Abstract:We propose new algorithms for singular value decomposition (SVD) of very large-scale matrices based on a low-rank tensor approximation technique called the tensor train (TT) format. The proposed algorithms can compute several dominant singular values and corresponding singular vectors for large-scale structured matrices given in a TT format. The computational complexity of the proposed methods scales logarithmically with the matrix size under the assumption that both the matrix and the singular vectors admit low-rank TT decompositions. The proposed methods, which are called the alternating least squares for SVD (ALS-SVD) and modified alternating least squares for SVD (MALS-SVD), compute the left and right singular vectors approximately through block TT decompositions. The very large-scale optimization problem is reduced to sequential small-scale optimization problems, and each core tensor of the block TT decompositions can be updated by applying any standard optimization methods. The optimal ranks of the block TT decompositions are determined adaptively during iteration process, so that we can achieve high approximation accuracy. Extensive numerical simulations are conducted for several types of TT-structured matrices such as Hilbert matrix, Toeplitz matrix, random matrix with prescribed singular values, and tridiagonal matrix. The simulation results demonstrate the effectiveness of the proposed methods compared with standard SVD algorithms and TT-based algorithms developed for symmetric eigenvalue decomposition.

Symmetric tensor decomposition by an iterative eigendecomposition algorithm

Iterative Methods for Symmetric Outer Product Tensor Decompositions

Partial orthogonal rank-one decomposition of complex symmetric tensors based on the Takagi factorization

Decomposition of linear tensor transformations

Decomposing tensors via rank-one approximations

Subspace power method for symmetric tensor decomposition

Locally Optimal Eigenpairs of Orthogonally Decomposable Tensors: A Generalized Proof

A parallel iterative method for solving symmetric tridiagonal extreme eigenpair problems

An Iterative Reweighted Method for Tucker Decomposition of Incomplete Multiway Tensors

A Parallel Iterative Method For Solving Symmetrical Tridiagonal Extreme Eigenpair Problems

On The Tensor Svd And The Optimal Low Rank Orthogonal Approximation Of Tensors

Rank Properties and Computational Methods for Orthogonal Tensor Decompositions

A chiseling algorithm for low-rank Grassmann decomposition of skew-symmetric tensors

Empirical Evaluation of Four Tensor Decomposition Algorithms

A Backward Stable Algorithm for Computing the CS Decomposition via the Polar Decomposition

Symmetry Breaking in Symmetric Tensor Decomposition

The eigenvalue decomposition of normal matrices by the decomposition of the skew-symmetric part with applications to orthogonal matrices

Very Large-Scale Singular Value Decomposition Using Tensor Train Networks

An Inverse Extreme Eigenpair Problem and Its Parallel Iterative Solution for Symmetric Tridiagonal Matrices

VeST: Very Sparse Tucker Factorization of Large-Scale Tensors

Robust Low-rank Tensor Decomposition with the L2 Criterion