Message Board

Dear readers, authors and reviewers,you can add a message on this page. We will reply to you as soon as possible!

2020 Volume 45 Issue 10
Article Contents

Jian-feng HE. On Subdirect Sums of Third-Order Diagonally Dominant Tensors[J]. Journal of Southwest China Normal University(Natural Science Edition), 2020, 45(10): 1-7. doi: 10.13718/j.cnki.xsxb.2020.10.001
Citation: Jian-feng HE. On Subdirect Sums of Third-Order Diagonally Dominant Tensors[J]. Journal of Southwest China Normal University(Natural Science Edition), 2020, 45(10): 1-7. doi: 10.13718/j.cnki.xsxb.2020.10.001

On Subdirect Sums of Third-Order Diagonally Dominant Tensors

More Information
  • Received Date: 21/02/2020
    Available Online: 20/10/2020
  • MSC: O151.21

  • According to the relation between the matrices and tensors, the subdirect sums of tensors has been introduced, which is the generalization of matrices case. Some conditions have been given to guarantee that the k-subdirect sums of third-order strictly diagonally dominant tensors(SDD) is also SDD. The same situation has been analyzed for S-SDD tensors, and examples been given to illustrate the theoretical results presented.
  • 加载中
  • [1] BRU R, PEDROCHE F, SZYLD D B. Subdirect Sums of Nonsingular M-Matrices and of Their Inverse[J]. Electron J Linear Algebra, 2005, 13(1): 162-174.

    Google Scholar

    [2] BRU R, PEDROCHE F, SZYLD D B. Additive Schwarz Iterations for Markov Chains[J]. SIAM J Matrix Anal Appl, 2005, 27(2): 445-458. doi: 10.1137/040616541

    CrossRef Google Scholar

    [3] FROMMER A, SZYLD D B. Weighted Max Norms, Splittings, and Overlapping Additive Schwarz Iterations[J]. Numer Math, 1999, 83(2): 259-278. doi: 10.1007/s002110050449

    CrossRef Google Scholar

    [4] FALLAT S M, JOHMSON C R. Sub-Direct Sums and Positivity Classes of Matrices[J]. Linear Algebra Appl, 1999, 288(1): 149-173. doi: 10.1016/S0024-3795(98)10194-5

    CrossRef Google Scholar

    [5] ZHU Y, HUANG T Z. Subdirect Sums of Doubly Diagonally Dominant Matrices[J]. Electron J Linear Algebra, 2007, 16(1): 171-182.

    Google Scholar

    [6] LI B, TSATSOMEROS M J. Doubly Diagonally Dominant Matrices[J]. Linear Algebra Appl, 1997, 261(1-3): 221-235. doi: 10.1016/S0024-3795(96)00406-5

    CrossRef Google Scholar

    [7] BRU R, CVETKOVIC' L, KOSTIC' V, et al. Characterization of α1 and α2-Matrices[J]. Cent Eur J Math, 2010, 8(1): 32-40. doi: 10.2478/s11533-009-0068-6

    CrossRef Google Scholar

    [8] LI C, LIU Q, GAO L, et al. Subdirect Sums of Nekrasov Matrices[J]. Linear and Multilinear Algebra, 2016, 64(2): 208-218. doi: 10.1080/03081087.2015.1032198

    CrossRef Google Scholar

    [9] ZHU Y, HUANG T, LIU J. Subdirect Sums of H-Matrices[J]. Int J Nonlinear Sci, 2009, 8(1): 50-58.

    Google Scholar

    [10] LI C, MA R, LIU Q, et al. Subdirect Sums of Weakly Chained Diagonally Dominant Matrices[J]. Linear and Multilinear Algebra, 2017, 65(6): 1220-1231. doi: 10.1080/03081087.2016.1233933

    CrossRef Google Scholar

    [11] GAO L, HUANG H, LI C. Subdirect Sums of QN-Matrices[J]. Linear and Multilinear Algebra, 2018, 2018: 1-20.

    Google Scholar

    [12] NIKIAS C L, MENDEL J M. Signal Processing with Higher-Order Spectra[J]. IEEE Signal Process Mag, 1993, 10(3): 10-37.

    Google Scholar

    [13] QI L, WANG F, WANG Y. Z-Eigenvalue Methods for a Global Polynomial Optimization Rroblem[J]. Math Program, 2009, 118: 301-316. doi: 10.1007/s10107-007-0193-6

    CrossRef Google Scholar

    [14] CHING W, HUANG X, MICHAEL KN, et al. Markov Chains: Models, Algorithms and Applications[M]. 2th ed. New York: Springer, 2013.

    Google Scholar

    [15] BOSE N K, KAMAT P S. Algorithm for Stability Test of Multidimensional Filters[J]. IEEE Transactions on Acoustics, Speech, and Signal Processing, 1974, 22(5): 307-314. doi: 10.1109/TASSP.1974.1162592

    CrossRef Google Scholar

    [16] DIAMOND R. A Note on the Rotational Superposition Problem[J]. Acta Cryst, 1988, 44(2): 211-216. doi: 10.1107/S0108767387010535

    CrossRef Google Scholar

    [17] 罗自炎, 祁力群.半正定张量[J].中国科学(数学), 2016, 46(5): 639-654.

    Google Scholar

    [18] QI L Q. Eigenvalues of a Real Supersymmetric Tensor[J]. Journal of Symbolic Computation, 2005, 40(6): 1302-1324. doi: 10.1016/j.jsc.2005.05.007

    CrossRef Google Scholar

    [19] LI C Q, JIAO A Q, LI Y T. An S-Type Eigenvalue Localization Set for Tensors[J]. Linear Algebra Appl, 2016, 493: 469-483. doi: 10.1016/j.laa.2015.12.018

    CrossRef Google Scholar

    [20] 桑彩丽, 赵建兴.非负矩阵张量最大奇异值的S-型上界[J].西南师范大学学报(自然科学版), 2018, 43(6): 1-5.

    Google Scholar

    [21] 桑彩丽, 赵建兴.矩阵张量的S-型奇异值包含集[J].西南师范大学学报(自然科学版), 2019, 44(10): 1-4.

    Google Scholar

  • 加载中
通讯作者: 陈斌, bchen63@163.com
  • 1. 

    沈阳化工大学材料科学与工程学院 沈阳 110142

  1. 本站搜索
  2. 百度学术搜索
  3. 万方数据库搜索
  4. CNKI搜索

Article Metrics

Article views(2536) PDF downloads(245) Cited by(0)

Access History

Other Articles By Authors

On Subdirect Sums of Third-Order Diagonally Dominant Tensors

Abstract: According to the relation between the matrices and tensors, the subdirect sums of tensors has been introduced, which is the generalization of matrices case. Some conditions have been given to guarantee that the k-subdirect sums of third-order strictly diagonally dominant tensors(SDD) is also SDD. The same situation has been analyzed for S-SDD tensors, and examples been given to illustrate the theoretical results presented.

  • 方阵的k-子直和是矩阵和的推广,在许多应用中均有出现,例如矩阵完全化问题、区域分解方法中的重叠子域、有限元中的整体刚度矩阵等[1-3].文献[4]引入了方阵k-子直和的概念,并对其进行了相关研究,证明了正定矩阵的子直和是正定矩阵,对称M-矩阵的子直和是对称M-矩阵等结论.关于方阵k-子直和的一些新的研究成果可参考文献[5-11].

    张量是矩阵的高阶推广,在许多科学领域,如信号图像处理[12]、非线性优化[13]、高阶统计学[14]、物理学中的弹性分析[15-16]和数据挖掘与处理等领域都有重要的应用.有关张量的研究成果可参考文献[17-21],关于张量的更多文献不在此逐一赘述.

    A=(ai1im),其中ai1im$\mathbb{C}$($\mathbb{R}$),ij=1,…,nj=1,…,m,则称A为一个mn维的复(实)张量,记作AC[mn](R[mn]).

    鉴于矩阵与张量之间的关系,本文将矩阵子直和的概念推广到张量上,提出张量子直和的概念,并讨论SDD张量、S-SDD-型张量子直和的性质.

    为方便讨论,引入以下符号:

1.   张量子直和的概念
  • 本节给出张量子直和及相关概念的定义.

    定义1[17]  设张量A=(ai1i2im)∈R[mn],且满足

    则称A为对角占优张量.

    若对每个i∈[n],有

    则称A为严格对角占优张量,简记为SDD张量.

    定义2  设张量A=(ai1im)∈R[mn]n≥2,S是[n]的一个真子集,张量A满足以下两个条件:

    (a) |aii|≥(>)riΔS(A),∀iS

    (b) (|aii|-riΔS(A))(|ajj|-rjΔS(A))≥(>)riΔS(A)rjΔS(A),∀iS,∀jS.

    则称张量AS-对角占优型张量(S-严格对角占优型张量),简记为S-SDD0-型张量(S-SDD-型张量).

    定义3  设张量A=(ai1i2i3)∈R[3,m]B=(bi1i2i3)∈R[3,n],1≤k≤min{mn},称张量

    为张量A与张量Bk-阶子直和,记为A$ \oplus $kB,其中

    例1  设张量A=[A(1,:,:),A(2,:,:),A(3,:,:)],B=[B(1,:,:),B(2,:,:),B(3,:,:)]∈R[3, 3],其中

    则张量A与张量B的2-阶子直和为

    其中

2.   对角占优张量的子直和
  • 本节我们讨论(严格)对角占优张量(S-(严格)对角占优型张量)的子直和仍然为(严格)对角占优张量(S-(严格)对角占优型张量)的条件.

    定理1  设张量A=(ai1i2i3)∈R[3,m]B=(bi1i2i3)∈R[3,n]均为对角占优张量,1≤k≤min{mn},且aiiibjjj>0,∀iS2,∀j∈[k],则张量A与张量Bk-阶子直和C=A$ \oplus $kB也是对角占优张量.

     当i1S1时,

    故有

    i1S2时,

    故有

    i1S3时,

    故有

    综合以上3种情形可知,结论成立.

    例2  设张量A=[A(1,:,:),A(2,:,:),A(3,:,:)],B=[B(1,:,:),B(2,:,:),B(3,:,:)],其中

    易知张量AB均为对角占优张量,而张量A与张量B的2-阶子直和为

    其中

    计算知

    故张量C为对角占优张量.

    类似定理1的证明可得如下结论:

    定理2 看设张量A=(ai1i2i3)∈R[3,m]B=(bi1i2i3)∈R[3,n]均为严格对角占优张量,1≤k≤min{mn},且aiiibjjj>0,∀iS2,∀j∈[k],则张量A与张量Bk-阶子直和也是严格对角占优张量.

    下面讨论S-SDD-型张量的子直和.先看一个例子.

    例3  设张量A=(ai1i2i3)∈R[3, 4],其中

    S={1,2},由于

    A为{1,2}-SDD-型张量.但是对于张量C=A$ \oplus $2A,其中

    i=1,j=5时,有

    即张量C不是{1,2}-SDD-型张量.

    例3表明,S-SDD-型张量的子直和不一定是S-SDD-型张量.因此,寻找S-SDD-型张量的子直和仍然为S-SDD-型张量的条件是有意义的.

    定理3  设张量A=(ai1i2i3)∈R[3,m]S-SDD-型张量,SS1的子集,B=(bi1i2i3)∈R[3,n]为SDD张量,1≤k≤min{mn},aiii>0,∀iS2bjjj>0,j∈[k],则张量A与张量Bk-阶子直和C=A$ \oplus $kB也是S-SDD-型张量.

      先证明S=S1时的情形.

    要证明CS-SDD-型张量,需证明:

    (i) |ciii|>riΔS(C),∀iS

    (ii) (|ciii|-riΔS(C))(|cjjj|-rjΔS(C))>riΔS(C)rjΔS(C),∀iS,∀jS.

    此时,由于S=S1,则S=S2S3.下面逐一证明这两个条件成立.

    (i) 由张量AS-SDD-型张量知|aiii|>riΔS(A),∀iS1,从而有

    (ii) 当jS2时,由iS=S1,有

    由张量AS-SDD-型张量得

    又因B为SDD张量,故|biii|>ri(B),∀i∈[n].从而有

    jS3时,由iS=S1rjΔS(C)=0,rjΔS(C)=rjtΔS(B),故

    综上所述,可知结论成立.

    当|S| < |S1|时,可类似证明结论成立.

    例4  取例3中的S-SDD-型张量A,例2中的SDD张量BS={1,2},则C=A$ \oplus $2B=[C(1,:,:),C(2,:,:),C(3,:,:),C(4,:,:),C(5,:,:)],其中

    由于

    所以,张量CS-SDD-型张量.

Reference (21)

Catalog

    /

    DownLoad:  Full-Size Img  PowerPoint
    Return
    Return