Numerical CP decomposition of some difficult tensors

DOI10.1016/J.CAM.2016.12.007zbMATH Open1357.65050arXiv1603.01372OpenAlexW2963931908MaRDI QIDQ508007FDOQ508007

Authors: Petr Tichavský, Anh-Huy Phan, Andrzej Cichocki 

Publication date: 9 February 2017

Published in: Journal of Computational and Applied Mathematics (Search for Journal in Brave)

Abstract: In this paper, a numerical method is proposed for canonical polyadic (CP) decomposition of small size tensors. The focus is primarily on decomposition of tensors that correspond to small matrix multiplications. Here, rank of the tensors is equal to the smallest number of scalar multiplications that are necessary to accomplish the matrix multiplication. The proposed method is based on a constrained Levenberg-Marquardt optimization. Numerical results indicate the rank and border ranks of tensors that correspond to multiplication of matrices of the size 2x3 and 3x2, 3x3 and 3x2, 3x3 and 3x3, and 3x4 and 4x3. The ranks are 11, 15, 23 and 29, respectively. In particular, a novel algorithm for multiplying the matrices of the sizes 3x3 and 3x2 with 15 multiplications is presented.

Full work available at URL: https://arxiv.org/abs/1603.01372

Recommendations

• A Practical Randomized CP Tensor Decomposition
• Numerical optimization for symmetric tensor decomposition
• Tensor decompositions for the bubbles and cube numerical framework
• Numerical tensor techniques for multidimensional convolution products
• scientific article; zbMATH DE number 1225903
• Decomposition Algorithms for Tensors and Polynomials
• Computing tensor product decompositions
• Tensor Decompositions and Applications
• Tensor numerical methods in scientific computing
• Triangular decomposition of CP factors of a third-order tensor with application to solving nonlinear systems of equations

zbMATH Keywords

algorithmnumerical resultLevenberg-Marquardt methodcanonical polyadic tensor decompositionsmall matrix multiplication

Mathematics Subject Classification ID

Multilinear algebra, tensor calculus (15A69)

Cites Work

• Convergence Properties of the Nelder--Mead Simplex Method in Low Dimensions
• Tensor Decompositions and Applications
• Gaussian elimination is not optimal
• Multiplying matrices faster than coppersmith-winograd
• On multiplication of 2 \(\times\) 2 matrices
• Title not available (Why is that?)
• A noncommutative algorithm for multiplying 3×3 matrices using 23 multiplications
• Duality Applied to the Complexity of Matrix Multiplication and Other Bilinear Forms
• Title not available (Why is that?)
• Low complexity damped Gauss-Newton algorithms for CANDECOMP/PARAFAC
• Enhanced Line Search: A Novel Method to Accelerate PARAFAC
• Optimization techniques for small matrix multiplication

Cited In (5)

• Equivalent polyadic decompositions of matrix multiplication tensors
• Comparison of Accuracy and Scalability of Gauss--Newton and Alternating Least Squares for CANDECOMC/PARAFAC Decomposition
• A Practical Randomized CP Tensor Decomposition
• Title not available (Why is that?)
• CP decomposition and weighted clique problem

Uses Software

• fminsearch

This page was built for publication: Numerical CP decomposition of some difficult tensors

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q508007)