Implementations

impl SbxCrossover

pub fn new(eta: f64) -> Self

Create a new SBX crossover with the given distribution index

pub fn with_probability(self, probability: f64) -> Self

Set the per-pair crossover probability reported by CrossoverOperator::crossover_probability.

This is the probability the driving algorithm uses to decide, once per parent pair, whether to apply the operator. It is not the per-gene exchange rate — for that see with_exchange_probability.

pub fn with_exchange_probability(self, probability: f64) -> Self

Set the per-gene exchange probability used inside SBX (canonical: 0.5).

Trait Implementations

impl BoundedCrossoverOperator<RealVector> for SbxCrossover

fn crossover_bounded<R: Rng>( &self, parent1: &RealVector, parent2: &RealVector, bounds: &MultiBounds, rng: &mut R, ) -> OperatorResult<(RealVector, RealVector)>

Apply bounded crossover to two parents

impl Clone for SbxCrossover

fn clone(&self) -> SbxCrossover

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl CrossoverOperator<RealVector> for SbxCrossover

fn crossover<R: Rng>( &self, parent1: &RealVector, parent2: &RealVector, rng: &mut R, ) -> OperatorResult<(RealVector, RealVector)>

Apply crossover to two parents and produce two offspring

fn crossover_probability(&self) -> f64

Get the probability of crossover being applied

impl Debug for SbxCrossover

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

Formats the value using the given formatter.