Skip to main content
Springer logoLink to Springer
. 2018 Apr 6;2018(1):76. doi: 10.1186/s13660-018-1672-4

An upper bound for the Z-spectral radius of adjacency tensors

Zhi-Yong Wu 1, Jun He 1,, Yan-Min Liu 1, Jun-Kang Tian 1
PMCID: PMC5889425  PMID: 29657511

Abstract

Let H be a k-uniform hypergraph on n vertices with degree sequence Δ=d1dn=δ. In this paper, in terms of degree di, we give a new upper bound for the Z-spectral radius of the adjacency tensor of H. Some examples are given to show the efficiency of the bound.

Keywords: Hypergraph, Z-eigenvalue, Bound, Nonnegative tensor

Introduction

Let A=(ai1i2im) be an mth order n-dimensional real square tensor, x be a real n-vector. Then we define the following real n-vector:

Axm1=(i2,,im=1naii2imxi2xim)1in,x[m1]=(xim1)1in.

If there exist a real vector x and a real number λ such that

Axm1=λx[m1],

then λ is called an H-eigenvalue of A and x is called an eigenvector of A associated with λ [1, 2]. If there exist a real vector x and a real number λ such that

Axm1=λx,xTx=1,

then λ is called a Z-eigenvalue of A and x is called an eigenvector of A associated with λ. You can see more about the eigenvalues of tensors in [37].

Let H be a hypergraph with a vertex set V(H) and an edge set E(H)={e1,e2,,et}. If every edge of H contains exactly k distinct vertices, then H is called a k-uniform hypergraph. The degree of a vertex i in H is the number of edges incident with i, denoted by di. If di=d for any iV(H), then the hypergraph H is called a regular hypergraph. Recently, the spectral radii of hypergraphs have been studied in [8, 9].

Let {i1,,ik}E(H) mean that there is an edge containing k distinct vertices i1,,ik. Then the adjacency tensor A(H)=(ai1ik) of a hypergraph H is a kth order n-dimensional tensor with entries:

ai1ik={1(k1)!,if {i1,,ik}E(H),0,otherwise.

Let D(H)=diag(d1,d2,,dn) be the degree diagonal tensor of the graph H. Then the tensor Q(H)=D(H)+A(H) is called the signless Laplacian tensor of the hypergraph H. The largest modulus of the Z-eigenvalues of the adjacency tensor A(H) is denoted by ρZ(H), which is called the Z-spectral radius of the adjacency tensor A(H).

For a k-uniform hypergraph H, let Δ=d1dn=δ be the degree sequence of the hypergraph H. In 2013, Xie and Chang [8] presented the following upper bound for the largest Z-eigenvalues ρZ(H) of adjacency tensors:

ρZ(H)Δ. 1

In this paper, we give a new upper bounds in terms of degree di for the Z-spectral radius of hypergraphs, which improves the bound as shown in (1). Then we give some examples to compare these bounds for Z-spectral radius of hypergraphs.

Preliminaries

Some basic definitions and useful results are listed as follows.

Definition 2.1

([10])

The tensor A is called reducible if there exists a nonempty proper index subset J{1,2,,n} such that ai1,i2,,im=0, i1J, i2,,imJ. If A is not reducible, then we call A to be irreducible.

Definition 2.2

Let A be an m-order and n-dimensional tensor. We define σ(A) the Z-spectrum of A by the set of all Z-eigenvalues of A. Assume σ(A), then the Z-spectral radius of A is denoted by

ρZ(A)=max{|λ|:λσ(A)}.

The concept of weakly symmetric was first introduced and used by Chang, Pearson, and Zhang [11] in order to study the following Perron–Frobenius theorem for the Z-eigenvalue of nonnegative tensors.

Lemma 2.1

([11])

Let A=(ai1i2im) be a weakly symmetric nonnegative tensor, then the spectral radius ρZ(A) is a positive Z-eigenvalue with a nonnegative Z-eigenvector x. Furthermore, if A is irreducible, x is positive.

|A| means that (|A|)i1im=|ai1im|. Two useful lemmas are given as follows.

Lemma 2.2

Let A and B be two weakly symmetric and irreducible tensors of order m and dimension n. If B and B|A| are nonnegative, then ρZ(B)ρZ(|A|).

Proof

Let y be the eigenvector associated with β, where β is a Z-eigenvalue of A. Then we can get

|β||y|=|Ay[m1]||A||y[m1]|B|y[m1]|.

By Theorem 4.7 of [11], we have

ρZ(B)=maxy0min|yi|>0(B|y|[m1])i|yi|min|yi|>0(B|y|[m1])i|yi||β|.

Then

ρZ(B)ρZ(|A|).

 □

Lemma 2.3

Let {Ak} be a sequence of nonnegative, weakly symmetric tensors of order m and dimension n, and AkAk+1 be nonnegative for each positive integer k. Then

limkρZ(Ak)=ρZ(limkAk).

Proof

Let A=limkAk. Since AkAk+1 is nonnegative, by Lemma 2.2, we know that {ρZ(Ak)} is a monotone decreasing sequence with a lower bound ρZ(A). So limkAk exists and

λ=limkρZ(Ak)ρZ(A).

Since {Ak} is nonnegative, weakly symmetric, then there exists a nonnegative vector x(k) such that Ak(x(k))m1=ρZ(Ak)x(k) and (x(k))Tx(k)=1. Then {x(k)} is a bounded sequence, it has a convergent subsequence {yt}. Suppose that y=limkyt. By Akytm1=ρZ(Ak)yt, we get Aym1=λy. So λ is an eigenvalue of A. Since λρZ(A), we have ρZ(A)=λ. □

The Z-spectral radius of tensors and hypergraphs

In this section, let ri(A)=i2,,im=1n|aii2im||aiii|, we give some bounds on the Z-spectral radius of tensors and hypergraphs.

Theorem 3.1

Let A be weakly symmetric nonnegative tensors of order m and dimension n. Then

ρZ(A)maxai1im0{j=1mrij1m(A)}.

Proof

Case 1. If A is irreducible, by Lemma 2.1, let u=(ui) be the positive eigenvector associated with the largest Z-eigenvalues ρZ(A) of A. Then

Aum1=ρZ(A)u.

Let uα=max{ui1uim:ai1im0,1i1,,imn}, then

ρZ(A)ui2=i2,,im=1naii2imuiui2uim=aii2ik0aii2ikuiui2uimri(A)uα. 2

Suppose that uα=uj1ujm. Then, from (2), we can get

ρZ(A)uj12rj1(A)uα,ρZ(A)ujm2rjm(A)uα.

Then, by uαmuα2, we have

l=1mρZm(A)ujl2uαml=1mril(A)uα2l=1mril(A).

Therefore,

ρZ(A)maxai1im0{j=1mrij1m(A)}.

Case 2. If A is reducible. Let T=(ti1i2im), ti1i2im=1 for all 1i1,i2,,imn. Then A+ϵT is an irreducible nonnegative tensor for any chosen positive real number ϵ. Now we substitute A+ϵT for A, respectively, in the previous case. When ϵ0, the result follows by the continuity of ρZ(A+ϵT). □

By Theorem 3.1, a bound on the Z-spectral radius of a uniform hypergraph is obtained, we also compare the bound with the result in (1).

Theorem 3.2

Let H be a k-uniform hypergraph on n vertices with the degree sequence Δ=d1dn=δ. Then

ρZ(H)max{i1,,ik}E(H){j=1kdij1k(A)}. 3

Proof

Case 1. A(H) is irreducible. In this case, by Lemma 2.1, there exists a positive eigenvector corresponding to the spectral radius ρZ(H). Then, by Theorem 3.1, we have

ρZ(H)max{i1,,ik}E(H){j=1kdij1k(A)}.

Case 2. If A(H) is reducible. Let T=(ti1i2ik), ti1i2ik=1, for all 1i1,i2,,ikn. Then A(H)+ϵT is an irreducible nonnegative tensor for any chosen positive real number ϵ. Now we substitute A(H)+ϵT for A(H), respectively, in the previous case. When ϵ0, the result follows by the continuity of ρZ(A(H)+ϵT). □

Remark

Obviously, we can get

max{i1,,ik}E(H){j=1kdij1k(A)}Δ.

That is to say, our bound in Theorem 3.2 is always better than the bound in (1).

We now show the efficiency of the new upper bound in Theorem 3.2 by the following examples.

Example 1

Consider 3-uniform hypergraph H1 with a vertex set V(H1)={1,2,3,4,5,6,7} and an edge set E(H1)={e1,e2,e3}, where e1={1,2,3}, e2={1,4,5}, e3={1,6,7}.

Example 2

Consider 3-uniform hypergraph H2 with a vertex set V(H2)={1,2,3,4,5,6,7} and an edge set E(H2)={e1,e2,e3}, where e1={1,6,7}, e2={2,6,7}, e3={3,6,7}.

From Table 1, we can find that bound (3) is always better than (1).

Table 1.

Upper bounds for the hypergraphs H1 and H2

(1) (3)
H1 3 313
H2 3 323

Conclusion

In this paper, we get a new bound for the Z-spectral radius of tensors. As applications, in terms of the degree sequence di, we obtain a new bound for the Z-spectral radius of hypergraphs, which is always better than the bound in [8]. We list two examples to show the efficiency of our new bound.

Acknowledgements

Wu is supported by the Research Center for Qianbei Culture of Guizhou Higher Education Humanistic and Social Science Research Base Foundation: Kelaofolk Mathematical Investigation and Cultural Inheritance[2015JD114]. He is supported by the Science and Technology Foundation of Guizhou Province (Qian ke he Ji Chu [2016]1161); Guizhou Province Natural Science Foundation in China (Qian Jiao He KY [2016]255); the Doctoral Scientific Research Foundation of Zunyi Normal College (BS[2015]09); High-level Innovative Talents of Guizhou Province (Zun Ke He Ren Cai [2017]8). Liu is supported by the National Science Foundation of China (71461027); Science and Technology Talent Training Object of Guizhou Province Outstanding Youth (Qian ke he ren zi [2015]06); Guizhou Province Natural Science Foundation in China (Qian Jiao He KY [2014]295); 2013, 2014, and 2015 Zunyi 15851 Talents Elite Project funding; Zhunyi Innovative Talent Team (Zunyi KH (2015)38). Tian is supported by Guizhou Province Natural Science Foundation in China (Qian Jiao He KY [2015]451); Science and Technology Foundation of Guizhou Province (Qian ke he J zi [2015]2147).

Authors’ contributions

All authors contributed equally to this work. All authors read and approved the final manuscript.

Competing interests

The authors declare that they have no competing interests.

Footnotes

Publisher’s Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Contributor Information

Zhi-Yong Wu, Email: 2390708509@qq.com.

Jun He, Email: hejunfan1@163.com.

Yan-Min Liu, Email: 546692653@qq.com.

Jun-Kang Tian, Email: 17261717@qq.com.

References

  • 1.Qi L. Eigenvalues of a real supersymmetric tensor. J. Symb. Comput. 2005;40:1302–1324. doi: 10.1016/j.jsc.2005.05.007. [DOI] [Google Scholar]
  • 2.Qi L. Eigenvalues and invariants of tensor. J. Math. Anal. Appl. 2007;325:1363–1377. doi: 10.1016/j.jmaa.2006.02.071. [DOI] [Google Scholar]
  • 3.Yang Y., Yang Q. Further results for Perron–Frobenius theorem for nonnegative tensors. SIAM J. Matrix Anal. Appl. 2010;31:2517–2530. doi: 10.1137/090778766. [DOI] [Google Scholar]
  • 4.Yang Y., Yang Q. Further results for Perron–Frobenius theorem for nonnegative tensors II. SIAM J. Matrix Anal. Appl. 2011;32:1236–1250. doi: 10.1137/100813671. [DOI] [Google Scholar]
  • 5.Li C., Li Y., Kong X. New eigenvalue inclusion sets for tensors. Numer. Linear Algebra Appl. 2014;21:39–50. doi: 10.1002/nla.1858. [DOI] [Google Scholar]
  • 6.He J., Huang T.Z. Upper bound for the largest Z-eigenvalue of positive tensors. Appl. Math. Lett. 2014;38:110–114. doi: 10.1016/j.aml.2014.07.012. [DOI] [Google Scholar]
  • 7.Li W., Liu D. Z-Eigenpair bounds for an irreducible nonnegative tensor. Linear Algebra Appl. 2015;483:182–199. doi: 10.1016/j.laa.2015.05.033. [DOI] [Google Scholar]
  • 8.Xie J., Chang A. On the Z-eigenvalues of the adjacency tensors for uniform hypergraphs. Linear Algebra Appl. 2013;439:2195–2204. doi: 10.1016/j.laa.2013.07.016. [DOI] [Google Scholar]
  • 9.Bu C., Jin X., Li H., Deng C. Brauer-type eigenvalue inclusion sets and the spectral radius of tensors. Linear Algebra Appl. 2017;512:234–248. doi: 10.1016/j.laa.2016.09.041. [DOI] [Google Scholar]
  • 10.Chang K.C., Zhang T., Pearson K. Perron–Frobenius theorem for nonnegative tensors. Commun. Math. Sci. 2008;6:507–520. doi: 10.4310/CMS.2008.v6.n2.a12. [DOI] [Google Scholar]
  • 11.Chang K.C., Zhang T., Pearson K. Some variational principles for Z-eigenvalues of nonnegative tensors. Linear Algebra Appl. 2013;438:4166–4182. doi: 10.1016/j.laa.2013.02.013. [DOI] [Google Scholar]

Articles from Journal of Inequalities and Applications are provided here courtesy of Springer

RESOURCES