/**
 * Dot product of two vectors: sum_i x[i] * y[i].
 * Operates up to the shorter length of the inputs.
 * @param x First vector
 * @param y Second vector
 * @returns Scalar dot product
 */
export declare function dot(x: ArrayLike<number>, y: ArrayLike<number>): number;
/**
 * Vector p-norm. Supports common p values (1, 2, Infinity) with a fast dense path.
 * When `skipna` is true the implementation ignores NaNs and returns NaN if no valid entries exist.
 * @param x Input vector
 * @param p Norm order (default 2)
 * @param skipna Whether to ignore NaNs (default true)
 * @returns Norm value or NaN
 */
export declare function norm(x: ArrayLike<number>, p?: number, skipna?: boolean): number;
/**
 * Ordinary least squares for a simple linear model y = intercept + slope * x.
 * Ignores paired NaN entries and returns NaN coefficients if no valid pairs or singular design.
 * @param x Predictor values
 * @param y Response values
 * @returns Object with `intercept` and `slope` or NaNs on failure
 */
export declare function ols(x: ArrayLike<number>, y: ArrayLike<number>): {
    intercept: number;
    slope: number;
};
/**
 * Multiple linear regression using normal equations (adds intercept column internally).
 * Returns coefficient vector or null if the normal matrix is singular or inputs are empty.
 * @param X Design matrix (rows = observations, cols = features)
 * @param y Response vector
 * @returns Coefficient array [intercept, beta1, beta2, ...] or null
 */
export declare function olsMulti(X: number[][], y: number[]): number[] | null;