# Best rank-k approximations for tensors: generalizing Eckart–Young

Jan Draisma, Giorgio Ottaviani, Alicia Tocino

### Uittreksel

Given a tensor f in a Euclidean tensor space, we are interested in the critical points of the distance function from f to the set of tensors of rank at most k, which we call the critical rank-at-most-k tensors for f. When f is a matrix, the critical rank-one matrices for f correspond to the singular pairs of f. The critical rank-one tensors for f lie in a linear subspace Hf, the critical space of f. Our main result is that, for any k, the critical rank-at-most-k tensors for a sufficiently general f also lie in the critical space Hf. This is the part of Eckart–Young Theorem that generalizes from matrices to tensors. Moreover, we show that when the tensor format satisfies the triangle inequalities, the critical space Hf is spanned by the complex critical rank-one tensors. Since f itself belongs to Hf, we deduce that also f itself is a linear combination of its critical rank-one tensors.

Taal Engels 27 13 Research in the Mathematical Sciences 5 2 10.1007/s40687-018-0145-1 Gepubliceerd - 1 jun 2018

### Vingerafdruk

Tensors
Tensor
Approximation
H-space
Triangle inequality
Distance Function
Linear Combination
Deduce
Critical point
Euclidean
Subspace
Generalise
Theorem

### Citeer dit

@article{50435cb7cec24f7fa0c8825ce26edc6d,
title = "Best rank-k approximations for tensors: generalizing Eckart–Young",
abstract = "Given a tensor f in a Euclidean tensor space, we are interested in the critical points of the distance function from f to the set of tensors of rank at most k, which we call the critical rank-at-most-k tensors for f. When f is a matrix, the critical rank-one matrices for f correspond to the singular pairs of f. The critical rank-one tensors for f lie in a linear subspace Hf, the critical space of f. Our main result is that, for any k, the critical rank-at-most-k tensors for a sufficiently general f also lie in the critical space Hf. This is the part of Eckart–Young Theorem that generalizes from matrices to tensors. Moreover, we show that when the tensor format satisfies the triangle inequalities, the critical space Hf is spanned by the complex critical rank-one tensors. Since f itself belongs to Hf, we deduce that also f itself is a linear combination of its critical rank-one tensors.",
keywords = "Best rank-k approximation, Eckart–Young Theorem, Tensor",
author = "Jan Draisma and Giorgio Ottaviani and Alicia Tocino",
year = "2018",
month = "6",
day = "1",
doi = "10.1007/s40687-018-0145-1",
language = "English",
volume = "5",
journal = "Research in the Mathematical Sciences",
issn = "2522-0144",
publisher = "Springer",
number = "2",

}

Best rank-k approximations for tensors : generalizing Eckart–Young. / Draisma, Jan; Ottaviani, Giorgio; Tocino, Alicia.

In: Research in the Mathematical Sciences, Vol. 5, Nr. 2, 27, 01.06.2018.

TY - JOUR

T1 - Best rank-k approximations for tensors

T2 - Research in the Mathematical Sciences

AU - Draisma,Jan

AU - Ottaviani,Giorgio

AU - Tocino,Alicia

PY - 2018/6/1

Y1 - 2018/6/1

N2 - Given a tensor f in a Euclidean tensor space, we are interested in the critical points of the distance function from f to the set of tensors of rank at most k, which we call the critical rank-at-most-k tensors for f. When f is a matrix, the critical rank-one matrices for f correspond to the singular pairs of f. The critical rank-one tensors for f lie in a linear subspace Hf, the critical space of f. Our main result is that, for any k, the critical rank-at-most-k tensors for a sufficiently general f also lie in the critical space Hf. This is the part of Eckart–Young Theorem that generalizes from matrices to tensors. Moreover, we show that when the tensor format satisfies the triangle inequalities, the critical space Hf is spanned by the complex critical rank-one tensors. Since f itself belongs to Hf, we deduce that also f itself is a linear combination of its critical rank-one tensors.

AB - Given a tensor f in a Euclidean tensor space, we are interested in the critical points of the distance function from f to the set of tensors of rank at most k, which we call the critical rank-at-most-k tensors for f. When f is a matrix, the critical rank-one matrices for f correspond to the singular pairs of f. The critical rank-one tensors for f lie in a linear subspace Hf, the critical space of f. Our main result is that, for any k, the critical rank-at-most-k tensors for a sufficiently general f also lie in the critical space Hf. This is the part of Eckart–Young Theorem that generalizes from matrices to tensors. Moreover, we show that when the tensor format satisfies the triangle inequalities, the critical space Hf is spanned by the complex critical rank-one tensors. Since f itself belongs to Hf, we deduce that also f itself is a linear combination of its critical rank-one tensors.

KW - Best rank-k approximation

KW - Eckart–Young Theorem

KW - Tensor

UR - http://www.scopus.com/inward/record.url?scp=85050407297&partnerID=8YFLogxK

U2 - 10.1007/s40687-018-0145-1

DO - 10.1007/s40687-018-0145-1

M3 - Article

VL - 5

JO - Research in the Mathematical Sciences

JF - Research in the Mathematical Sciences

SN - 2522-0144

IS - 2

M1 - 27

ER -