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.

Algebraic Lyapunov and Stein Stability Results for Tensors

Numerical solution to a linear equation with tensor product structure.

Algebraic Riccati Tensor Equations with Applications in Multilinear Control Systems

Numerical stability and tensor nuclear norm

Robust D-Stability Lmi Conditions of Matrix Polytopes Via Affine Parameter-Dependent Lyapunov Functions

Stability Criteria For Matrices

Modified Z-eigenvalue inclusion intervals for tensors with applications

A Kind of Nonnegative Matrices and Its Application on the Stability of Discrete Dynamical Systems

Two-DOF Lifted LMI Conditions for Robust D-stability of Polynomial Matrix Polytopes

Z -eigenvalues based structured tensors: ℳ_z -tensors and strong ℳ_z -tensors

Lyapunov Differential Equations and Inequalities for Stability and Stabilization of Linear Time-Varying Systems

On Nonnegative Solution of Multi-Linear System with Strong Mz -Tensors

On the Stability of Spatially Distributed Systems

Z-eigenvalues Based Structured Tensors: $$\mathcal {M}_z$$-Tensors and Strong $$\mathcal {M}_z$$-Tensors.

The Intrinsic Toeplitz Structure and Its Applications in Algebraic Riccati Equations

Existence and uniqueness of positive solution for multilinear systems with generalized strong M-tensor

Sensitivity Analysis of the Lyapunov Tensor Equation

On discrete-time polynomial dynamical systems on hypergraphs

Random Quadralinear Forms and Schur Product on Tensors.

A Solvable Lie Algebra Condition for Stability of Linear Multidimensional Systems

Solving Multi-linear Systems with ℳ -Tensors