Struct CharacteristicVector 

pub struct CharacteristicVector {
    pub counts: [u32; 23],
    pub max_depth: u32,
    pub node_count: u32,
}

Shape-tag counts plus tree size and depth: a candidate filter.

The count vector is a cheap lower-bound proxy for tree edit distance: two subtrees within edit distance d differ by at most 2 * d in L1 count distance, so a large CharacteristicVector::l1_distance rules a pair out without touching either tree. That is what CharacteristicVector::shape_divergence gates candidate pairs on, and what CharacteristicVector::cosine_similarity contributes to the structural dimension of a verdict.

Fields

counts: [u32; 23]

Node count per shape tag; index = tag, index 0 unused.

max_depth: u32

Number of nodes on the longest root-to-leaf path of the unit subtree; a lone node has depth 1.

node_count: u32

Total number of nodes in the unit subtree.

Implementations

impl CharacteristicVector

pub fn l1_distance(&self, other: &Self) -> u64

L1 distance between the two count vectors. Depth and node count do not participate.

pub fn shape_divergence(&self, other: &Self) -> f64

How far apart the two shape mixes are, on a 0.0–1.0 scale: the L1 count distance over the nodes the two units have between them. 0.0 when they hold the same shapes in the same numbers, 1.0 when they share no shape at all. 0.0 for two empty vectors, which are not divergent — they are simply nothing to tell apart.

Size is part of it, and deliberately so: the vectors sum to their unit node counts, so the distance is at least |na - nb| / (na + nb) and a pair whose sizes differ by a factor of r scores at least (r - 1) / (r + 1) before any difference in shape mix is counted. A limit of 0.5 therefore says exactly what max_length_ratio’s 3.0 says about size, and says it about the shape mix too.

pub fn cosine_similarity(&self, other: &Self) -> f64

Cosine similarity of the two count vectors, 0.0 when either vector is all-zero. Depth and node count do not participate.

Trait Implementations

impl Clone for CharacteristicVector

fn clone(&self) -> CharacteristicVector

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Debug for CharacteristicVector

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

Formats the value using the given formatter.

impl Default for CharacteristicVector

fn default() -> CharacteristicVector

Returns the “default value” for a type.

impl Eq for CharacteristicVector

impl PartialEq for CharacteristicVector

fn eq(&self, other: &CharacteristicVector) -> bool

Equality operator ==.

fn ne(&self, other: &Rhs) -> bool

Inequality operator !=.

impl StructuralPartialEq for CharacteristicVector