756
Views
5
CrossRef citations to date
0
Altmetric
Articles

Graphs determined by signless Laplacian spectra

, &

Abstract

In the past decades, graphs that are determined by their spectrum have received more attention, since they have been applied to several fields, such as randomized algorithms, combinatorial optimization problems and machine learning. An important part of spectral graph theory is devoted to determining whether given graphs or classes of graphs are determined by their spectra or not. So, finding and introducing any class of graphs which are determined by their spectra can be an interesting and important problem. A graph is said to be DQS if there is no other non-isomorphic graph with the same signless Laplacian spectrum. For a DQS graph G, we show that GrK1sK2 is DQS under certain conditions, where r, s are natural numbers and K1 andK2 denote the complete graphs on one vertex and two vertices, respectively. Applying these results, some DQS graphs with independent edges and isolated vertices are obtained.

1 Introduction

Let G=(V,E) be a simple graph with vertex set V=V(G)=v1,,vn and edge set E=E(G)=e1,,em. Denote by d(v) the degree of vertex v. All graphs considered here are simple and undirected. All notions on graphs that are not defined here can be found in [Citation1–5]. The join of two graphs G and H is a graph formed from disjoint copies of G and H by connecting each vertex of G to each vertex of H. We denote the join of two graphs G and H by GH. The complement of a graph G is denoted by G¯.

Let A(G) be the (0,1)-adjacency matrix of graph G. The characteristic polynomial of G is det(λIA(G)), and it is denoted by PG(λ). Let λ1, λ2, …, λn be the distinct eigenvalues of G with multiplicities m1, m2, …, mn, respectively. The multi-set of eigenvalues of Q(G) is called the signless Laplacian spectrum of G. The matrices L(G)=D(G)A(G) and Q(G)=SL(G)=D(G)+A(G) are called the Laplacian matrix and the signless Laplacian matrix of G, respectively, where D(G) denotes the degree matrix. Note that D(G) is diagonal. The multi-set SpecQ(G)=[λ1]m1,[λ2]m2,,[λn]mn of eigenvalues of Q(G) is called the signless Laplacian spectrum of G, where mi denote the multiplicities of λi. The Laplacian spectrum is defined analogously.

For any bipartite graph, its Q-spectrum coincides with its L-spectrum. Two graphs are Q-cospectral (resp. L-cospectral, A-cospectral) if they have the same Q-spectrum (resp. L-spectrum, A-spectrum). A graph G is said to be DQS (resp. DLS, DAS) if there is no other non-isomorphic graph Q-cospectral (resp. L-cospectral, A-cospectral) with G. Van Dam and Haemers [Citation6] conjectured that almost all graphs are determined by their spectra. Nevertheless, the set of graphs that are known to be determined by their spectra is too small. So, discovering infinite classes of graphs that are determined by their spectra can be an interesting problem. About the background of the question “Which graphs are determined by their spectrum?”, we refer to [Citation6]. It is interesting to construct new DQS (DLS) graphs from known DQS (DLS) graphs. For a DLS graph G, the join GrK1 is also DLS under some conditions [Citation7]. Actually, a graph is DLS if and only if its complement is DLS. Hence we can obtain DLS graphs from known DLS graphs by adding independent edges.

Up to now, only some graphs with special structures are shown to be determined by their spectra (DS, for short) (see [Citation1,8–30] and the references cited in them).

In this paper, we investigate signless Laplacian spectral characterization of graphs with independent edges and isolated vertices. For a DQS graph G, we show that GrK1sK2 is DQS under certain conditions. Applying these results, some DQS graphs with independent edges and isolated vertices are obtained.

2 Some definitions and preliminaries

Some useful established results about the spectrum are presented in this section, will play an important role throughout this paper.

Lemma 2.1

[Citation4,9,17] For the adjacency matrix, the Laplacian matrix and the signless Laplacian matrix of a graph G, the following can be deduced from the spectrum:

(1) The number of vertices.

(2) The number of edges.

(3) Whether G is regular.

For the Laplacian matrix, the following follows from the spectrum:

(4) The number of components.

For the signless Laplacian matrix, the following follow from the spectrum:

(5) The number of bipartite components.

(6) The sum of the squares of degrees of vertices.

Lemma 2.2

[Citation17] Let G be a graph with n vertices, m edges and t triangles and vertex degrees d1,d2,,dn. Let Tk=i=1n(qi(G))k, then T0=n,T1=i=1ndi=2m,T2=2m+i=1ndi2 and T3=6t+3i=1ndi2+i=1ndi3.

For a graph G, let PL(G) and PQ(G) denote the product of all nonzero eigenvalues of LG and QG, respectively. We assume that PL(G)=PQ(G)=1 if G has no edges.

Lemma 2.3

[Citation4] For any connected bipartite graph G of order n, we have PQ(G)=PL(G)=nτ(G), where τ(G) is the number of spanning trees of G.

For a connected graph G with n vertices and m edges, G is called unicyclic (resp. bicyclic) if m=n (resp. m=n+1). If G is a unicyclic graph that contains an odd (resp. even) cycle, then G is called odd unicyclic (resp. even unicyclic).

Lemma 2.4

[Citation31] For any graph G, det(QG)=4 if and only if G is an odd unicyclic graph. If G is a non-bipartite connected graph and |E(G)|>|V(G)|, then det(QG)>16, with equality if and only if G is a non-bipartite bicyclic graph with C4 as its induced subgraph.

Lemma 2.5

[Citation32] Let H be a proper subgraph of a connected graph G. Then, q1(G)>q1(H).

3 Main results

We first investigate spectral characterizations of the union of a tree and several complete graphs K1 and K2.

Theorem 3.1

Let T be a DLS( DQS) tree of order n. Then TrK1sK2 is DLS. TrK1sK2 isDQS ifn is not divisible by 2 and s=1.

Proof

Let G be any graph L-cospectral with TrK1sK2. By Lemma 2.1, G has n+r+2s vertices, n1+s edges and r+s+1 components. So each component of G is a tree. Suppose that G=G0G1Gr+s, where Gi is a tree with ni vertices and n0n1nsnr+s1. Since G is L-cospectral with TrK1sK2, by Lemma 2.3, we get n0n1nr+s=PL(G)=n2s. We claim that ns=2. Suppose not and so ns3. Therefore, n0n1ns3 and since ns+1nr+s1, one may deduce that n2s=n0n1nr+s3s+1 or n(23)s3. Now if s, then 03, a contradiction. Hence ns=2. By a similar argument one may show that n1=n2==ns1=2 and so n0=n and ns+1=ns+2==ns+r=1. Hence G=G0rK1sK2. Since G and TrK1sK2 are L-cospectral, G0 and T are L-cospectral. Since T is DLS, we have G0=T, G=TrK1sK2. Hence TrK1sK2 is DLS. Let H be any graph Q-cospectral with TrK1sK2. By Lemma 2.1, H has n+r+2s vertices, n1+s edges and r+s+1 bipartite components. So one of the following holds:

(i) H has exactly r+s+1 components, and each component of H is a tree.

(ii) H has r+s+1 components which are trees, the other components of H are odd unicyclic.

If (i) holds, then H and TrK1sK2 are both bipartite, so they are also L-cospectral. Since TrK1sK2 is DLS, we have H=TrK1sK2.

If (ii) holds, then by Lemma 2.4, PQ(H) is divisible by 4. Since T is a tree of order n, by Lemma 2.3, PQ(H)=n2s is divisible by 4. Hence TrK1sK2 is DQS when n is not divisible by 2 and s=1.

Remark 1

Some DLS trees are given in [33–38]. We can obtain DLS (DQS) graphs with independent edges and isolated vertices from Theorem 3.1.

Theorem 3.2

Let G be a DQS odd unicyclic graph of order n7. Then GrK1sK2 is DQS.

Proof

Let H be any graph Q-cospectral with GrK1sK2. By Lemma 2.4, PQ(H)=4(2s). By Lemma 2.1, H has n+r+2s vertices, n+r edges and r+s bipartite components. So one of the following holds:

(i) H has exactly r+s components, and each component of H is a tree.

(ii) H has r+s components which are trees, the other components of H are odd unicyclic.

If (i) holds, then we can let H=H1Hr+s, where Hi is a tree with ni vertices and n1nr+s1. Since PQ(H)=4(2s), by Lemma 2.3, we have n1nr+s=4(2s), n18.

Since G contains a cycle, we have q1(H)=q1(G)4. Let Δ(H) be the maximum degree of H. If Δ(H)2, then all components of H are paths, i.e., q1(H)<4, a contradiction. So Δ(H)>3. From n18 and n1nr+s=4(2s)=2(s+2), we know that H1=K1,7 (without loss of generality), H2==Hs=K2 and Hs+1==Hr+s=K1. Since H=K1,7(s1)K2rK1 has n+r+2s vertices, we get n=6, a contradiction to n>6.

If (ii) holds, then we can let H=U1UcH1Hr, where Ui is odd unicyclic, Hi is a tree with ni vertices. By Lemmas 2.3 and 2.4, 4(2s)=PQ(H)=4cn1nr. So c=1, H1==Hs=K2 and Hs+1==Hr+s=K1. Since H=U1rK1sK2 and GrK1sK2 are Q-cospectral, U1 and G are Q-cospectral. Since G is DQS, we have U1=G, H=GrK1sK2.

Remark 2

Some DQS unicyclic graphs are given in [39–44]. We can obtain DQS graphs with independent edges and isolated vertices from Theorem 3.2.

Theorem 3.3

Let G be a non-bipartite DQS bicyclic graph with C4 as its induced subgraph and n5. Then GrK1sK2 is DQS.

Proof

Let H be any graph Q-cospectral with GrK1sK2. By Lemma 2.4, we have PQ(H)=16(2r). By Lemma 2.1, H has n+r+2s vertices, n+1+s edges and r+s bipartite components, where n=|V(G)|. So H has at least r+s1 components which are trees. Suppose that H1,H2,,Hr+s are r+s bipartite components of H, where H2,,Hr are trees. If H1 contains an even cycle, then by Lemma 2.3, we have PQ(H)PQ(H1)16, and PQ(H)=16(2s1) if and only if H=C4(s1)K2rK1. Since H has n+r+2s vertices, we get n=2, a contradiction (G contains C4). Hence H1,H2,,Hr+s are trees. Since H has n+r+2s vertices, n+1+r+2s edges and r+s bipartite components, H has a non-bipartite component H0 which is a bicyclic graph. Lemma 2.4 implies that PQ(H)>PQ(H0)>16, and PQ(H)=16(2s) if and only if H=H0rK1sK2 and H0 contains C4 as its induced subgraph. By PQ(H)=16(2s), we have H=H0rK1sK2. Since H and GrK1sK2 are Q-cospectral, H0 and G are Q-cospectral. Since G is DQS, we have H0=G,H=GrK1sK2. Hence GrK1sK2 is DQS.

Remark 3

Some DQS bicyclic graphs are given in [45–48]. We can obtain DQS graphs with independent edges and isolated vertices from Theorem 3.3.

Theorem 3.4

Let G be a DQS connected non-bipartite graph with n3 vertices. If H is Q-cospectral with GrK1sK2, then H is a DQS graph.

Proof

By Lemma 2.1, H has n+r+2s vertices and at least r+s bipartite components. We perform the mathematical induction on s.

H has r+s components. Since H has at least r+s bipartite components, each component of H is bipartite. Suppose that H=H1Hr+s, where Hi is a connected bipartite graph with ni vertices, and n1nsnr+s1. Since H and GrK1sK2 are Q-cospectral, by Lemma 2.1, G is a connected non-bipartite graph.

Let s=1. For n3, q1(G)3, since G has K1,2 or K3 as its subgraph. Obviously SpecQ(H) has exactly r+s eigenvalues that are zero. We show that if H is Q-cospectral with GrK1K2, then H is a DQS graph. First we show that there is no connected graph Q-cospectral with SpecQ(G)=SpecQ(G)21. In fact we prove that G cannot have 2 as its eigenvalue. Obviously, SpecQ(H)=SpecQ(G)0r+1. But, in this case |E(G)|=|E(G)|+1 and |V(G)|=|V(G)|+1, which means that G must be connected. Otherwise, G contains 0 as its signless eigenvalues, a contradiction. Therefore, G is a proper subgraph of G and so q1(G)q1(G)3 (see Lemma 2.5), a contradiction. Therefore, G cannot have 2 as its eigenvalue. By what was proved one can easily conclude that SpecQ(H)=SpecQ(G)SpecQ(K2)SpecQ(rK1), since G is not a bipartite graph and so has not 0 as an its signless Laplacian eigenvalue. Therefore, H=GK2rK1.

Now, let the theorem be true for s; that is, if SpecQ(G1)=SpecQ(G)SpecQ(rK1sK2), then G1=GrK1sK2. We show that it follows from SpecQ(K)=SpecQ(G)SpecQ(rK1(s+1)K2) that K=GrK1(s+1)K2. Obviously, K has 2 vertices, one edge and one bipartite component more than G1. So, we must have K=G1K2.

Remark 4

In the following results graph G in GrK1sK2 is a connected non-bipartite.

Corollary 3.1

The graph KnrK1sK2 is DQS.

Proof

From [Citation6] (Proposition 7), if n=1,2, then KnrK1sK2 is DQS. For n3, by Theorem 3.4 the result follows.

In [Citation49], Cámara and Haemers proved that a graph obtained from Kn by deleting a matching is DAS. In [Citation50], it have been shown that this graph is also DQS.

Corollary 3.2

Let G be the graph obtained from Kn by deleting a matching. Then GrK1sK2 is DQS.

Proof

From [Citation6] (Proposition 7), if n=1,2, then KnrK1sK2 is DQS. For n3, by Theorem 3.4 the result follows.

A regular graph is DQS if and only if it is DAS [Citation6]. It is known that a k-regular graph of order n is DAS when k=0,1,2,n1,n2,n3 [Citation17]. Hence a k-regular graph of order n is DQS when k=0,1,2,n1,n2,n3.

Corollary 3.3

Let G be a connected (n2)-regular graph of order n. Then GrK1sK2 is DQS.

Corollary 3.4

Let G be a connected (n3)-regular graph of order n. Then GrK1sK2 is DQS.

Corollary 3.5

Let G be a connected (n4)-regular DAS graph. Then GrK1sK2 is DQS.

Remark 5

Some 3-regular DAS graphs are given in [Citation6,51]. We can obtain DQS graphs with independent edges and isolated vertices and isolated vertices from Corollary 3.4.

Corollary 3.6

Let Fn denote the friendship graph and G be Q-spectral with Fn, then GrK1sK2 is DQS.

Proof

It is well-known that Fn is DQS. By Theorem 3.4 the proof is completed.

References

  • LiuX., PengliL., Signless Laplacian spectral characterization of some joins, Electron. J. Linear Algebra, 3012015 30
  • BapatR.B., Graphs and Matrices 2010Springer-Verlag New York
  • BiggsN.L., Algebraic Graph Theory 1933Cambridge University press Cambridge
  • CvetkovićD., RowlinsonP., SimićS., An introduction to the theory of graph spectra London Mathematical Society Student Teyts, vol. 75 2010Cambridge University Press Cambridge
  • WestD.B., Introduction to Graph Theory 2001Prentice hall Upper Saddle River
  • Van DamE.R., HaemersW.H., Which graphs are determined by their spectrum?, Linear Algebra. Appl., 373 2003 241–272
  • LizhenX., HeC., On the signless Laplacian spectral determination of the join of regular graphs., Discrete Math. Algorithms Appl., 6042014 1450050
  • A.Z. Abdian, Lowell W. Beineke, A. Behmaram On the spectral determinations of the connected multicone graphs Kr▽sKt, arXiv preprintarXiv:1806.02625.
  • AbdianA.Z., MirafzalS.M., On new classes of multicone graph determined by their spectrums, Alg. Struc. Appl., 2 2015 23–34
  • AbdianA.Z., Graphs which are determined by their spectrum, Konuralp J. Math., 4 2016 34–41
  • AbdianA.Z., Two classes of multicone graphs determined by their spectra, J. Math. Ext., 10 2016 111–121
  • AbdianA.Z., Graphs cospectral with multicone graphs Kw▽L(P), TWMS. J. App and Eng. Math., 7 2017 181–187
  • A.Z. Abdian, The spectral determination of the multicone graphs Kw▽P, 2017. arXiv preprintarXiv:1703.08728.
  • AbdianA.Z., MirafzalS.M., The spectral characterizations of the connected multicone graphs Kw▽LHS and Kw▽LGQ(3,9), Discrete Math. Algorithms Appl., 10 2018 1850019
  • AbdianA.Z., MirafzalS.M., The spectral determinations of the connected multicone graphs Kw▽mP17 and Kw▽mS, Czechoslovak Math. J.2018 1–14 10.21136/CMJ.2018.0098-17
  • A.Z. Abdian, The spectral determinations of the multicone graphs Kw▽mCn, arXiv preprintarXiv:1703.08728.
  • BuC., ZhouJ., Signless Laplacian spectral characterization of the cones over some regular graphs, Linear Algebra Appl., 436 2012 3634–3641
  • CvetkovićD., RowlinsonP., SimićS., Signless Laplacians of finite graphs, Linear Algebra Appl., 42312007 155–171
  • CvetkovićD., SimićS., Towards a spectral theory of graphs based on the signless Laplacian, II, Linear Algebra Appl., 432 2010 2257–2272
  • DasK.C., On conjectures involving second largest signless Laplacian eigenvalue of graphs, Linear Algebra Appl., 43 2010 2
  • HaemersW.H., LiuX.G., ZhangY.P., Spectral characterizations of lollipop graphs, Linear Algebra Appl., 428 2008 2415–2423 3018–3029
  • MerrisR., Laplacian matrices of graphs: A survey, Linear Algebra Appl., 197 1994 143–176
  • MirafzalS.M., AbdianA.Z., Spectral characterization of new classes of multicone graphs, Stud. Univ. Babeş-Bolyai Math, 6232017 275–286
  • MirafzalS.M., AbdianA.Z., The spectral determinations of some classes of multicone graphs, J. Discrete Math. Sci. Cryptogr., 2112018 179–189
  • SharafdiniR., AbdianA.Z., Signless Laplacian determinations of some graphs with independent edges, Carpathian Mathematical Publication2018 in press
  • YiW., YizhengF., YingyingT., On graphs with three distinct Laplacian eigenvalues, Appl. Math. J. Chinese Univ. Ser. A, 22 2007 478–484
  • WangJ., ZhaoH., HuangQ., Spectral charactrization of multicone graphs, Czechoslovak Math. J., 62 2012 117–126
  • GünthardHS.H., PrimasH., Zusammenhang von Graph theory und Mo-Theorie von Molekeln mit Systemen konjugierter Bindungen, Helv. Chim. Acta, 39 1925 1645–1653
  • LiuM.H., LiuB.L., Some results on the Laplacian spectrum, J. Comput. Appl. Math., 59 2010 3612–3616
  • ZhouJ., BuC., Laplacian spectral characterization of some graphs obtained by product operation, Discrete Math., 312 2012 1591–1595
  • MirzakhahM., KianiD., The sun graph is determined by its signless Laplacian spectrum, Electron. J. Linear Algebra., 20 2010 610–620
  • ChenM., ZhouB., On the signless Laplacian spectral radius of cacti, Croat. Chem. Acta, 8942016 1–6
  • AalipourG., AkbariS., ShajariN., Laplacian spectral characterization of two families of trees, Linear Multilinear Algebra, 62 2014 965–977
  • BouletR., The centipede is determined by its Laplacian spectrum, C. R. Acad. Sci., Paris I, 346 2008 711–716
  • BuC., ZhouJ., LiH., Spectral determination of some chemical graphs, Filomat, 26 2012 1123–1131
  • LuP.L., ZhangX.D., ZhangY., Determination of double quasi-star tree from its Laplacian spectrum, J. Shanghai Univ., 1432010 163–166
  • ShenX.L., HouY.P., Some trees are determined by their Laplacian spectra, J. Nat. Sci. Hunan Norm. Univ., 2912006 21–24 (in Chinese)
  • StanićZ., On determination of caterpillars with four terminal vertices by their Laplacian spectrum, Linear Algebra Appl., 431 2009 2035–2048
  • BuC., ZhouJ., LiH., WangW., Spectral characterizations of the corona of a cycle and two isolated vertices, Graphs Combin., 30 2014 1123–1133
  • LiuM.H., Some graphs determined by their (signless) Laplacian spectra, Czechoslovak Math. J., 621372012 1117–1134
  • LiuM.H., ShanH.Y., DasK.C., Some graphs determined by their (signless) Laplacian spectra, Linear Algebra Appl., 449 2014 154–165
  • LiuX., WangS., ZhangY., YongX., On the spectral characterization of some unicyclic graphs, Discrete Math., 311 2011 2317–2336
  • WangJ.F., BelardoF., ZhangQ., Signless Laplacian spectral characterization of line graphs of T-shape trees, Linear Multilinear Algebra, 62 2014 1529–1545
  • ZhangY., LiuX., ZhangB., YongX., The lollipop graph is determined by its Q-spectrum, Discrete Math., 309 2009 3364–3369
  • GuoG., WangG., On the (signless) Laplacian spectral characterization of the line graphs of lollipop graphs, Linear Algebra Appl., 438 2013 4595–4605
  • LiuX., ZhouS., Spectral characterizations of propeller graphs, Electron. J. Linear Algebra., 27 2014 19–38
  • WangJ.F., HuangQ.X., BelardoF., Li MarziE.M., On the spectral characterizations of 1-graphs, Discrete Math., 310 2010 1845–1855
  • WangJ.F., HuangQ.X., BelardoF., Li MarziE.M., Spectral characterizations of dumbbell graphs, Electron. J. Combin., 17 2010
  • CámaraM., HaemersW.H., Spectral characterizations of almost complete graphs, Discrete Appl. Math., 176 2014 19–23
  • HuangS., ZhouJ., BuC., Signless Laplacian spectral characterization of graphs with isolated vertices, Filomat, 30142017
  • LiuF.J., HuangQ.X., LaiH.J., Note on the spectral characterization of some cubic graphs with maximum number of triangles, Linear Algebra Appl., 438 2013 1393–1397