Struct EmbeddingSimilarity 

pub struct EmbeddingSimilarity;

Utility functions for vector embedding similarity.

Implementations

impl EmbeddingSimilarity

pub fn cosine(a: &[f64], b: &[f64]) -> f64

Cosine similarity between two vectors.

Returns 0.0 if either vector is the zero vector.

pub fn dot_product(a: &[f64], b: &[f64]) -> f64

Raw dot product of two vectors.

pub fn euclidean(a: &[f64], b: &[f64]) -> f64

Euclidean distance converted to a similarity score 1 / (1 + distance).

pub fn manhattan(a: &[f64], b: &[f64]) -> f64

Manhattan (L1) distance converted to a similarity score 1 / (1 + distance).

pub fn chebyshev(a: &[f64], b: &[f64]) -> f64

Chebyshev (L∞) distance converted to a similarity score 1 / (1 + distance).

pub fn compute(a: &[f64], b: &[f64], metric: SimilarityMetric) -> f64

Compute the similarity between a and b using the given metric.

pub fn top_k( query: &[f64], corpus: &[Vec<f64>], k: usize, metric: SimilarityMetric, ) -> Vec<SimilarityResult>

Return the top-k most similar vectors from corpus relative to query.

Results are sorted descending by score (most similar first). If k exceeds corpus.len(), all entries are returned.

pub fn normalize(v: &[f64]) -> Vec<f64>

L2-normalise a vector (return a unit vector).

If the input is the zero vector, returns a zero vector.

pub fn pairwise(corpus: &[Vec<f64>], metric: SimilarityMetric) -> Vec<Vec<f64>>

Compute the N×N pairwise similarity matrix for corpus.

result[i][j] is the similarity between corpus[i] and corpus[j].