Struct EnergyScaleTransmissionModel 

pub struct EnergyScaleTransmissionModel { /* private fields */ }

Transmission model with energy-scale calibration parameters (t₀, L_scale).

Carries per-isotope resonance data (NOT a precomputed σ grid) and rebuilds the TRUE cross-section at the corrected energies on each evaluation (issue #608), matching forward_model:

1. Convert nominal energy → TOF: t = TOF_FACTOR * L / √E_nom
2. Apply calibration: t_corr = t - t₀, E_corr = (TOF_FACTOR * L * L_scale / t_corr)²
3. Evaluate σ(E_corr) via reich_moore + Doppler on the working grid built from E_corr.
4. Beer-Lambert + resolution on the working grid, then extract the data points last.

This is equivalent to SAMMY’s TZERO parameters.

The Jacobian for t₀ and L_scale defaults to partial-GAL since issue #489: central FD on t0 only (2 evals) plus an inline rank-1 derivation of the L_scale column. The previous central-FD-on-both (4-eval) behaviour is reachable via with_jacobian_method, NEREIDS_TZERO_JACOBIAN=fd2, or tzero_jacobian="fd2" Python kwarg. See EnergyScaleJacobianMethod for full method documentation.

Implementations

impl EnergyScaleTransmissionModel

pub fn new( resonance_data: Arc<Vec<ResonanceData>>, density_indices: Arc<Vec<usize>>, density_ratios: Arc<Vec<f64>>, temperature_k: f64, nominal_energies: Vec<f64>, flight_path_m: f64, t0_index: usize, l_scale_index: usize, instrument: Option<Arc<InstrumentParams>>, ) -> Self

Create a new energy-scale transmission model.

Arguments

• resonance_data — Resonance parameters per isotope; σ is evaluated at the corrected energies via reich_moore + Doppler (issue #608).
• density_indices — Maps isotope index → density parameter index.
• density_ratios — Fractional ratio per isotope (1.0 when ungrouped).
• temperature_k — Sample temperature (K) for Doppler broadening.
• nominal_energies — Energy grid in eV (ascending).
• flight_path_m — Nominal flight path in meters.
• t0_index — Index of t₀ parameter.
• l_scale_index — Index of L_scale parameter.
• instrument — Optional resolution function.

pub fn with_jacobian_method(self, method: EnergyScaleJacobianMethod) -> Self

Override the t0 / L_scale Jacobian method for this model instance. Bypasses the NEREIDS_TZERO_JACOBIAN env var.

pub fn with_temperature_index( self, temperature_index: Option<usize>, ) -> Result<Self, FittingError>

Fit the sample temperature jointly with the energy scale (issue #634). Some(idx) makes params[idx] the free temperature (K); None keeps temperature fixed at the constructor’s temperature_k.

Errors

FittingError::InvalidConfig if Some(idx) collides with t0_index, l_scale_index, or any density index — a mis-wired index would otherwise Doppler-broaden at a nonsense “temperature” (e.g. the t0 value) with no error. Mirrors TransmissionFitModel::new’s density-overlap rejection (issue #634 review, sibling-parity class).

pub fn effective_jacobian_method(&self) -> EnergyScaleJacobianMethod

The Jacobian method that will actually be used for the t0 and L_scale columns.

The rank-1 identity J[:, L_scale] = ((tof - t0)/L_scale)·J[:, t0] holds only when L_scale reaches the prediction through the corrected energy grid and nowhere else. With a resolution kernel it reaches it twice: the kernel is read against L·L_scale, so its width moves with the same parameter. That second route is absent from the identity, so deriving the column understates it.

The distinction is physical rather than a special case — it is whether an instrument response exists at all. Without one the identity is exact to f64 roundoff (partial_gal_no_resolution_matches_fd2).

Trait Implementations

impl FitModel for EnergyScaleTransmissionModel

fn analytical_jacobian( &self, params: &[f64], free_param_indices: &[usize], _y_current: &[f64], ) -> Option<FlatMatrix>

Jacobian: analytical for density parameters, finite-difference for t₀ and L_scale.

fn evaluate(&self, params: &[f64]) -> Result<Vec<f64>, FittingError>

Evaluate the model for the given parameters.

impl ForwardModel for EnergyScaleTransmissionModel

fn predict(&self, params: &[f64]) -> Result<Vec<f64>, FittingError>

Predict model output for the given parameter vector.

fn jacobian( &self, params: &[f64], free_param_indices: &[usize], y_current: &[f64], ) -> Option<Vec<Vec<f64>>>

Analytical Jacobian (column-major layout).

fn n_data(&self) -> usize

Number of data points in the model output.

fn n_params(&self) -> usize

Number of parameters (total, including fixed).