Member Junction
    Preparing search index...

    A nearest-neighbour search over a store snapshot.

    interface VectorSearchJob {
        Candidates: Int32Array<ArrayBufferLike>;
        Metric: DistanceMetric;
        Query: Float64Array<ArrayBufferLike>;
        QueryNormSq: number;
        RowEnd?: number;
        RowStart?: number;
        Snapshot: VectorStoreSnapshot;
        Source?: VectorStoreChangeSource;
        Threshold: number;
        TopK: number;
    }

    Hierarchy (View Summary)

    Index

    Properties

    Candidates: Int32Array<ArrayBufferLike>

    Ascending row indexes to consider, or null for every live row in [RowStart, RowEnd)

    Query: Float64Array<ArrayBufferLike>

    The query, with values already rounded to the store's precision

    QueryNormSq: number

    Sum of squares of the query

    RowEnd?: number

    Row after the last one scanned when Candidates is null (default RowCount)

    RowStart?: number

    First row scanned when Candidates is null (default 0)

    The live store, for accelerators that maintain derived indexes

    Threshold: number

    Minimum score (inclusive), or null for no threshold

    TopK: number

    Maximum rows to return, or null to return every row that passes the threshold