Struct BayesianConfig 

pub struct BayesianConfig {
    pub ncomp: usize,
    pub tau2: f64,
    pub ig_a0: f64,
    pub ig_b0: f64,
    pub n_iter: usize,
    pub burn_in: usize,
    pub thin: usize,
    pub seed: u64,
}

Configuration for the Bayesian FOSR Gibbs sampler (REG-06-04).

Uses conjugate Normal-Inverse-Gamma priors. Defaults match the weakly-informative settings recommended by Jiang et al. (2025): τ² = 100, IG(0.001, 0.001).

Divergence from refund: refund’s Bayesian FOSR uses spline basis priors; this implementation uses FPCA score compression via fdata_to_pc_1d for simplicity and zero new dependencies.

Fields

ncomp: usize

Number of FPC components for score compression (must be ≥ 1).

tau2: f64

Prior variance τ² on FPC-space coefficients (default: 100.0).

Large τ² gives a weakly informative prior; very small values are dogmatic.

ig_a0: f64

Inverse-Gamma prior shape a₀ (default: 0.001 — weakly informative).

ig_b0: f64

Inverse-Gamma prior rate b₀ (default: 0.001 — weakly informative).

n_iter: usize

Number of Gibbs iterations retained after burn-in (must be ≥ 1).

burn_in: usize

Burn-in iterations discarded (must be < n_iter iterations will run).

thin: usize

Thinning interval — keep every thin-th draw (must be ≥ 1).

seed: u64

RNG seed — chain is fully deterministic: StdRng::seed_from_u64(seed).

Trait Implementations

impl Clone for BayesianConfig

fn clone(&self) -> BayesianConfig

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Debug for BayesianConfig

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

Formats the value using the given formatter.

impl PartialEq for BayesianConfig

fn eq(&self, other: &BayesianConfig) -> bool

Equality operator ==.

fn ne(&self, other: &Rhs) -> bool

Inequality operator !=.

impl StructuralPartialEq for BayesianConfig