Approximate matrix and tensor diagonalization by unitary transformations: convergence of Jacobi-type algorithms

DOI10.1137/19M125950XzbMATH Open1453.90168arXiv1905.12295OpenAlexW3041284015MaRDI QIDQ5131963FDOQ5131963

Authors: Konstantin Usevich, Pierre Comon, Jianze Li 

Publication date: 9 November 2020

Published in: SIAM Journal on Optimization (Search for Journal in Brave)

Abstract: We propose a gradient-based Jacobi algorithm for a class of maximization problems on the unitary group, with a focus on approximate diagonalization of complex matrices and tensors by unitary transformations. We provide weak convergence results, and prove local linear convergence of this algorithm.The convergence results also apply to the case of real-valued tensors.

Full work available at URL: https://arxiv.org/abs/1905.12295

Recommendations

• Jacobi-type algorithms for homogeneous polynomial optimization on Stiefel manifolds with applications to tensor approximations
• Convergence of a Jacobi-type method for the approximate orthogonal tensor diagonalization
• Globally convergent Jacobi-type algorithms for simultaneous orthogonal symmetric tensor diagonalization
• Block-Jacobi methods with Newton-steps and non-unitary joint matrix diagonalization
• On approximate diagonalization of third order symmetric tensors by orthogonal transformations

zbMATH Keywords

unitary grouplocal convergenceGivens rotationsoptimization on manifoldsapproximate tensor diagonalizationŁojasiewicz gradient inequality

Mathematics Subject Classification ID

Numerical optimization and variational techniques (65K10) Complexity and performance of numerical algorithms (65Y20) Multilinear algebra, tensor calculus (15A69) Nonlinear programming (90C30) Methods of local Riemannian geometry (53B21) Local Riemannian geometry (53B20)

Cites Work

• Manopt, a Matlab toolbox for optimization on manifolds
• Title not available (Why is that?)
• Title not available (Why is that?)
• The Geometry of Algorithms with Orthogonality Constraints
• Jacobi Angles for Simultaneous Diagonalization
• Title not available (Why is that?)
• Title not available (Why is that?)
• Function theory of several complex variables
• Morse Theory. (AM-51)
• Title not available (Why is that?)
• Convergence of the Iterates of Descent Methods for Analytic Cost Functions
• Convergence of descent methods for semi-algebraic and tame problems: proximal algorithms, forward-backward splitting, and regularized Gauss-Seidel methods
• Title not available (Why is that?)
• Title not available (Why is that?)
• Geodesic convexity in nonlinear optimization
• Optimization and dynamical systems
• Tensor decompositions for learning latent variable models
• Pointwise convergence of gradient‐like systems
• Nondegenerate critical manifolds
• An Extrinsic Look at the Riemannian Hessian
• A proof of the Morse-Bott lemma.
• Numerical Methods for Simultaneous Diagonalization
• Jacobi algorithm for the best low multilinear rank approximation of symmetric tensors
• A new convergence proof for the higher-order power method and generalizations
• Convergence results for projected line-search methods on varieties of low-rank matrices via Łojasiewicz inequality
• A global convergence proof for cyclic Jacobi methods with block rotations
• On the Convergence of the Jacobi Method for Arbitrary Orderings
• Title not available (Why is that?)
• Contrasts, independent component analysis, and blind deconvolution
• Complex-Valued Matrix Differentiation: Techniques and Key Results
• Characterizing real-valued multivariate complex polynomials and their symmetric tensor representations
• Lie Groups, Lie Algebras, and Representations
• Steepest Descent Algorithms for Optimization Under Unitary Matrix Constraint
• Jacobi's Algorithm on Compact Lie Algebras
• Global rates of convergence for nonconvex optimization on manifolds
• Convergence rate analysis for the higher order power method in best rank one approximations of tensors
• Convergence of the cyclic and quasi-cyclic block Jacobi methods
• Quotient geometry with simple geodesics for the manifold of fixed-rank positive-semidefinite matrices
• Globally convergent Jacobi-type algorithms for simultaneous orthogonal symmetric tensor diagonalization
• Hermitian tensor decompositions

Cited In (15)

• Convergence of a Jacobi-type method for the approximate orthogonal tensor diagonalization
• Jacobi-type algorithms for homogeneous polynomial optimization on Stiefel manifolds with applications to tensor approximations
• Globally convergent Jacobi-type algorithms for simultaneous orthogonal symmetric tensor diagonalization
• On approximate diagonalization of third order symmetric tensors by orthogonal transformations
• Convergence and worst-case complexity of adaptive Riemannian trust-region methods for optimization on manifolds
• Projectively and Weakly Simultaneously Diagonalizable Matrices and their Applications
• Convergence of Gradient-Based Block Coordinate Descent Algorithms for Nonorthogonal Joint Approximate Diagonalization of Matrices
• Title not available (Why is that?)
• Block-Jacobi methods with Newton-steps and non-unitary joint matrix diagonalization
• Sparse additive function decompositions facing basis transforms
• Conjugate gradient algorithm for optimization under unitary matrix constraint
• Trace maximization algorithm for the approximate tensor diagonalization
• Alternating Mahalanobis Distance Minimization for Accurate and Well-Conditioned CP Decomposition
• Linear convergence of an alternating polar decomposition method for low rank orthogonal tensor approximations
• A Riemannian gradient ascent algorithm with applications to orthogonal approximation problems of symmetric tensors

Uses Software

• Manopt

This page was built for publication: Approximate matrix and tensor diagonalization by unitary transformations: convergence of Jacobi-type algorithms

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q5131963)