Struct BayesianNetworkParams 

pub struct BayesianNetworkParams {
    pub genome_dim: usize,
    pub max_parents: usize,
    pub init_prob: f32,
    pub smoothing_count: usize,
}

Per-run configuration for the BayesianNetwork model.

Held inside EdaParams::model for the lifetime of a run. Use BayesianNetworkParams::default_for for typical binary-optimisation defaults.

Fields

genome_dim: usize

Number of bits per genome; the number of nodes D in the network and the length of BayesianNetworkState::order and BayesianNetworkState::parents.

max_parents: usize

Maximum number of parents per node (κ); bounds each node’s CPT to 2^κ cells and caps the greedy edge-addition search.

init_prob: f32

Prior marginal probability of a 1 gene, used to seed every edgeless CPT on the prior path (prev = None).

smoothing_count: usize

Laplace pseudo-count s added per CPT cell during estimation; s ≥ 1 keeps every probability strictly inside (0, 1). Applies only to CPT estimation for sampling, never to the BIC structure score. The value is floored to 1 inside fit, so a supplied 0 is treated as 1 to uphold the strictly-interior guarantee.

Implementations

impl BayesianNetworkParams

pub fn default_for(genome_dim: usize) -> Self

Sensible BOA defaults for a genome_dim-bit problem.

Trait Implementations

impl Clone for BayesianNetworkParams

fn clone(&self) -> BayesianNetworkParams

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Debug for BayesianNetworkParams

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

Formats the value using the given formatter.