Enum BoundaryDetection 

#[non_exhaustive]
pub enum BoundaryDetection {
    LocalMaxMinima {
        min_peak_separation: f64,
    },
    Adaptive {
        sensitivity: f64,
    },
    Manual(Vec<f64>),
}

How to locate the N Fourier-domain boundaries ω_n ∈ (0, π) that partition the signal’s spectrum into adaptive bands.

Variants (Non-exhaustive)

Non-exhaustive enums could have additional variants added in future. Therefore, when matching against variants of non-exhaustive enums, an extra wildcard arm must be added to account for any future variants.

LocalMaxMinima

Gilles 2013 §III.A: find local maxima of the (preprocessed) magnitude spectrum, keep the strongest peaks (subject to min_peak_separation in normalized frequency [0, π]), and place each boundary at the argmin of the spectrum between consecutive peaks.

Fields

min_peak_separation: f64

Minimum allowed separation between retained peaks, measured in the same [0, π] units as the boundaries. Lower values admit more (and finer-grained) bands; higher values force broader, fewer bands. A reasonable default is π / (2 · max_bands).

Adaptive

Gilles & Heal 2014 scale-space approach. The (preprocessed) spectrum is Gaussian-smoothed with a width controlled by sensitivity; peaks surviving the smoothing seed the boundaries. The current implementation is a single-scale simplification of the paper’s persistence-tracking algorithm — it is enough to keep the recovered band count stable under SNR ≥ 20 dB noise (issue acceptance criterion #4) but does not yet expose per-mode lifetimes.

Fields

sensitivity: f64

0 < sensitivity ≤ 1. Smaller values smooth more aggressively (fewer, stabler peaks) — this is the inverse of the DSP convention where higher sensitivity means more responsive detection; here it means less smoothing applied before detection, so a peakier spectrum survives. Default 0.5 smooths to roughly 2.5 % of the half-spectrum width; at sensitivity = 1 σ drops to zero and the detector falls back to gaussian_smooth’s one-sample floor.

Manual(Vec<f64>)

Caller-supplied boundaries in (0, ω_max) where ω_max = 2π · (n / 2) / n is the highest representable DFT bin frequency for the signal — exactly π for even n, but π · (n − 1) / n for odd n. Boundaries must be strictly increasing and contain at most EWTConfig::max_bands entries. Boundaries above ω_max are rejected by ewt rather than silently producing an all-zero last component.

Trait Implementations

impl Clone for BoundaryDetection

fn clone(&self) -> BoundaryDetection

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Debug for BoundaryDetection

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

Formats the value using the given formatter.