jax-js
    Preparing search index...
    • Singular value decomposition of real matrices.

      This computes a thin SVD using a symmetric eigendecomposition of A.T @ A or A @ A.T. It is intended as a baseline implementation; it is less numerically stable than a dedicated SVD routine for ill-conditioned inputs.

      With computeUv: true, returns [u, s, vh] such that a ~= u @ diag(s) @ vh. Only fullMatrices: false is supported for non-square matrices.

      Parameters

      • a: ArrayLike
      • __namedParameters: { computeUv?: boolean; fullMatrices?: boolean } = {}

      Returns Array | [Array, Array, Array]