Struct SpatialQuality 

pub struct SpatialQuality {
    pub evr: f64,
    pub cap_areas: Vec<f64>,
    pub exclusivities: Vec<f64>,
    pub voronoi_cells: Vec<VoronoiCell>,
    pub pairwise_intersections: HashMap<(usize, usize), f64>,
    pub coverage: CoverageReport,
    pub bridge_threshold: f64,
    pub bridge_quality_matrix: Vec<Vec<f64>>,
}

Pre-computed spatial properties of the category layout on S².

Every field here is derived from the category centroids and angular spreads — no embedding-space information, pure sphere geometry. This struct is computed once during CategoryLayer::build() and informs bridge detection, edge weights, and confidence scoring.

Fields

evr: f64

Global explained variance ratio of the projection.

cap_areas: Vec<f64>

Solid angle of each category’s cap (2π(1 − cos α)).

exclusivities: Vec<f64>

Per-category exclusivity: fraction of cap not overlapped by any other. 1.0 = isolated, 0.0 = completely overlapped.

voronoi_cells: Vec<VoronoiCell>

Voronoi cell for each category (area + neighbor indices).

pairwise_intersections: HashMap<(usize, usize), f64>

Pairwise cap intersection areas in steradians, keyed by (min(i, j), max(i, j)). Only pairs with measurable overlap (> 1e-15 sr) are stored, keeping the map sparse.

coverage: CoverageReport

Coverage report: what fraction of S² is claimed by any category.

bridge_threshold: f64

EVR-adaptive bridge threshold. Higher EVR → looser threshold. Formula: 0.5 + (1 − EVR)² × 0.4

bridge_quality_matrix: Vec<Vec<f64>>

C×C matrix of spatially-adjusted bridge quality between category pairs. matrix[i][j] = max_bridge_strength(i,j) × territorial_factor(i,j). Empty until Self::set_bridge_quality_matrix is called with a built CategoryGraph (done during CategoryLayer::build).

Implementations

impl SpatialQuality

pub fn compute( centroids: &[SphericalPoint], half_angles: &[f64], evr: f64, ) -> Self

👎Deprecated:

use compute_with_config; this uses PipelineConfig::default() sample counts

Compute spatial quality from category centroids and angular spreads, using the legacy default Monte Carlo sample counts.

pub fn compute_with_config( centroids: &[SphericalPoint], half_angles: &[f64], evr: f64, config: &PipelineConfig, ) -> Self

Compute spatial quality using configurable sample counts and bridge threshold parameters.

Cost at default sample counts: ~100-200ms for 31 categories. This is a one-time build cost, not per-query.

pub fn set_bridge_quality_matrix(&mut self, graph: &CategoryGraph)

Populate the C×C bridge_quality_matrix from a freshly built graph.

Each cell is edge.max_bridge_strength × territorial_factor(i, j), left at 0.0 where no edge exists (including the diagonal).

pub fn territorial_factor(&self, cat_a: usize, cat_b: usize) -> f64

Exclusivity-based territorial factor for a category pair.

Bridges between categories that heavily overlap (low exclusivity) are discounted — they’re shared territory, not genuine connectors. Returns a value in (0, 1].

pub fn are_voronoi_neighbors(&self, cat_a: usize, cat_b: usize) -> bool

Whether two categories are Voronoi neighbors (geometrically adjacent on S²).

pub fn voronoi_area(&self, cat: usize) -> f64

Voronoi cell area for a category.

pub fn territorial_efficiency(&self, cat: usize, item_count: usize) -> f64

Territorial efficiency: items per steradian of Voronoi cell.

pub fn intersection_area(&self, cat_a: usize, cat_b: usize) -> f64

Cap intersection area between two categories.

Trait Implementations

impl Clone for SpatialQuality

fn clone(&self) -> SpatialQuality

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Debug for SpatialQuality

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

Formats the value using the given formatter.