# Theory and Computation of Tensors. Multi-Dimensional Arrays by Yimin Wei, Weiyang Ding

By Yimin Wei, Weiyang Ding

*Theory and Computation of Tensors: Multi-Dimensional Arrays* investigates theories and computations of tensors to develop views on matrices. facts within the monstrous information period isn't just turning out to be greater but additionally turning into even more complex. Tensors (multi-dimensional arrays) come up obviously from many engineering or medical disciplines simply because they could symbolize multi-relational info or nonlinear relationships.

- Provides an advent of contemporary effects approximately tensors
- Investigates theories and computations of tensors to expand views on matrices
- Discusses easy methods to expand numerical linear algebra to numerical multi-linear algebra
- Offers examples of the way researchers and scholars can have interaction in study and the functions of tensors and multi-dimensional arrays

**Read or Download Theory and Computation of Tensors. Multi-Dimensional Arrays PDF**

**Similar linear books**

**A first course in linear algebra**

A primary direction in Linear Algebra is an creation to the elemental thoughts of linear algebra, besides an advent to the suggestions of formal arithmetic. It starts off with platforms of equations and matrix algebra sooner than entering into the idea of summary vector areas, eigenvalues, linear ameliorations and matrix representations.

**Measure theory/ 3, Measure algebras**

Fremlin D. H. degree thought, vol. three (2002)(ISBN 0953812936)(672s)-o

**Elliptic Partial Differential Equations**

Elliptic partial differential equations is likely one of the major and such a lot energetic components in arithmetic. In our e-book we research linear and nonlinear elliptic difficulties in divergence shape, with the purpose of delivering classical effects, in addition to newer advancements approximately distributional options. accordingly the e-book is addressed to master's scholars, PhD scholars and someone who desires to start study during this mathematical box.

- Linear algebra c-2
- Handbook of Linear Algebra, Second Edition (Discrete Mathematics and Its Applications)
- Functional Analysis in Modern Applied Mathematics
- Algebra Can Be Fun
- Linear Spaces and Approximation / Lineare Räume und Approximation: Proceedings of the Conference held at the Oberwolfach Mathematical Research Institute, Black Forest, August 20–27,1977 / Abhandlungen zur Tagung im Mathematischen Forschungsinstitut Oberwo
- Groups of Lie Type and their Geometries (London Mathematical Society Lecture Note Series)

**Additional info for Theory and Computation of Tensors. Multi-Dimensional Arrays**

**Sample text**

H2 (b) H:,:,2 = . . .. 2) .. .. HJ1 HJ1 +1 · · · HJ1 +J2 −2 .. .. .. . . HJ3 −1 HJ3 ··· HJ2 +J3 −3 . . . . .. HJ3 (b) H:,:,J3 = .. . .. .. HJ1 +J3 −2 HJ1 +J3 −1 · · · HJ1 +J2 +J3 −4 HJ2 +J3 −2 .. . . HJ1 +J2 +J3 −4 HJ1 +J2 +J3 −3 Then the BHHB tensor H has the level-2 Vandermonde decomposition ⊗ Z1,1 ) ×2 (Z2,2 ⊗ Z1,2 ) · · · ×m (Z2,m ⊗ Z1,m ) , H = C ×1 (Z2,1 KR KR KR 44 CHAPTER 3. FAST TENSOR-VECTOR PRODUCTS where C is a diagonal tensor with diagonal entries {ck }K k=1 , each a Vandermonde matrix Ip −1 2 2 1 z1,1 z1,1 · · · z1,1 1 z2,1 z2,1 Ip −1 2 2 1 z1,2 z1,2 · · · z1,2 1 z2,2 z2,2 Z1,p = ..

Let A be a complex tensor, E be a nonnegative tensor, b be a complex vector, and d be a nonnegative vector. Define Σ := x ∈ Cn : A + E xm−1 = b + d, E ≤ E, d ≤ d . Then Σ = x ∈ Cn : Axm−1 − b ≤ E|x|m−1 + d . Proof. On the one hand, if x ∈ Σ, then Axm−1 − b = −Exm−1 + d ≤ E|x|m−1 + d. On the other hand, if Axm−1 − b ≤ E|x|m−1 + d, then there exist two signature matrices S1 and S2 such that S1 Axm−1 − b = Axm−1 − b and S2 x = |x|, which indicates that S1 Axm−1 − b ≤ ES2m−1 xm−1 + d. Thus there is a diagonal matrix D with |D| ≤ I such that Axm−1 − b = S1∗ DES2m−1 xm−1 + S1∗ Dd.

We first construct • 3rd -order square Hankel tensors of size n × n × n (n = 10, 20, . . , 100), and • 3rd -order square BHHB tensors of level-1 size n1 × n1 × n1 and level-2 size n2 × n2 × n2 (n1 , n2 = 5, 6, . . , 12). Then we compute the tensor-vector products H ×2 x2 ×3 x3 using • our proposed fast algorithm based on FFT, and • the non-structured algorithm based on the definition. The average running times of 1000 products are shown in Fig. 2. From the results, we can see that the running time of our algorithm increases far more slowly than that of the non-structured algorithm just as predicted by the theoretical analysis.