Skip to content

svd

svd(A, options?): object

Defined in: svd.js:23

Thin SVD: A = U diag(s) V^T with U m×k, s length k, V n×k, k = min(m, n). Singular values are non-negative and descending.

Parameters

A

number[][]

Matrix (any shape)

options?

maxSweeps?

number

Maximum Jacobi sweeps

tol?

number

Column-pair orthogonality tolerance

Returns

object

s

s: number[]

U

U: number[][]

V

V: number[][]