Abstract:We present a notion of bilinear stability, which is to numerical stability what bilinear complexity is to time complexity. In bilinear complexity, an algorithm for evaluating a bilinear operator $\beta : \mathbb{U} \times \mathbb{V} \to \mathbb{W}$ is a decomposition $\beta = \varphi_1 \otimes \psi_1 \otimes w_1 + \dots + \varphi_r \otimes \psi_r \otimes w_r $; the number of terms $r$ captures the speed of the algorithm; and its smallest possible value, i.e., the tensor rank of $\beta$, quantifies the speed of a fastest algorithm. Bilinear stability introduces norms to the mix: The growth factor of the algorithm $\lVert \varphi_1 \rVert_* \lVert \psi_1 \rVert_* \lVert w_1 \rVert + \dots + \lVert \varphi_r \rVert_* \lVert \psi_r \rVert_* \lVert w_r \rVert$ captures the accuracy of the algorithm; and its smallest possible value, i.e., the tensor nuclear norm of $\beta$, quantifies the accuracy of a stablest algorithm. To substantiate this notion, we establish a bound for the forward error in terms of the growth factor and present numerical evidence comparing various fast algorithms for matrix and complex multiplications, showing that larger growth factors correlate with less accurate results. Compared to similar studies of numerical stability, bilinear stability is more general, applying to any bilinear operators and not just matrix or complex multiplications; is more simplistic, bounding forward error in terms of a single (growth) factor; and is truly tensorial like bilinear complexity, invariant under any orthogonal change of coordinates. As an aside, we study a new algorithm for computing complex multiplication in terms of real, much like Gauss's, but is optimally fast and stable in that it attains both tensor rank and nuclear norm.

Tensor logarithmic norm and its applications.

On the tensor spectral -norm and its higher order power method

Numerical stability and tensor nuclear norm

The Stochastic Logarithmic Norm for Stability Analysis of Stochastic Differential Equations

On the applicability of logarithmic tensor category theory

Tensor network renormalization: application to dynamic correlation functions and non-hermitian systems

Spectral Radius and Infinity Norm of Matrices

Sum of squared logarithms - an inequality relating positive definite matrices and their matrix logarithm

ℋ-tensors and Nonsingular ℋ-tensors

Applications of Induced Tensor Norms to Guidance Navigation and Control

Modified Z-eigenvalue inclusion intervals for tensors with applications

Sharp Bounds on Toeplitz Determinants for Starlike and Convex Functions Associated with Bilinear Transformations

Tensor Recovery Based on a Novel Non-Convex Function Minimax Logarithmic Concave Penalty Function

Log-majorization and matrix norm inequalities with application to quantum information

Tensor norms on ordered normed spaces, polarization constants, and exchangeable distributions

Logarithmic Norm Regularized Low-Rank Factorization for Matrix and Tensor Completion

On Stability of Tensor Networks and Canonical Forms

Generalized Eigenvalue for Even Order Tensors Via Einstein Product and Its Applications in Multilinear Control Systems.

Derivatives of Multilinear Functions of Matrices

On some tensor inequalities based on the t-product

Linear to multi-linear algebra and systems using tensors