Struct ProductSearchIndex 

pub struct ProductSearchIndex { /* private fields */ }

Multi-vector product search index.

Implementations

impl ProductSearchIndex

pub fn new(config: ProductSearchConfig) -> Self

Create an empty index.

pub fn insert(&mut self, item: MultiVector)

Insert an item into the index.

pub fn search(&self, query: &MultiVector, k: usize) -> Vec<SearchCandidate>

Search for the k nearest items to query across all sub-vectors.

The combined score is the mean of per-sub-vector scores. Items are returned sorted by combined score (descending for similarity metrics).

pub fn search_sub( &self, query_sub: &[f32], sub_idx: usize, k: usize, ) -> Vec<SearchCandidate>

Search using only the sub-vector at index sub_idx.

pub fn item_count(&self) -> usize

Number of items in the index.

pub fn sub_dimension_count(&self) -> usize

The configured number of sub-dimensions.

pub fn remove(&mut self, id: usize) -> bool

Remove an item by id. Returns true if the item existed.