Struct ScalarChebyshevPlan 

pub struct ScalarChebyshevPlan { /* private fields */ }

Chebyshev-in-density interpolant of T_i(n) for scalar (k = 1) forward models. For each row i, pre-samples T_i(n_j) at M Chebyshev-of-the-first-kind nodes in [0, n_max], then stores the Chebyshev coefficients. Online evaluation is Clenshaw recurrence with M multiply-adds per row.

Unlike the Gauss quadrature, the Chebyshev representation is a scalar interpolant in density space — one pass evaluates the interpolant at n, and the derivative needs a separate derivative-coefficient series.

Implementations

impl ScalarChebyshevPlan

pub fn build( source_resolution_plan: Arc<ResolutionPlan>, sigma: &[f64], n_max: f64, m: usize, ) -> Result<Self, ScalarSurrogateBuildError>

Build an M-node Chebyshev-in-density plan from a shared crate::resolution::ResolutionPlan + scalar σ + density box [0, n_max].

The source_resolution_plan Arc is stored on the plan so the dispatch-time eligibility check can use Arc::ptr_eq to refuse stale plans on the same grid (independent review on PR #475). A matching σ fingerprint is also computed and stored for the same reason: same-grid σ-mismatch would otherwise trigger silently-wrong transmissions.

Internally calls source_resolution_plan.compile_to_matrix() once, then crate::resolution::apply_r M times (one per Chebyshev node) to get exact row evaluations, then runs a per-row discrete cosine transform to extract Chebyshev coefficients.

Cost: one matrix compile + M × N × avg_nnz_per_row FMAs for the exact sampling pass, plus M^2 per row for the DCT.

pub fn len(&self) -> usize

pub fn is_empty(&self) -> bool

pub fn target_energies(&self) -> &[f64]

pub fn n_max(&self) -> f64

pub fn m(&self) -> usize

pub fn density_box(&self) -> Option<f64>

pub fn source_resolution_plan(&self) -> &Arc<ResolutionPlan>

Accessor for the shared crate::resolution::ResolutionPlan the plan was built from. Dispatch uses Arc::ptr_eq between this and the model’s currently attached resolution_plan as the O(1) identity check that refuses stale plans on the same grid.

pub fn sigma_fingerprint(&self) -> u64

FNV-1a-64 fingerprint of the σ slice (by to_bits()) the plan was built from. Dispatch recomputes from the model’s current σ and compares to catch same-grid σ-mismatch.

pub fn forward_scalar(&self, n: f64) -> Vec<f64>

Evaluate the Chebyshev interpolant at density n. Density outside [0, n_max] extrapolates (caller responsibility — dispatch should reject via the density-box check).

pub fn forward_and_derivative_scalar(&self, n: f64) -> (Vec<f64>, Vec<f64>)

Evaluate forward + derivative in one pass. The derivative of a Chebyshev series can be evaluated via a modified Clenshaw recurrence that internally tracks the derivative coefficients — or we use the standard identity T_k'(x) = k · U_{k-1}(x) (Chebyshev-of-the-second-kind recurrence). Here we run two parallel Clenshaw sweeps: one for T(x) and one for d/dx T(x), then scale by dx/dn = 2 / n_max.

Trait Implementations

impl Clone for ScalarChebyshevPlan

fn clone(&self) -> ScalarChebyshevPlan

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Debug for ScalarChebyshevPlan

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

Formats the value using the given formatter.