Struct TransmissionFitModel 

pub struct TransmissionFitModel { /* private fields */ }

Forward model for fitting isotopic areal densities from transmission data.

The model computes T(E) for a set of isotopes with variable areal densities. Each isotope’s resonance data and the energy grid are fixed; only the areal densities are adjusted during fitting.

Optionally, the sample temperature can also be fitted by setting temperature_index to the parameter slot holding the temperature value. When temperature_index is Some(idx), the Doppler broadening kernel is recomputed at params[idx] when the temperature changes (cached across calls at the same temperature), and the analytical Jacobian provides density columns directly plus a single FD column for temperature.

instrument uses Arc so that parallel pixel loops can share one copy of a potentially large tabulated resolution kernel via cheap reference-count increments instead of deep-cloning per pixel.

Implementations

impl TransmissionFitModel

pub fn new( energies: Vec<f64>, resonance_data: Vec<ResonanceData>, temperature_k: f64, instrument: Option<Arc<InstrumentParams>>, density_mapping: (Vec<usize>, Vec<f64>), temperature_index: Option<usize>, external_base_xs: Option<Arc<Vec<Vec<f64>>>>, ) -> Result<Self, FittingError>

Create a validated TransmissionFitModel.

When external_base_xs is Some, uses those precomputed unbroadened cross-sections instead of computing them (expensive Reich-Moore). spatial_map precomputes once for all pixels and passes them here.

Errors

Returns FittingError::InvalidConfig if temperature_index overlaps with density_indices, or if external_base_xs has a mismatched shape.

Trait Implementations

impl FitModel for TransmissionFitModel

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

Analytical Jacobian for the transmission model with temperature fitting.

When base_xs is available (temperature fitting path):

• Density columns: ∂T/∂nᵢ = -σᵢ(E)·T(E) using cached broadened XS from the most recent evaluate() call. Same formula as PrecomputedTransmissionModel, zero extra broadening calls.
• Temperature column: analytical chain rule via on-demand ∂σ/∂T. ∂T/∂T_temp = -T(E) · Σᵢ nᵢ·rᵢ·∂σᵢ/∂T. The derivative is computed once per temperature via broadened_cross_sections_with_analytical_derivative_from_base() and cached until temperature changes. Costs one broadening call per Jacobian (same as the old FD approach, but exact).

Returns None for the no-base_xs path (full forward model), which falls back to finite-difference in the LM solver. Analytical Jacobian for density and temperature fitting.

Without resolution: ∂T/∂N_g = -(Σ_{i∈g} rᵢ σᵢ) · T ∂T/∂Temp = -T · Σᵢ nᵢ rᵢ ∂σᵢ/∂T

With resolution (R is a linear operator): ∂T_obs/∂N_g = R[-(Σ_{i∈g} rᵢ σᵢ) · T] ∂T_obs/∂Temp = R[-T · Σᵢ nᵢ rᵢ ∂σᵢ/∂T]

Returns None only when base_xs is not available (full forward model path falls back to FD) or when the temperature cache is stale.

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

Evaluate the model for the given parameters.

impl ForwardModel for TransmissionFitModel

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).