Enum ResolutionFunction 

pub enum ResolutionFunction {
    Gaussian(ResolutionParams),
    Tabulated(Arc<TabulatedResolution>),
    IkedaCarpenter(Arc<IkedaCarpenter>),
}

Resolution function: analytical Gaussian, tabulated from Monte Carlo, or analytical Ikeda–Carpenter moderator model.

The Tabulated and IkedaCarpenter variants wrap an Arc so that cloning (e.g., per-pixel in spatial mapping) is a cheap reference-count bump rather than a deep copy.

IkedaCarpenter synthesizes a TabulatedResolution at construction and is applied through the same per-call convolution path as Tabulated (broaden / broaden_presorted / plan) — only the kernel source differs (analytic IC pulse vs Monte-Carlo file). This keeps the three-way resolution cross-validation (Gaussian | tabulated-UDR | Ikeda–Carpenter) fair on the reference broadening path. Note: IkedaCarpenter does not opt into the spatial-map surrogate fast-paths (the scalar/cubature plans gate on Tabulated); it falls back to the general path, which is correct but unoptimized — see the resolution-calibration notes for the W6 follow-up.

Variants

Gaussian(ResolutionParams)

Analytical Gaussian resolution from instrument parameters.

Tabulated(Arc<TabulatedResolution>)

Tabulated resolution from Monte Carlo instrument simulation.

IkedaCarpenter(Arc<IkedaCarpenter>)

Analytical Ikeda–Carpenter moderator resolution model.

Implementations

impl ResolutionFunction

pub fn grid_bounds_ev(&self, energies: &[f64]) -> (f64, f64)

The energies a working grid for the data window energies has to span, as (low, high) in eV with low ≤ e_min and high ≥ e_max: SAMMY’s Wdsint limits at the two ends for a Gaussian, the extremes of TabulatedResolution::gather_bounds_ev over the grid for a sampled kernel. Returns (0.0, 0.0) for an empty grid.

pub fn flight_path_m(&self) -> f64

Flight path used to map true neutron energy to detector time.

pub fn with_flight_path( &self, flight_path_m: f64, ) -> Result<Self, ResolutionParseError>

The same resolution function read against a different flight path.

A fit that frees L_scale evaluates the theory on an energy grid built with L·L_scale. The kernel has to be read against the same flight path or its width is wrong by that factor — on every family, since all three convert between energy and detector time through L. No variant resynthesizes: the flight path is not part of what the kernel IS, only of the map it is applied through.

Errors

Returns ResolutionParseError::InvalidFormat if flight_path_m is not positive and finite.

pub fn detector_bin_probabilities( &self, true_energy_ev: f64, detector_time_edges_us: &[f64], timing_offset_us: f64, ) -> Result<Vec<f64>, ResolutionParseError>

Probability that one neutron of known true energy is recorded in each supplied detector-time bin.

The tabulated and Ikeda–Carpenter variants are evaluated directly in detector time. In particular, the analytical IC variant does not pass through its legacy synthesized TabulatedResolution broadening table. The older Gaussian energy-broadening model has no physical detector-time probability law and is therefore rejected rather than silently treated as one.

timing_offset_us is convention-dependent and NOT transferable between variants: a mode-centred tabulated (UDR) kernel places its pulse mode at the nominal arrival, so the offset must absorb the calibrated moderator mean delay, while the causal Ikeda–Carpenter pulse rises from the nominal arrival onward and its offset is a pure clock/detector shift. Swapping response models under one calibrated offset shifts every bin systematically.

Trait Implementations

impl Clone for ResolutionFunction

fn clone(&self) -> ResolutionFunction

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Debug for ResolutionFunction

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

Formats the value using the given formatter.