Struct LayerVerdict 

pub struct LayerVerdict {
    pub layer: Layer,
    pub confidence: f64,
    pub margin: f64,
    pub uncertain: bool,
    pub candidates: Vec<(Layer, f64)>,
    pub basis: String,
    pub support_fns: usize,
}

Fields

layer: Layer

confidence: f64

Absolute fit: how close layer is to the observed histogram, normalised against the max possible Euclidean distance in the 5-bucket simplex. Penalised on small modules (<5 fns capped at 0.2, <10 fns softened by 0.7×).

margin: f64

Distance gap between the chosen layer and the runner-up. High margin = clear winner. Low margin = classifier is hesitating between multiple fingerprints — read with care.

uncertain: bool

True when the verdict shouldn’t be read as a hard label — either confidence is low OR margin to the runner-up is small. Renderers should surface this with explicit “uncertain” wording.

candidates: Vec<(Layer, f64)>

Top three nearest layers with their raw distances. Lets the caller show “next: Domain Δ4.1, RenderUi Δ8.3” so the verdict is auditable without pretending the second-best didn’t exist.

basis: String

support_fns: usize

Trait Implementations

impl Clone for LayerVerdict

fn clone(&self) -> LayerVerdict

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Debug for LayerVerdict

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

Formats the value using the given formatter.