import type { SparseVec } from "./vec.js";
/**
 * Returns the Jaccard similarity of given multi-hot vectors, a value in the
 * `[0,1]` interval: 1.0 if `a` and `b` are equal, or 0.0 if none of the
 * components match.
 *
 * @remarks
 * The sizes of both input vectors MUST be equal. All non-zero vector component
 * values are treated equal.
 *
 * References:
 *
 * - https://en.wikipedia.org/wiki/Jaccard_index
 * - https://en.wikipedia.org/wiki/One-hot
 *
 * @param a -
 * @param b -
 */
export declare const jaccardSimilarity: ({ data: adata }: SparseVec, { data: bdata }: SparseVec) => number;
/**
 * Cosine **similarity** metric. Result always in `[-1,1]` interval (or in [0,1]
 * range if `a` and `b` are multi-hot vectors).
 *
 * @remarks
 * Similar to: `dot(a, b) / (magSq(a) * magSq(b))`. Returns zero if one of the
 * vectors is a zero-vector.
 *
 * Reference: https://en.wikipedia.org/wiki/Cosine_similarity
 *
 * @param a -
 * @param b -
 */
export declare const cosineSimilarity: (a: SparseVec, b: SparseVec) => number;
//# sourceMappingURL=similarity.d.ts.map