Struct StageResult 

pub struct StageResult {
    pub label: &'static str,
    pub n_obs: u32,
    pub n_calm_obs: u32,
    pub n_false_alarms: u32,
    pub n_detections: u32,
    pub first_detection_k: Option<u32>,
    pub lambda_at_detection: Option<f32>,
    pub lambda_event_peak: f32,
    pub ground_truth_onset_k: u32,
}

Detection statistics for one stage of the Continuous Rigor pipeline.

Populated by the per-stage loops in each example.

Fields

label: &'static str

Human-readable stage label (e.g. "Stage I: Physics Baseline").

n_obs: u32

Total observations processed in this stage (including calibration).

n_calm_obs: u32

Observations in the “calm” pre-event segment (for FA counting).

n_false_alarms: u32

Policy::Review or Policy::Escalate events in the calm segment (false alarms).

n_detections: u32

Policy::Review or Policy::Escalate events in the event / post-onset segment (detections).

first_detection_k: Option<u32>

Sample index k of the first detection (None if no detection).

lambda_at_detection: Option<f32>

Lyapunov exponent λ at the first detection (None if no detection).

lambda_event_peak: f32

Maximum |λ| observed in the event segment.

ground_truth_onset_k: u32

Ground-truth onset sample index (from SigMF annotation or scenario model).

Implementations

impl StageResult

pub const fn new(label: &'static str, ground_truth_onset_k: u32) -> Self

Construct a zeroed-out result with a label and ground-truth onset.

pub fn false_alarm_rate(&self) -> f32

False-alarm rate: n_false_alarms / n_calm_obs.

Returns 0.0 if no calm observations.

pub fn lead_time_samples(&self) -> Option<i32>

Lead-time advantage in samples: ground_truth − first_detection.

Positive → DSFB detected BEFORE the ground-truth reference time. Negative → DSFB detected AFTER (degraded performance or late onset). Returns None if no detection was made.

pub fn lead_time_ms(&self, sample_rate_hz: f32) -> Option<f32>

Lead-time expressed as milliseconds, given the sample rate in Hz.

pub fn meets_1e5_fa_threshold(&self) -> bool

Whether this stage meets the SBIR 10⁻⁵ false-alarm threshold.

At a calm segment of N observations, FA rate < 10⁻⁵ requires < N/10⁵ false alarms. This method uses the observed rate as a proxy. Formal FA validation requires large-N Monte Carlo beyond what a single pipeline run provides — see paper §L5 for caveat.

Trait Implementations

impl Clone for StageResult

fn clone(&self) -> StageResult

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Debug for StageResult

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

Formats the value using the given formatter.

impl Copy for StageResult