Struct PairSurfaceFit 

pub struct PairSurfaceFit {
    pub phi_a: Array2<f64>,
    pub phi_b: Array2<f64>,
    pub surface: TensorSurfaceFit,
    pub backend: PairSurfaceBackend,
    pub lower_corner: [f64; 2],
    pub cell_widths: [f64; 2],
}

A pair-component fit from RAW coordinates: the factor bases it was fit on (the grid engine’s per-axis uniform cubic B-splines, evaluated on the sample — exactly what CarveInput consumes, one measure end to end) plus the TensorSurfaceFit carve product and which backend produced it.

Fields

phi_a: Array2<f64>

Axis-1 basis on the sample (n × (K+3), partition of unity).

phi_b: Array2<f64>

Axis-2 basis on the sample (n × (K+3), partition of unity).

surface: TensorSurfaceFit

The carve product: coefficients, covariances, λ, EDF.

backend: PairSurfaceBackend

lower_corner: [f64; 2]

Lower corner of the per-axis uniform knot range (the data’s bounding box) — with Self::cell_widths, everything needed to rebuild a basis row at an arbitrary point.

cell_widths: [f64; 2]

Knot-cell width per axis.

Implementations

impl PairSurfaceFit

pub fn predict( &self, dim: usize, x1: f64, x2: f64, ) -> Result<(f64, f64), String>

Posterior (mean, variance) of response dimension dim at an arbitrary point, through the carve-facing posterior objects — valid for BOTH backends, since both populate the same surface contract: mean = b₁ᵀ C_d b₂ and variance = σ̂²_d · xᵀUx with U the shared scale-free coefficient covariance, σ̂²_d the residual variance at n − edf, and x the 16-entry tensor basis row. Outside the data bounding box the boundary cell’s cubic polynomial extends (the grid engine’s convention).

Trait Implementations

impl Clone for PairSurfaceFit

fn clone(&self) -> PairSurfaceFit

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Debug for PairSurfaceFit

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

Formats the value using the given formatter.