Struct NeighborhoodSearch 

pub struct NeighborhoodSearch { /* private fields */ }

HNSW neighborhood search algorithms

Provides high-performance k-nearest neighbor search using the HNSW algorithm with dynamic candidate management and layer-based navigation.

Implementations

impl NeighborhoodSearch

pub fn new(distance_metric: DistanceMetric) -> Self

Create a new neighborhood search instance

Arguments

• distance_metric - Distance metric to use for similarity computation

Returns

New NeighborhoodSearch instance

Examples

use sqlitegraph::hnsw::{NeighborhoodSearch, DistanceMetric};

let search = NeighborhoodSearch::new(DistanceMetric::Cosine);
// Ready for k-NN search operations

pub fn search_layer( &self, layer: &HnswLayer, query_vector: &[f32], vectors: &[Vec<f32>], entry_points: &[u64], k: usize, ) -> Result<SearchResult, HnswError>

Find k-nearest neighbors in a single layer

Performs greedy search within a specific layer to find the nearest neighbors to the query vector.

Arguments

• layer - Target layer for search
• query_vector - Query vector for similarity comparison
• vectors - Vector storage (node_id -> vector mapping)
• entry_points - Initial entry points for search
• k - Number of neighbors to find

Returns

SearchResult with k nearest neighbors or error

Examples

use sqlitegraph::hnsw::{NeighborhoodSearch, DistanceMetric};

let search = NeighborhoodSearch::new(DistanceMetric::Cosine);
let vectors: Vec<Vec<f32>> = create_test_vectors();
let layer = create_test_layer();

let result = search.search_layer(
    &layer,
    &query_vector,
    &vectors,
    &[0, 1],  // entry points
    5         // k = 5
)?;

Trait Implementations

impl Default for NeighborhoodSearch

fn default() -> Self

Returns the “default value” for a type.