Let A(G) and D(G) be the adjacency matrix and the degree diagonal matrix of a graph G, respectively. For any real number α∈[0,1], Nikiforov recently defined the Aα-matrix of G as Aα(G)=αD(G)+(1−α)A(G). The graph invariant Spα(G) is the sum of the p-th power of the Aα-eigenvalues of G for 12<α<1, which has a close relation to the α-Estrada index. In this paper, we establish some bounds on Spα(G) and characterize the extremal graphs. In particular, we present some bounds on Spα(G) in terms of the degree sequences, order and size of G by using majorization techniques. Moreover, we give lower and upper bounds for Spα(G) of a bipartite graph and characterize the extremal graphs.
Citation: Zhen Lin. On the sum of powers of the Aα-eigenvalues of graphs[J]. Mathematical Modelling and Control, 2022, 2(2): 55-64. doi: 10.3934/mmc.2022007
[1] | Min Zeng, Yongxin Yuan . On the solutions of the dual matrix equation $ A^\top XA = B $. Mathematical Modelling and Control, 2023, 3(3): 210-217. doi: 10.3934/mmc.2023018 |
[2] | Qian Wang, Xue Han . Comparing the number of ideals in quadratic number fields. Mathematical Modelling and Control, 2022, 2(4): 268-271. doi: 10.3934/mmc.2022025 |
[3] | A. Refaie Ali, Rashid Jan, H. Alotaibi, Nesreen A. Yaseen . Analyticity and uniqueness of the fractional electromagnetic boundary value problem. Mathematical Modelling and Control, 2024, 4(1): 101-109. doi: 10.3934/mmc.2024009 |
[4] | Aidong Ge, Zhen Chang, Jun-e Feng . Solving interval type-2 fuzzy relation equations via semi-tensor product of interval matrices. Mathematical Modelling and Control, 2023, 3(4): 331-344. doi: 10.3934/mmc.2023027 |
[5] | Salma Al-Tuwairqi, Asma Badrah . A qualitative analysis of a model on alpha-synuclein transport and aggregation in neurons. Mathematical Modelling and Control, 2023, 3(2): 104-115. doi: 10.3934/mmc.2023010 |
[6] | M. Sathish Kumar, M. Deepa, J Kavitha, V. Sadhasivam . Existence theory of fractional order three-dimensional differential system at resonance. Mathematical Modelling and Control, 2023, 3(2): 127-138. doi: 10.3934/mmc.2023012 |
[7] | Qian Lin, Yan Zhu . Unicyclic graphs with extremal exponential Randić index. Mathematical Modelling and Control, 2021, 1(3): 164-171. doi: 10.3934/mmc.2021015 |
[8] | Jianhua Sun, Ying Li, Mingcui Zhang, Zhihong Liu, Anli Wei . A new method based on semi-tensor product of matrices for solving reduced biquaternion matrix equation $ \sum\limits_{p = 1}^l A_pXB_p = C $ and its application in color image restoration. Mathematical Modelling and Control, 2023, 3(3): 218-232. doi: 10.3934/mmc.2023019 |
[9] | Xueling Fan, Ying Li, Wenxv Ding, Jianli Zhao . $ \mathcal{H} $-representation method for solving reduced biquaternion matrix equation. Mathematical Modelling and Control, 2022, 2(2): 65-74. doi: 10.3934/mmc.2022008 |
[10] | Xiaomeng Wei, Haitao Li, Guodong Zhao . Kronecker product decomposition of Boolean matrix with application to topological structure analysis of Boolean networks. Mathematical Modelling and Control, 2023, 3(4): 306-315. doi: 10.3934/mmc.2023025 |
Let A(G) and D(G) be the adjacency matrix and the degree diagonal matrix of a graph G, respectively. For any real number α∈[0,1], Nikiforov recently defined the Aα-matrix of G as Aα(G)=αD(G)+(1−α)A(G). The graph invariant Spα(G) is the sum of the p-th power of the Aα-eigenvalues of G for 12<α<1, which has a close relation to the α-Estrada index. In this paper, we establish some bounds on Spα(G) and characterize the extremal graphs. In particular, we present some bounds on Spα(G) in terms of the degree sequences, order and size of G by using majorization techniques. Moreover, we give lower and upper bounds for Spα(G) of a bipartite graph and characterize the extremal graphs.
Let G be a simple finite undirected connected graph with vertex set V(G) and edge set E(G), where |V(G)| is the order and |E(G)| is the size of G. Let A(G) and D(G) be the adjacency matrix and the degree diagonal matrix of a graph G, respectively. Then L(G)=D(G)−A(G), Q(G)=D(G)+A(G) and L(G)=D−12(G)L(G)D−12(G) are called the Laplacian matrix, the signless Laplacian matrix and the normalized Laplacian matrix of the graph G, respectively.
The investigation on the sum of the p-th power of the eigenvalues of graphs is a topic of interest in Mathematical Chemistry. Based on the mathematical methods, scholars get many bounds for the sum of the p-th power of the eigenvalues of graphs. For a non-zero real number p, Zhou [1] introduced the sum of the p-th power of the non-zero Laplacian eigenvalues of G, denoted by SpL(G). Since SpL(G) has close relation with the Laplacian-energy-like invariant [2], the Laplacian Estrada index [3] and the Kirhhoff index [4], there are considerable results regarding SpL(G) in the literature. For related results, one may refer to [1,5,6,7,8,9] and references therein. For a non-zero real number p, M. Liu and B. Liu [10] defined SpQ(G) as the sum of the p-th power of the non-zero signless Laplacian eigenvalues of G, which has close relation with the incidence energy [11] and the signless Laplacian Estrada index [12]. For details on SpQ(G), see the papers [13,14] and the references cited therein. Moreover, Akbari et al. [15,16] compared between SpL(G) and SpQ(G) when the parameter p takes different values. For a non-zero real number p, Ş.B. Bozkurt and D. Bozkurt [17] defined SpL(G) as the sum of the p-th power of the normalized Laplacian eigenvalues of G, which has close relation with the degree-Kirchhoff index [18] and the general Randić index [19]. For related results, one may refer to [20,21].
For any real number α∈[0,1], Nikiforov [22] defined the Aα-matrix of G as
Aα(G)=αD(G)+(1−α)A(G). |
It is easy to see that Aα(G) is the adjacency matrix A(G) if α=0, and Aα(G) is essentially equivalent to signless Laplacian matrix Q(G) if α=12. The new matrix Aα(G) not only can underpin a unified theory of A(G) and Q(G), but it also brings many new interesting problems, see for example [22,23,24,25]. In particular, Aα(G) is a positive definite matrix for 12<α<1, which is a hitherto uncharted territory of worth our investigation and exploration, see [22]. Moreover, X. Liu and S. Liu [26] found that Aα-eigenvalues (especially, 12<α<1) are much more efficient than A-eigenvalues and Q-eigenvalues when we use them to distinguish graphs, by enumerating the Aα-characteristic polynomials for all graphs on at most ten vertices. The Aα-matrix has been an interesting topic in mathematical literature and has been studied extensively, see for example [22,23,27,28,29,30] and references therein.
Let λ1(Aα(G))≥λ2(Aα(G))≥⋯≥λn(Aα(G)) be the Aα-eigenvalues of a graph G of order n. Motivated by the above work, we define Spα(G) as the sum of the p-th power of the Aα-eigenvalues of G, that is,
Spα(G)=n∑i=1λpi(Aα(G)), |
where 12<α<1 and p is a real number. Spα(G) can be regard as a generalization of SpQ(G) due to the fact that our results are correct for the sum of the p-th power of the non-zero A12-eigenvalues of G. By using the Maclaurin development, we have
Eα(G)=n∑i=1eλi(Aα(G))=∞∑p=0Spα(G)p!, |
where p is an integer and Eα(G) is called the α-Estrada index defined by Cardoso et al. [31]. Thus the bound for Spα(G) can be naturally converted to the bound of the α-Estrada index. In addition, we find that Spα(G) is connected with the first general Zagreb index, which is a useful topological index and has important applications in chemistry.
The primary purpose of this paper is to establish the bounds of Spα(G). The cases p=0 and p=1 are trivial as S0α(G)=n and S1α(G)=2αm, where m is the size of G. We will not consider both cases in the following results. The rest of the paper is organized as follows. In Section 2, we recall some useful notions and lemmas used further. In Section 3, some bounds on the Spα(G) are presented. In Sections 4 and 5, several bounds for Spα(G) related to degree sequences, order and size are given through majorization techniques. In Section 6, lower and upper bounds for Spα(G) of a bipartite graph G are obtained, and the extremal graphs characterized.
Let G−e denote the graph that arises from G by deleting the edge e∈E(G). A connected graph is called a c-cyclic graph if it contains n vertices and n+c−1 edges. For vi∈V(G), dG(vi)=di(G) denotes the degree of vertex vi in G. The minimum and the maximum degree of G are denoted by δ=δ(G) and Δ=Δ(G), respectively. A pendant vertex is a vertex of degree one and a quasi-pendant vertex is a vertex adjacent to a pendant vertex. Li and Zheng [32] defined the first general Zagreb index of G as Zp=Zp(G)=∑v∈V(G)dp(v), where p is an arbitrary real number except 0 and 1. A subset I of V(G) is called an independent set of a graph G if no two vertices in I are adjacent in G. Given a graph G, the independence number θ(G) of G is the numbers of vertices of the largest independent set. Denote by Kn, Ka,b and ¯G the complete graph, the complete bipartite graph and the complement of a graph G, respectively. The join G1∨G2 of two vertex-disjoint graphs G1 and G2 is the graph formed from the union of G1 and G2 by joining each vertex of G1 to each vertex of G2.
Lemma 2.1. ([33]) Let G be a graph with n vertices. If e∈E(G) and 12≤α≤1, thenλi(Aα(G))≥λi(Aα(G−e))for 1≤i≤n.
Lemma 2.2. ([34,35]) Let G be a graph of order n and size m. Then
Z2(G)≥4m2n+12(Δ−δ)2 |
with equality if and only if G has the property d2=d3=⋯=dn−1=Δ+δ2, which includes also the regular graphs.
Lemma 2.3. ([36]) Let s1,s2,…,sn be the singular values of a matrix M=(mij)∈Mn. Then
n∑j=1spj≤n∑i,j=1|mij|pfor0<p≤2,n∑j=1spj≥n∑i,j=1|mij|pforp≥2. |
Lemma 2.4. ([37]) For c-cyclic graphs with n vertices, the minimal degree sequences with respect to the majorization order are given by (2,2,…,2,1,1), in case c=0 and n>2, (2,2,…,2), in case c=1 and n>2, (3,3,…,3⏟2c−2,2,2,…,2), in case 2≤c≤6 and n>2c−2.
Lemma 2.5. ([38]) Let a(G), b(G) and mG(α) be the number of pendant vertices, quasi-pendant vertices of G and the multiplicity of α as an eigenvalue of Aα(G), respectively. ThenmG(α)≥a(G)−b(G)with equality if each internal vertex is a quasi-pendant vertex.
Lemma 2.6. ([39]) Let G be a graph of order n and size m. If α∈(12,1), then
λ1(Aα(G))≤2mn−1(1−α)+αn−1, |
the equality holds if and only if G≅Kn.
Theorem 3.1. Let 12<α<1, G be a connected graph with n vertices and e∈E(G).
(i) If p>0 and p≠1, thenSpα(G−e)<Spα(G).
(ii) If p<0, thenSpα(G−e)>Spα(G).
Proof. By Perron-Frobenius Theorem, we have λ1(Aα(G))>λ1(Aα(G−e)). By Lemma 2.1, the result follows.
Corollary 3.1. Let 12<α<1, G be a connected graph of order n.
(i) If p>0 and p≠1, then
Spα(G)≤(n−1)p+(n−1)(αn−1)p |
with equality if and only if G≅Kn.
(ii) If p<0, then
Spα(G)≥(n−1)p+(n−1)(αn−1)p |
with equality if and only if G≅Kn.
Proof. From Proposition 36 in [22], it follows that λ1(Aα(Kn))=n−1 and λi(Aα(Kn))=αn−1 for 2≤i≤n. By Theorem 3.1, we have the proof.
Corollary 3.2. Let 12<α<1, G be a connected graph of order n with independence number θ.
(i) If p>0 and p≠1, then
Spα(G)≤(n−θ−1)(αn−1)p+(θ−1)(n−θ)pαp+xp1+xp2, |
where x1 and x2 are the roots of the equation
x2−(αn+n−θ−1)x+αθ+αn2−αn−αθ2−θn+θ2=0 |
and equality holds if and only if G≅¯Kθ∨Kn−θ.
(ii) If p<0, then
Spα(G)≥(n−θ−1)(αn−1)p+(θ−1)(n−θ)pαp+xp1+xp2, |
where x1 and x2 are the roots of the equation
x2−(αn+n−θ−1)x+αθ+αn2−αn−αθ2−θn+θ2=0 |
and equality holds if and only if G≅¯Kθ∨Kn−θ.
Proof. Let ϕα(G,x) be the characteristic polynomial of Aα(G). By direct computation, we have
ϕα(¯Kθ∨Kn−θ,x)=(x−αn+1)n−θ−1[x−(n−θ)α]θ−1[x2−(αn+n−θ−1)x+αθ+αn2−αn−αθ2−θn+θ2]. |
Thus
Spα(¯Kθ∨Kn−θ)=(n−θ−1)(αn−1)p+(θ−1)(n−θ)pαp+xp1+xp2, |
where x1 and x2 are the roots of the equation x2−(αn+n−θ−1)x+αθ+αn2−αn−αθ2−θn+θ2=0. By Theorem 3.1, we have the proof.
Theorem 3.2. Let G be a connected graph of order n and size m. If 12<α<1 and p≠0 and p≠1, then
Spα(G)≥(2mn)p+(n−1)(ndet(Aα(G))2m)pn−1 | (3.1) |
with equality if and only if G≅Kn.
Proof. By the arithmetic-geometric mean inequality, we have
Spα(G)=λp1(Aα(G))+n∑i=2λpi(Aα(G))≥λp1(Aα(G))+(n−1)(n∏i=2λi(Aα(G)))pn−1=λp1(Aα(G))+(n−1)(det(Aα(G))λ1(Aα(G)))pn−1. |
Let h(x)=xp+(n−1)(det(Aα(G))x)pn−1. Then h′(x)=p(xp−1−det(Aα(G))pn−1x−pn−1−1). It is easy to see that h(x) is increasing on [det(Aα(G))1n,+∞) whether p>0 or p<0. From Corollary 19 in [22], it follows that
λ1(Aα(G))≥2mn>2αmn=n∑i=1λi(Aα(G))n≥(n∏i=1λi(Aα(G)))1n=det(Aα(G))1n. |
Thus
Spα(G)≥h(λ1(Aα(G)))≥h(2mn)=(2mn)p+(n−1)(ndet(Aα(G))2m)pn−1 |
with equality if and only if λ1(Aα(G))=2mn and λ2(Aα(G))=⋯=λn(Aα(G)). From Corollary 33 in [22], the diameter of G is 1. Thus, G≅Kn. Conversely, if G≅Kn, then λ1(Aα(G))=n−1, and λi(Aα(G))=αn−1 for 2≤i≤n. It is easy to check that equality holds in (3.1). This completes the proof.
Theorem 3.3. Let 12<α<1, G be a connected graph of order n and size m.
(i) If p<0 or p>1, then
Spα(G)≥(Z2n)p2+1(n−1)p−1(2αm−√Z2n)p | (3.2) |
with equality if and only if G≅Kn.
(ii) If 0<p<1, then
Spα(G)≤(Z2n)p2+1(n−1)p−1(2αm−√Z2n)p | (3.3) |
with equality if and only if G≅Kn.
Proof. Since p<0 or p>1, we know that f(x)=xp is a strictly convex function. By Jensen's inequality, we have
(n∑i=21n−1λi(Aα(G)))p≤n∑i=21n−1λpi(Aα(G)), |
that is,
n∑i=2λpi(Aα(G))≥1(n−1)p−1(2αm−λ1(Aα(G)))p. |
Thus
Spα(G)=λp1(Aα(G))+n∑i=2λpi(Aα(G))≥λp1(Aα(G))+1(n−1)p−1(2αm−λ1(Aα(G)))p. |
Let g(x)=xp+1(n−1)p−1(2αm−x)p. Then g′(x)=p(xp−1−(2αm−x)p−1(n−1)p−1)≥0 for x≥2αmn. Hence g(x) is increasing on [2αmn,+∞). From Lemma 2.2 and Corollary 19 in [22], it follows that λ1(Aα(G))≥√Z2n≥2αmn. Thus
Spα(G)≥g(λ1(Aα(G)))≥g(√Z2n)=(Z2n)p2+1(n−1)p−1(2αm−√Z2n)p |
with equality if and only if λ1(Aα(G))=√Z2n and λ2(Aα(G))=⋯=λn(Aα(G)). From Corollary 33 in [22], the diameter of G is 1. Thus, G≅Kn. Conversely, if G≅Kn, then λ1(Aα(G))=n−1, and λi(Aα(G))=αn−1 for 2≤i≤n. It is easy to check that equality holds in (3.2).
Now suppose that 0<p<1. Then
(n∑i=21n−1λi(Aα(G)))p≥n∑i=21n−1λpi(Aα(G)), |
with equality if and only if λ2(Aα(G))=⋯=λn(Aα(G)), and g(x) is decreasing on [2αmn,+∞). By similar arguments as above, the second part of the theorem follows.
Combining the above arguments, we have the proof.
By Lemma 2.2 and Theorem 3.3, we have
Corollary 3.3. Let 12<α<1, G be a connected graph of order n and size m.
(i) If p<0 or p>1, then
Spα(G)≥(4m2n2+12n(Δ−δ)2)p2+1(n−1)p−1(2αm−√4m2n2+12n(Δ−δ)2)p |
with equality if and only if G≅Kn.
(ii) If 0<p<1, then
Spα(G)≤(4m2n2+12n(Δ−δ)2)p2+1(n−1)p−1(2αm−√4m2n2+12n(Δ−δ)2)p |
with equality if and only if G≅Kn.
Theorem 3.4. Let 12<α<1, G be a connected graph of order n and size m.
(i) If 0<p≤2, thenSpα(G)≤αpZp+2m(1−α)p.
(ii) If p>2, thenSpα(G)≥αpZp+2m(1−α)p.
Proof. Since Aα(G) is a real symmetric and positive definite matrix for 12<α<1, the singular values of Aα(G) are equal to the eigenvalues of Aα(G). By Lemma 2.3, we have the proof.
Suppose x=(x1,x2,…,xn) and y=(y1,y2,…,yn) are two non-increasing sequences of real numbers, we say x is majorized by y, denoted by x⪯y, if n∑i=1xi=n∑i=1yi and j∑i=1xi≤j∑i=1yi for j=1,2,…,n−1. For a real-valued function f defined on a set in Rn, if f(x)≤f(y) whenever x⪯y but x≠y, then f is said to be Schur-convex.
Theorem 4.1. Let 12<α<1, G be a connected graph of order n and size m with the degree sequence d1≥d2≥⋯≥dn.
(i) If p<0 or p>1, then
(2αm)pnp−1≤αpZp(G)≤Spα(G). |
(ii) If 0<p<1, then
Spα(G)≤αpZp(G)≤(2αm)pnp−1. |
Proof. Let x=2αmn(1,…,1), y=(αd1,…,αdn) and z=(λ1(Aα(G)),…,λn(Aα(G))). It is well known that the spectrum of any symmetric, positive semi-definite matrix majorizes its main diagonal [40], hence x⪯y⪯z. Since p<0 or p>1, f(x)=xp is a convex function. From [41], we know that if the real-valued function f defined on an interval in R is a convex then n∑i=1f(xi) is Schur-convex. Thus (2αm)pnp−1≤n∑i=1αpdpi≤Spα(G).
If 0<p<1, then g(x)=−xp is a convex function. By similar arguments as above, the second part of the theorem follows.
By Lemma 2.4 and Theorem 4.1, we have
Corollary 4.1. Let 12<α<1, 0≤c≤6 and G be a c-cyclic graph with n vertices.
(i) If p<0 or p>1, c=0 and n>2, then
Spα(G)≥(n−2)(2α)p+2αp. |
If p<0 or p>1, c=1 and n>2, then
Spα(G)≥n(2α)p. |
If p<0 or p>1, 2≤c≤6 and n>2c−2, then
Spα(G)≥αp((2c−2)3p+(n−2c+2)2p). |
(ii) If 0<p<1, c=0 and n>2, then
Spα(G)≤(n−2)(2α)p+2αp. |
If 0<p<1, c=1 and n>2, thenSpα(G)≤n(2α)p.
If 0<p<1, 2≤c≤6 and n>2c−2, then
Spα(G)≤αp((2c−2)3p+(n−2c+2)2p). |
Theorem 4.2. Let 12<α<1, G be a connected graph of order n and size m with the degree sequence d1≥d2≥⋯≥dn=δ.
(i) If p<0 or p>1, then
Spα(G)≤n−1∑i=1(αdi+(1−α)(n−i))p+12p(2αδ−(1−α)n(n−1))p. |
(ii) If 0<p<1, then
Spα(G)≥n−1∑i=1(αdi+(1−α)(n−i))p+12p(2αδ−(1−α)n(n−1))p. |
Proof. Let x=(λ1(Aα(G)),λ2(Aα(G)),…,λn(Aα(G))) and y=(αd1+(1−α)(n−1),αd2+(1−α)(n−2),…,αdn−1+(1−α),2αm−α(2m−δ)−(1−α)n(n−1)2). From Theorem 3.1 in [28], it follows that λi(Aα(G))≤αdi+(1−α)(n−i) for 1≤i≤n. Thus x⪯y. Similar to the method used in Theorem 4.1, we have the proof.
Theorem 4.3. Let 12<α<1, G be a connected graph of order n and size m with the degree sequence Δ=d1≥d2≥⋯≥dn.
(i) If p<0 or p>1 and d2≥d3≥⋯≥dk≥αn−1, then
Spα(G)≤Δp+(k−1)(αn−1)p |
+[2αm−Δ−(k−1)(αn−1)]p, | (4.1) |
where 2≤k≤n.If p<0 or p>1 and d2≤αn−1, then
Spα(G)≤k∑i=1dpi+(2αm−k∑i=1di)p, |
where 2≤k≤n.
(ii) If 0<p<1 and d2≥d3≥⋯≥dk≥αn−1, then
Spα(G)≥Δp+(k−1)(αn−1)p |
+[2αm−Δ−(k−1)(αn−1)]p, | (4.2) |
where 2≤k≤n.If 0<p<1 and d2≤αn−1, then
Spα(G)≥k∑i=1dpi+(2αm−k∑i=1di)p, |
where 2≤k≤n.
Proof. Let x=(λ1(Aα(G)),λ2(Aα(G)),…,λn(Aα(G))) and y=(d1,αn−1,…,αn−1,2αm−d1−(k−1)(αn−1),0,…,0). From Proposition 10 in [22], it follows that λ1(Aα(G))≤d1. By Lemma 2.1, we have λi(Aα(G))≤λi(Aα(Kn))=αn−1 for 2≤i≤n. Thus x⪯y. Similar to the method used in Theorem 4.1, we have the proof.
Remark 4.1. It is easy to see that the equality in (4.1) and (4.2) holds if G≅Kn.
Theorem 4.4. Let 12<α<1, G be a connected graph of order n and size m with the degree sequence Δ=d1≥d2≥⋯≥dn.
(i) If p<0 or p>1 and n−1>d1≥d2≥d3≥⋯≥dk≥α(n−2), then
Spα(G)≤Δp+αp(k−1)(n−2)p+(2αm−Δ−α(k−1)(n−2))p, |
where 2≤k≤n.If p<0 or p>1, d1<n−1 and d2≤α(n−2), then
Spα(G)≤k∑i=1dpi+(2αm−k∑i=1di)p, |
where 2≤k≤n.
(ii) If 0<p<1 and
n−1>d1≥d2≥d3≥⋯≥dk≥α(n−2), |
then
Spα(G)≥Δp+αp(k−1)(n−2)p+(2αm−Δ−α(k−1)(n−2))p, |
where 2≤k≤n.If 0<p<1, d1<n−1 and d2≤α(n−2), then
Spα(G)≥k∑i=1dpi+(2αm−k∑i=1di)p, |
where 2≤k≤n.
Proof. Let x=(λ1(Aα(G)),λ2(Aα(G)),…,λn(Aα(G))) and y=(d1,α(n−2),…,α(n−2),2αm−d1−α(k−1)(n−2),0,…,0). From Proposition 10 in [22] and Theorem 3.1 in [27], it follows that λ1(Aα(G))≤d1 and λi(Aα(G))≤α(n−2) for 2≤i≤n. Thus x⪯y. Similar to the method used in Theorem 4.1, we have the proof.
Theorem 4.5. Let 12<α<1, G be a connected graph of order n and size m with the degree sequence Δ=d1≥d2≥⋯≥dn.
(i) If p<0 or p>1 and mG(λj) is the multiplicity of λj as an eigenvalue of Aα(G), then
Spα(G)≤Δp+k∑i=2dpi+mG(λj)λpj+(2αm−Δ−k∑i=2di−mG(λj)λj)p, |
where 2≤k≤j≤n.
(ii) If 0<p<1 and mG(λj) is the multiplicity of λj as an eigenvalue of Aα(G), then
Spα(G)≥Δp+k∑i=2dpi+mG(λj)λpj+(2αm−Δ−k∑i=2di−mG(λj)λj)p, |
where 2≤k≤j≤n.
Proof. Let x=(λ1(Aα(G)),…,λn(Aα(G))) and y=(d1,d2,…,dk,λj,…,λj,2αm−d1−k∑i=2dk−mG(λj)λj,0,…,0). From Proposition 10 in [22], it follows that λi(Aα(G))≤di for 1≤i≤n. Then x⪯y. Similar to the method used in Theorem 4.1, we have the proof.
By Lemma 2.5 and Theorem 4.5, we have
Corollary 4.2. Let 12<α<1, G be a connected graph of order n and size m with the degree sequence Δ=d1≥d2≥⋯≥dn, and let a and b be the number of pendant vertices and quasi-pendant vertices of G, respectively.
(i) If p<0 or p>1 and a−b≥1, then
Spα(G)≤Δp+n−a+b−1∑i=2dpi+(a−b)αp+(2αm−Δ−n−a+b−1∑i=2di−(a−b)α)p. |
(ii) If 0<p<1 and a−b≥1, then
Spα(G)≥Δp+n−a+b−1∑i=2dpi+(a−b)αp+(2αm−Δ−n−a+b−1∑i=2di−(a−b)α)p. |
Theorem 5.1. Let 12<α<1, G be a connected graph of order n and size m.
(i) If p<0 or p>1 and there is c such that λ1(Aα(G))≥c>0, then
Spα(G)≥cp+(2αm−c)p(n−1)p−1. |
(ii) If 0<p<1 and there is c such that λ1(Aα(G))≥c>0, then
Spα(G)≤cp+(2αm−c)p(n−1)p−1. |
Proof. Let x=(c,2αm−cn−1,…,2αm−cn−1) and
y=(λ1(Aα(G)),λ2(Aα(G)),…,λn(Aα(G))). |
Since λ1(Aα(G))≥c>0, we have x⪯y. Similar to the method used in Theorem 4.1, we have the proof.
Corollary 5.1. Let 12<α<1, G be a connected graph of order n and size m.
(i) If p<0 or p>1, then
Spα(G)≥1αp(α2Δ+(1−α)2)p |
+(2α2m−α2Δ−(1−α)2)pαp(n−1)p−1. | (5.1) |
(ii) If 0<p<1, then
Spα(G)≤1αp(α2Δ+(1−α)2)p |
+(2α2m−α2Δ−(1−α)2)pαp(n−1)p−1. | (5.2) |
Proof. From Corollary 13 in [22], it follows that
λ1(Aα(G))≥αΔ+(1−α)2α. |
By Theorem 5.1, we have the proof.
Corollary 5.2. Let 12<α<1, G be a connected graph of order n and size m.
(i) If p<0 or p>1, then
Spα(G)≥(2mn)p(1+(αn−1)p(n−1)p−1). |
(ii) If 0<p<1, then
Spα(G)≤(2mn)p(1+(αn−1)p(n−1)p−1). |
Proof. From Corollary 19 in [22], it follows that
λ1(Aα(G))≥2mn. |
By Theorem 5.1, we have the proof.
Remark 5.1. It is easy to see that the equality in Corollary 5.2 holds if G≅Kn.
Corollary 5.3. Let 12<α<1, G be a connected graph of order n and size m with chromatic number χ.
(i) If p<0 or p>1, then
Spα(G)≥(χ−1)p+(2αm−χ+1)p(n−1)p−1. | (5.3) |
(ii) If 0<p<1, then
Spα(G)≤(χ−1)p+(2αm−χ+1)p(n−1)p−1. | (5.4) |
The equality holds in (5.3) and (5.4) if and only if G≅Kn.
Proof. It is well known that λ1(A(G))≥χ−1 with equality if and only if G is a complete graph or an odd cycle, see [42]. From Proposition 18 in [22], it follows that
λ1(Aα(G))≥λ1(A(G))≥χ−1 |
with equality if and only if G is a complete graph or an odd cycle. By Theorem 5.1, we have the proof.
Theorem 5.2. Let 12<α<1, G be a connected graph of order n and size m.
(i) If p<0 or p>1, then
Spα(G)≤(2(1−α)mn−1+αn−1)p+(n−2)(αn−1)p |
+(2αm−2(1−α)mn−1−(n−1)(αn−1))p. | (5.5) |
(ii) If 0<p<1, then
Spα(G)≥(2(1−α)mn−1+αn−1)p+(n−2)(αn−1)p |
+(2αm−2(1−α)mn−1−(n−1)(αn−1))p. | (5.6) |
The equality holds in (5.5) and (5.6) if and only if G≅Kn.
Proof. Let x=(λ1(Aα(G)),…,λn(Aα(G))) and y=(2mn−1(1−α)+αn−1,αn−1,…,αn−1,2αm−2mn−1(1−α)−(n−1)(αn−1)). By Lemmas 2.1 and 2.6, we have
λ1(Aα(G))≤2mn−1(1−α)+αn−1 |
and
λi(Aα(G))≤λi(Aα(Kn))=αn−1 |
for 2≤i≤n. Thus x⪯y. Similar to the method used in Theorem 4.1, we have the proof.
Theorem 6.1. Let 12<α<1, G be a connected bipartite graph with n vertices.
(i) If p>1, then Spα(G)≤Spα(K1,n−1) with equality if and only if G≅K1,n−1.
(ii) If 0<p<1, then Spα(G)≤Spα(K⌈n/2⌉,⌊n/2⌋) with equality if and only if G≅K⌈n/2⌉,⌊n/2⌋.
(iii) If p<0, then Spα(G)≥Spα(K⌈n/2⌉,⌊n/2⌋) with equality if and only if G≅K⌈n/2⌉,⌊n/2⌋.
Proof. Let G=(X,Y) be a connected bipartite graph on n vertices and suppose that |X|=a, |Y|=b and a≥b≥1, where a+b=n. From Proposition 38 in [22], it follows that
Spα(Ka,b)=12p(αn+√α2n2+4a(n−a)(1−2α))p+αpap+αp(n−a)p+12p(αn−√α2n2+4a(n−a)(1−2α))p. |
Let
f(x)=12p(αn+√α2n2+4x(n−x)(1−2α))p+αpxp+αp(n−x)p+12p(αn−√α2n2+4x(n−x)(1−2α))p. |
Then
f′(x)=p(1−2α)(n−2x)2p−1√α2n2+4x(n−x)(1−2α)[(αn+√α2n2+4x(n−x)(1−2α))p−1−(αn−√α2n2+4x(n−x)(1−2α))p−1]+pαp(xp−1−(n−x)p−1). |
If p>1, then f(x) is decreasing for 1≤x≤n2 and increasing for n2≤x≤n−1. Hence f(n/2)≤f(x)≤f(1). By Theorem 3.1, we have Spα(G)<Spα(Ka,b)≤Spα(K1,n−1) for p>0, p≠1 and G≠Ka,b.
If 0<p<1, then f(x) is increasing for 1≤x≤n2 and decreasing for n2≤x≤n−1. Hence f(1)≤f(x)≤f(n/2). By Theorem 3.1, we have Spα(G)<Spα(Ka,b)≤Spα(K⌈n/2⌉,⌊n/2⌋) for p>0, p≠1 and G≠Ka,b.
If p<0, then f(x) is decreasing for 1≤x≤n2 and increasing for n2≤x≤n−1. Hence f(n/2)≤f(x)≤f(1). By Theorem 3.1, we have Spα(G)>Spα(Ka,b)≥Spα(K⌈n/2⌉,⌊n/2⌋) for p<0 and G≠Ka,b.
Combining the above arguments, we have the proof.
The author is grateful to the anonymous referee for careful reading and valuable comments which result in an improvement of the original manuscript. This work was supported by the National Natural Science Foundation of China (No. 12071411).
The authors declared that they have no conflicts of interest to this work.
[1] |
B. Zhou, On sum of powers of the Laplacian eigenvalues of graphs, Linear Algebra Appl., 429 (2008), 2239–2246. https://doi.org/10.1016/j.laa.2008.06.023 doi: 10.1016/j.laa.2008.06.023
![]() |
[2] | J. Liu, B. Liu, A Laplacian-energy-like invariant of a graph, MATCH Commun. Math. Comput. Chem., 59 (2008), 355–372. |
[3] | G. H. Fath-Tabar, A. R. Ashrafi, I. Gutman, Note on Estrada and L-Estrada indices of graphs, Bull. Cl. Sci. Math. Nat. Sci. Math. No., 34 (2009), 1–16. |
[4] |
I. Gutman, B. Mohar, The quasi-Wiener and the Kirchhoff indices coincide, J. Chem. Inf. Comput. Sci., 36 (1996), 982–985. https://doi.org/10.1021/ci960007t doi: 10.1021/ci960007t
![]() |
[5] |
X. Chen, K. C. Das, Characterization of extremal graphs from Laplacian eigenvalues and the sum of powers of the Laplacian eigenvalues of graphs, Discrete Math., 338 (2015), 1252–1263. https://doi.org/10.1016/j.disc.2015.02.006 doi: 10.1016/j.disc.2015.02.006
![]() |
[6] |
K. C. Das, K. Xu, M. Liu, On sum of powers of the Laplacian eigenvalues of graphs, Linear Algebra Appl., 439 (2013), 3561–3575. https://doi.org/10.1016/j.laa.2013.09.036 doi: 10.1016/j.laa.2013.09.036
![]() |
[7] |
M. Liu, B. Liu, A note on sum of powers of the Laplacian eigenvalues of graphs, Appl. Math. Lett., 24 (2011), 249–252. https://doi.org/10.1016/j.aml.2010.09.013 doi: 10.1016/j.aml.2010.09.013
![]() |
[8] |
G. Tian, T. Huang, B. Zhou, A note on sum of powers of the Laplacian eigenvalues of bipartite graphs, Linear Algebra Appl., 430 (2009), 2503–2510. https://doi.org/10.1016/j.laa.2008.12.030 doi: 10.1016/j.laa.2008.12.030
![]() |
[9] |
B. Zhou, A. Ilić, On the sum of powers of Laplacian eigenvalues of bipartite graphs, Czechoslovak Math. J., 60 (2010), 1161–1169. https://doi.org/10.1007/s10587-010-0081-8 doi: 10.1007/s10587-010-0081-8
![]() |
[10] | M. Liu, B. Liu, On sum of powers of the signless Laplacian eigenvalues of graphs, Hacet. J. Math. Stat., 41 (2012), 527–536. |
[11] | M. R. Jooyandeh, D. Kiani, M. Mirzakhah, Incidence energy of a graph, MATCH Commun. Math. Comput. Chem., 62 (2009), 561–572. |
[12] | S. K. Ayyaswamy, S. Balachandran, Y. B. Venkatakrishnan, I. Gutman, Signless Laplacian Estrada index, MATCH Commun. Math. Comput. Chem., 66 (2011), 785–794. |
[13] |
F. Ashraf, On two conjectures on sum of the powers of signless Laplacian eigenvalues of a graph, Linear Multilinear Algebra, 64 (2016), 1314–1320. https://doi.org/10.1080/03081087.2015.1083525 doi: 10.1080/03081087.2015.1083525
![]() |
[14] | L. You, J. Yang, Notes on the sum of powers of the signless Laplacian eigenvalues of graphs, Ars Combin., 117 (2014), 85–94. |
[15] |
S. Akbari, E. Ghorbani, J. H. Koolen, M. R. Oboudi, A relation between the Laplacian and signless Laplacian eigenvalues of a graph, J. Algebraic Combin., 32 (2010), 459–464. https://doi.org/10.1007/s10801-010-0225-9 doi: 10.1007/s10801-010-0225-9
![]() |
[16] |
S. Akbari, E. Ghorbani, J. H. Koolen, M. R. Oboudi, On sum of powers of the Laplacian and signless Laplacian eigenvalues of graphs, Electron. J. Combin., 17 (2010), R115. https://doi.org/10.37236/387 doi: 10.37236/387
![]() |
[17] | Ş. B. Bozkurt, D. Bozkurt, On the sum of powers of normalized Laplacian eigenvalues of graphs, MATCH Commun. Math. Comput. Chem., 68 (2012), 917–930. |
[18] |
H. Chen, F. Zhang, Resistance distance and the normalized Laplacian spectrum, Discrete Appl. Math., 155 (2007), 654–661. https://doi.org/10.1016/j.dam.2006.09.008 doi: 10.1016/j.dam.2006.09.008
![]() |
[19] | B. Bollobás, P. Erdös, Graphs of extremal weights, Ars Combin., 50 (1998), 225–233. |
[20] |
G. P. Clemente, A. Cornaro, New bounds for the sum of powers of normalized Laplacian eigenvalues of graphs, Ars Math. Contemp., 11 (2016), 403–413. https://doi.org/10.26493/1855-3974.845.1b6 doi: 10.26493/1855-3974.845.1b6
![]() |
[21] |
J. Li, J. Guo, W. C. Shiu, Ş. B. B. Altındaǧ, D. Bozkurt, Bounding the sum of powers of normalized Laplacian eigenvalues of a graph, Appl. Math. Comput., 324 (2018), 82–92. https://doi.org/10.1016/j.amc.2017.12.003 doi: 10.1016/j.amc.2017.12.003
![]() |
[22] |
V. Nikiforov, Merging the A- and Q-spectral theories, Appl. Anal. Discrete Math., 11 (2017), 81–107. https://doi.org/10.2298/AADM1701081N doi: 10.2298/AADM1701081N
![]() |
[23] |
S. Liu, K. C. Das, S. Sun, J. Shu, On the least eigenvalue of Aα-matrix of graphs, Linear Algebra Appl., 586 (2020), 347–376. https://doi.org/10.1016/j.laa.2019.10.025 doi: 10.1016/j.laa.2019.10.025
![]() |
[24] |
H. Lin, X. Liu, J. Xue, Graphs determined by their Aα-spectra, Discrete Math., 342 (2019), 441–450. https://doi.org/10.1016/j.disc.2018.10.006 doi: 10.1016/j.disc.2018.10.006
![]() |
[25] |
V. Nikiforov, O. Rojo, A note on the positive semidefiniteness of Aα(G), Linear Algebra Appl., 519 (2017), 156–163. https://doi.org/10.1016/j.laa.2016.12.042 doi: 10.1016/j.laa.2016.12.042
![]() |
[26] |
X. Liu, S. Liu, On the Aα-characteristic polynomial of a graph, Linear Algebra Appl., 546 (2018), 274–288. https://doi.org/10.1016/j.laa.2018.02.014 doi: 10.1016/j.laa.2018.02.014
![]() |
[27] |
Y. Chen, D. Li, J. Meng, On the second largest Aα-eigenvalues of graphs, Linear Algebra Appl., 580 (2019), 343–358. https://doi.org/10.1016/j.laa.2019.06.027 doi: 10.1016/j.laa.2019.06.027
![]() |
[28] |
S. Liu, K.C. Das, J. Shu, On the eigenvalues of Aα-matrix of graphs, Discrete Math., 343 (2020), 111917. https://doi.org/10.1016/j.disc.2020.111917 doi: 10.1016/j.disc.2020.111917
![]() |
[29] |
L. Wang, X. Fang, X. Geng, F. Tian, On the multiplicity of an arbitrary Aα-eigenvalue of a connected graph, Linear Algebra Appl., 589 (2020), 28–38. https://doi.org/10.1016/j.laa.2019.12.021 doi: 10.1016/j.laa.2019.12.021
![]() |
[30] |
S. Wang, D. Wong, F. Tian, Bounds for the largest and the smallest Aα eigenvalues of a graph in terms of vertex degrees, Linear Algebra Appl., 590 (2020), 210–223. https://doi.org/10.1016/j.laa.2019.12.039 doi: 10.1016/j.laa.2019.12.039
![]() |
[31] |
D. M. Cardoso, G. Pastén, O. Rojo, Graphs with clusters perturbed by regular graphs−Aα-spectrum and applications, Discuss. Math. Graph Theory, 40 (2020), 451–466. https://doi.org/10.7151/dmgt.2284 doi: 10.7151/dmgt.2284
![]() |
[32] | X. Li, J. Zheng, A unified approach to the extremal trees for different indices, MATCH Commun. Math. Comput. Chem., 54 (2005), 195–208. |
[33] |
H. Lin, J. Xue, J. Shu, On the Aα-spectra of graphs, Linear Algebra Appl., 556 (2018), 210–219. https://doi.org/10.1016/j.laa.2018.07.003 doi: 10.1016/j.laa.2018.07.003
![]() |
[34] | B. Borovićanin, K. C. Das, B. Furtula, I. Gutman, Bounds for Zagreb indices, MATCH Commun. Math. Comput. Chem., 78 (2017), 17–100. |
[35] |
T. Mansour, M. A. Rostami, E. Suresh, G. B. A. Xavier, New sharp lower bounds for the first Zagreb index, Appl. Math.Inform. Mech., 8 (2016), 11–19. https://doi.org/10.5937/SPSUNP1601011M doi: 10.5937/SPSUNP1601011M
![]() |
[36] | X. Zhan, Matrix inequalities, Berlin: Springer Press, 2002. |
[37] |
M. Bianchi, A. Cornaro, J. L. Palacios, A. Torriero, New bounds of degree-based topological indices for some classes of c-cyclic graphs, Discrete Appl. Math., 184 (2015), 62–75. https://doi.org/10.1016/j.dam.2014.10.037 doi: 10.1016/j.dam.2014.10.037
![]() |
[38] |
D. M. Cardoso, G. Pastén, O. Rojo, On the multiplicity of α as an eigenvalue of Aα(G) of graphs with pendant vertices, Linear Algebra Appl., 552 (2018), 52–70. https://doi.org/10.1016/j.laa.2018.04.013 doi: 10.1016/j.laa.2018.04.013
![]() |
[39] |
X. Huang, H. Lin, J. Xue, The Nordhaus-Gaddum type inequalities of Aα-matrix, Appl. Math. Comput., 365 (2020), 124716. https://doi.org/10.1016/j.amc.2019.124716 doi: 10.1016/j.amc.2019.124716
![]() |
[40] | I. Schur, Über eine Klasse von Mittelbildungen mit Anwendungen die Determinanten, Theorie Sitzungsber. Berlin. Math. Gesellschaft, 22 (1923), 9–20. |
[41] | A. W. Marshall, I. Olkin, B. C. Arnold, Inequalities: theory of majorization and its applications, 2 Eds., New York: Springer Press, 2011. https://doi.org/10.1007/978-0-387-68276-1 |
[42] |
H. S. Wilf, The eigenvalues of a graph and its chromatic number, J. London Math. Soc., 42 (1967), 330–332. https://doi.org/10.1112/jlms/s1-42.1.330 doi: 10.1112/jlms/s1-42.1.330
![]() |