Low rank Tucker-type tensor approximation to classical potentials
- 1 September 2007
- journal article
- Published by Walter de Gruyter GmbH in Central European Journal of Mathematics
- Vol. 5 (3), 523-550
- https://doi.org/10.2478/s11533-007-0018-0
Abstract
This paper investigates best rank-(r 1,..., r d ) Tucker tensor approximation of higher-order tensors arising from the discretization of linear operators and functions in ℝ d . Super-convergence of the best rank-(r 1,..., r d ) Tucker-type decomposition with respect to the relative Frobenius norm is proven. Dimensionality reduction by the two-level Tucker-to-canonical approximation is discussed. Tensor-product representation of basic multi-linear algebra operations is considered, including inner, outer and Hadamard products. Furthermore, we focus on fast convolution of higher-order tensors represented by the Tucker/canonical models. Optimized versions of the orthogonal alternating least-squares (ALS) algorithm is presented taking into account the different formats of input data. We propose and test numerically the mixed CT-model, which is based on the additive splitting of a tensor as a sum of canonical and Tucker-type representations. It allows to stabilize the ALS iteration in the case of “ill-conditioned” tensors. The best rank-(r 1,..., r d ) Tucker decomposition is applied to 3D tensors generated by classical potentials, for example \(\tfrac{1}{{\left| {x - y} \right|}}, e^{ - \alpha \left| {x - y} \right|} , \tfrac{{e^{ - \left| {x - y} \right|} }}{{\left| {x - y} \right|}}\) and \(\tfrac{{erf(|x|)}}{{|x|}}\) with x, y ∈ ℝ d . Numerical results for tri-linear decompositions illustrate exponential convergence in the Tucker rank, and robustness of the orthogonal ALS iteration.
Keywords
This publication has 24 references indexed in Scilit:
- A Link between the Canonical Decomposition in Multilinear Algebra and Simultaneous Matrix DiagonalizationSIAM Journal on Matrix Analysis and Applications, 2006
- Low-rank Kronecker-product Approximation to Multi-dimensional Nonlocal Operators. Part II. HKT Representation of Certain OperatorsComputing, 2005
- Low-rank Kronecker-product Approximation to Multi-dimensional Nonlocal Operators. Part I. Separable Approximation of Multi-variate FunctionsComputing, 2005
- Hierarchical Tensor-Product Approximation to the Inverse and Related Operators for High-Dimensional Elliptic ProblemsComputing, 2004
- MATLAB tensor classes for fast algorithm prototyping.Published by Office of Scientific and Technical Information (OSTI) ,2004
- Computation of the Canonical Decomposition by Means of a Simultaneous Generalized Schur DecompositionSIAM Journal on Matrix Analysis and Applications, 2004
- Numerical operator calculus in higher dimensionsProceedings of the National Academy of Sciences, 2002
- A Multilinear Singular Value DecompositionSIAM Journal on Matrix Analysis and Applications, 2000
- On the Best Rank-1 and Rank-(R1 ,R2 ,. . .,RN) Approximation of Higher-Order TensorsSIAM Journal on Matrix Analysis and Applications, 2000
- Analysis of individual differences in multidimensional scaling via an n-way generalization of “Eckart-Young” decompositionPsychometrika, 1970