• No results found

On two energy like invariants of line graphs and related graph operations

N/A
N/A
Protected

Academic year: 2020

Share "On two energy like invariants of line graphs and related graph operations"

Copied!
15
0
0

Loading.... (view fulltext now)

Full text

(1)

R E S E A R C H

Open Access

On two energy-like invariants of line

graphs and related graph operations

Xiaodan Chen

1,2*

, Yaoping Hou

2

and Jingjian Li

1

*Correspondence: x.d.chen@live.cn 1College of Mathematics and

Information Science, Guangxi University, Daxue Road 100, Nanning, 530004, P.R. China

2Department of Mathematics,

Hunan Normal University, Lushan Road 36, Changsha, 410081, P.R. China

Abstract

For a simple graphGof ordern, let

μ

1≥

μ

2≥ · · · ≥

μ

n= 0 be its Laplacian

eigenvalues, and letq1≥q2≥ · · · ≥qn≥0 be its signless Laplacian eigenvalues. The Laplacian-energy-like invariant and incidence energy ofGare defined as, respectively,

LEL(G) =

n–1

i=1

μ

i and IE(G) = n

i=1

qi.

In this paper, we present some new upper and lower bounds onLELandIEof line graph, subdivision graph, para-line graph and total graph of a regular graph, some of which improve previously known results. The main tools we use here are the Cauchy-Schwarz inequality and the Ozeki inequality.

MSC: 05C50; 05C90

Keywords: Laplacian-energy-like invariant; incidence energy; line graph; subdivision graph; para-line graph; total graph

1 Introduction

There are a large number of remarkable chemical applications in graph theory, one of which is based on the close correspondence between the graph eigenvalues and the molec-ular orbital energy levels ofπ-electrons in conjugated hydrocarbons. For the Hüchkel molecular orbital (HMO) approximation, the totalπ-electron energy in conjugated hy-drocarbons is given by the sum of absolute values of the eigenvalues of the corresponding molecular graph in which the degree of each vertex is not more than three in general. In , Gutman [] extended the concept of energy to all simple graphs, and defined the energy of a graphG(of ordern) as

E(G) =

n

i= |λi|,

whereλ≥λ≥ · · · ≥λnare the eigenvalues of the adjacency matrixA(G) ofG(also called

the eigenvalues of the graphG). The concept of graph energy was further extended to any matrix in []: the energyE(M) of a (not necessarily square) matrixMis the sum of its singular values, where the singular values ofMare the square roots of the eigenvalues of

(2)

the (square) matrixMMT(MTstands for the transpose ofM). Obviously,E(G) =E(A(G)).

For more details in the theory of graph energy, see [] by Liet al.

The Laplacian matrixL(G) of a graphGis another important and well-studied matrix in graph theory, which is defined to beL(G) =D(G) –A(G), whereD(G) is the diagonal matrix of vertex degrees ofG. The eigenvalues ofL(G) are also called the Laplacian eigenvalues of the graphG, which are always denoted byμ≥μ≥ · · · ≥μn–≥μn= . Based on the

Laplacian eigenvalues, Liu and Liu [] conceived the following invariant:

LEL(G) =

n–

i= √

μi. ()

They called this invariant the Laplacian-energy-like invariant, which was later proved to be an energy like invariant []. Nowadays,LELhas become a successful molecular descrip-tor []. A great deal of work has been done onLEL, see [] for a comprehensive survey.

Recall that giving an arbitrary orientation to each edge of Gwould yield an oriented graphG. LetB(G) be the (vertex-edge) incidence matrix ofG. ThenB(G)B(G)T=L(G), and

henceLEL(G) =E(B(G)). This provides a new interpretation forLEL: oriented incidence energy [], and furnishes a new insight into its possible physical or chemical meaning.

LetB(G) be the (vertex-edge) incidence matrix ofG. Motivated by Nikiforov’s idea and

LEL, Jooyandehet al.[] introduced the concept of incidence energy of a graphG: if the singular values ofB(G) areσ,σ, . . . ,σnthen the incidence energy ofGisIE(G) =

n i=σi=

E(B(G)). Note thatB(G)B(G)T=A(G) +D(G) =Q(G) is the signless Laplacian matrix ofG,

with its eigenvaluesq≥q≥ · · · ≥qn≥ (also called the signless Laplacian eigenvalues

of the graphG). Thus, it follows that

IE(G) =

n

i= σi=

n

i= √

qi. ()

Observe thatLEL(G) =IE(G). Furthermore, we have the following relation betweenLEL

andIE.

Theorem .(see []) Let G be a graph of order n.Then LEL(G)≤IE(G),with the equality holding if and only if G is bipartite.

For more results onIE, we refer the reader to [, –].

The line graph, subdivision graph, para-line graph and total graph are the well-known operations on graphs, which can produce many new types of graphs. In [], Ramaneet al.

(3)

in-cidence energy of (iterated) line graphs of a regular graph were also obtained by Gutman

et al.[].

In this paper, we give some new upper and lower bounds on the Laplacian-energy-like invariant and the incidence energy of line graph, subdivision graph, para-line graph and total graph of a regular graph, some of which improve the corresponding bounds in [] and [].

2 Preliminaries

In this section, we recall the concepts of line graphs and related graph operations, and list some lemmas that will be used in the subsequent sections.

The line graphL(G) of a graphGis the graph whose vertices are the edges ofG, with two vertices inL(G) adjacent whenever the corresponding edges inGhave exactly one vertex in common. IfGis a regular graph of degreerwithnvertices andm(=nr/) edges, then characteristic polynomial of L(G) can be expressed in terms of the characteristic polynomial ofG, namely []

PA(L(G))(x) = (x+ )mnPA(G)(xr+ ), ()

where we writePM(x) for the characteristic polynomial of a matrixM. By (), we have the

following lemma immediately.

Lemma . Let G be a regular graph of degree r with n vertices.If the eigenvalues of G are

λ,λ, . . . ,λn,then the eigenvalues ofL(G)are r–  +λ,r–  +λ, . . . ,r–  +λn,and–

(with multiplicity n(r– )/).

The subdivision graphS(G) of a graphGis the graph obtained by inserting a new vertex into every edge ofG. Clearly,S(G) is a bipartite graph. Moreover, ifGis a regular graph of degreerwithnvertices andmedges, then []

PL(S(G))(x) = (–)m( –x)mnPL(G)

x(r+  –x). ()

From () we arrive at the next lemma (see also []).

Lemma . Let G be a regular graph of degree r with n vertices.If the Laplacian eigenval-ues of G areμ,μ, . . . ,μn,thenS(G)has n(r– )/Laplacian eigenvalues equal toand

the followingn Laplacian eigenvalues:

r+ ±(r+ )– μ

i

 (i= , , . . . ,n).

The para-line graphC(G) of a graphG, which was first introduced in [], is defined to be the line graph of the subdivision graph ofG. In [], the para-line graph is also called the clique-inserted graph.

(4)

 (with multiplicity n(r– )/),and

r– ±r+  + λ

i

 (i= , , . . . ,n).

The total graphT(G) of a graphGis the graph whose vertices are the vertices and edges ofG, with two vertices ofT(G) adjacent if and only if the corresponding elements ofGare adjacent or incident.

Lemma .(see []) Let G be a regular graph of degree r(r> )having n vertices.If the eigenvalues of G areλ,λ, . . . ,λn,thenT(G)has n(r– )/eigenvalues equal to–and the

followingn eigenvalues:

r–  + λi±

r+  + λ

i

 (i= , , . . . ,n).

The following is known as the Ozeki inequality.

Lemma .(see []) Leta¯= (a,a, . . . ,an)andb¯= (b,b, . . . ,bn)be two positive n-tuples

with <aaiA and <bbiB,where i= , , . . . ,n.Then

n

i= ai

n

i= bi

n

i= aibi

 ≤ 

n

(ABab).

Lemma .(see [, ]) Let G be a graph with n vertices,m edges and maximum degree.

Then

(m)

n+ mLEL(G)≤ √

+  +(n– )(m– ),

with the left equality if and only if G∼=Knand the right if and only if G∼=Knor G∼=Sn.

From Lemma ., it follows directly that ifGbe a regular graph of degreerwithn ver-tices, then

nr

r+ ≤LEL(G)≤

r+  +(n– )(nrr– ), ()

with both equalities if and only ifG∼=Kn.

Lemma .(see []) Let G be a graph of order n with at least one edge.Thenμ=μ= · · ·=μn–if and only if G∼=Kn.

Finally, we should remark that ifGis a regular graph of degreer= , thenGis nothing but the disjoint union of independent edges. Hence, in the following, for avoiding the triviality we always assume thatr≥.

3 The Laplacian-energy-like invariant

(5)

Theorem .(see []) Let G be a regular graph of degree r with n vertices.Then

LELL(G)=LEL(G) +n(r– ) 

r.

By Theorem . and (), we have the following corollary directly.

Corollary . Let G be a regular graph of degree r with n vertices.Then

nr

r+ +

n(r– ) 

rLELL(G)≤√r+  +(n– )(nrr– ) +n(r– ) 

r,

with both equalities holding if and only if G∼=Kn.

Remark  In [], Wang and Luo proved that ifGis a regular graph ofnvertices and of degreer, then

n(r– ) 

r<LELL(G)≤(n– )nr+n(r– ) 

r, ()

with the right equality if and only ifG∼=Kn. Evidently, the lower bound in Corollary . is

better than that in (). For the upper bound, our bound is also better than that in (). To see this, we only need to show that

r+  +(n– )(nrr– )≤(n– )nr, ()

that is,

(n– )(nrr– )≤(n– )nr–√r+ ,

that is,

(n– )(r+ )nrnr+ (n– )(r+ ) = (√nr)+(n– )(r+ ),

which is clearly obeyed.

We now consider the case of subdivision graphs.

Theorem . Let G be a regular graph of degree r with n vertices.Then

(i) LEL(S(G))≤(n– )

r+  + 

n–LEL(G) + √

r+  +

n(r–)

 ,with the equality if and

only ifG∼=Kn;

(ii) LEL(S(G)) > (n– )

r+

+ 

n–LEL(G) + √

r+  +

n(r–)  .

Proof Suppose thatμ ≥μ≥ · · · ≥μn–≥μn=  are the Laplacian eigenvalues of G.

Then from () and Lemma ., it follows that

LELS(G)=

n–

i=

r+  +(r+ )– μ

i

 +

r+  –(r+ )– μ

i

+√r+  +

(6)

= n– i=

r+  +(r+ )– μ

i

 +

r+  –(r+ )– μ

i

+√r+  +

n(r– )  = n– i=

r+  + √μi+

r+  +

n(r– )

 . ()

Further, by the Cauchy-Schwarz inequality, we obtain

LELS(G)≤

(n– )n–

i=

(r+  + √μi) +

r+  +

n(r– ) 

= (n– ) r+  + 

n– LEL(G) +

r+  +

n(r– )

 .

Moreover, it is easy to see that the equality holds if and only ifμ=μ=· · ·=μn–, which,

by Lemma ., is equivalent toG∼=Kn. Hence, (i) follows.

We now prove (ii). Take ai =

r+  + √μi and bi = , i= , , . . . ,n– . Set A=

r+  + √r, a=√r+ , andB=b= . Clearly,  <aaiA(as ≤μn–≤ · · · ≤ μ≤r), and  <bbiB. Furthermore, we get

(ABab)= r

(r+  + √r+√r+ ) <

r

(r+ )< .

Then by Lemma ., it follows from () that

LELS(G)≥

(n– )n–

i=

(r+  + √μi) –

 (n– )

(ABab)

+√r+  +

n(r– ) 

> (n– ) r+ +

n– LEL(G) +

r+  +

n(r– )

 ,

as desired, completing the proof.

Theorem ., together with (), would yield the next immediate corollary.

Corollary . Let G be a regular graph of degree r with n vertices.Then

(i) LEL(S(G))≤(n– )

r+  +(

r++√(n–)(nrr–))

n– + √

r+  +

n(r–)

 ,with the

equality if and only ifG∼=Kn;

(ii) LEL(S(G)) > (n– )r++(n–)nr

r++ √

r+  +

n(r–)  .

Remark  It was showed in [] that ifGis a regular graph ofnvertices and of degreer, then

n(r– )

 +n

r+  <LELS(G)≤(n– )√r+√r+  +

(nr– )

(7)

the right equality holds if and only ifG∼=K. Note that the lower and upper bounds in

Corollary . are better than those in (), respectively. Indeed, for the lower bound, since

r r+≥

 andn>n– , we have

r+

+

nr

(n– )√r+ +

r+  +

n(r– ) 

> (n– ) r+ +

r+√r+  +

n(r– )  >n

r+  +

n(r– )

 .

For the upper bound, using () and the fact that nn–≤, we obtain

(n– )

r+  +(

r+  +√(n– )(nrr– ))

n–  +

r+  +

n(r– ) 

≤(n– )

r+  +

(n– )nr n–  +

r+  +

n(r– ) 

≤(n– )

r+  + √r+√r+  +

n(r– ) 

= (n– )√r+√r+  +

(nr– )

 .

Next, we turn our attention to the case of para-line graphs. Notice that ifGisr-regular, thenC(G) is alsor-regular, and hence it follows from Lemma . that the Laplacian eigen-values of C(G) are r+  (with multiplicity n(r– )/),r (with multiplicityn(r– )/), and

r+ ±(r+ )– μ

i

 (i= , , . . . ,n),

whereμ≥μ≥ · · · ≥μnare the Laplacian eigenvalues ofG. Using the same argument

as the proof of Theorem ., we may arrive at the following result.

Theorem . Let G be a regular graph of degree r with n vertices.Then

(i) LEL(C(G))≤(n– )

r+  +n–LEL(G) + (n(r–)+ )√r+  +n(r–)√r,with the equality if and only ifG∼=Kn;

(ii) LEL(C(G)) > (n– )

r++n–LEL(G) + (n(r–)+ )√r+  +n(r–)√r.

Likewise, the next corollary follows evidently from Theorem . and ().

Corollary . Let G be a regular graph of degree r with n vertices.Then

(i) LEL(C(G))≤(n– )

r+  +(

r++√(n–)(nrr–))

n– + (

n(r–)  + )

r+  +n(r–)√r,with the equality if and only ifG∼=Kn;

(ii) LEL(C(G)) >r++ nr

(n–)√r++ (

n(r–)  + )

r+  +n(r–)√r.

(8)

Theorem . Let G be a regular graph of degree r with n vertices.Then

(i) LEL(T(G))≤nr+  +√(n– )nr+n(r–)√r+ ,with the equality if and only if

G∼=Kn;

(ii) LEL(T(G)) > (n– )√r+  +√r+  +n(r–)√r+ .

Proof Note first thatT(G) is a regular graph of degree rwith n(r+)vertices. Then from Lemma ., it follows thatT(G) hasn(r–)Laplacian eigenvalues equal to r+  and the following nLaplacian eigenvalues:

r+  + μi±

(r+ )– μ

i

 (i= , , . . . ,n),

whereμ≥μ≥ · · · ≥μn=  are the Laplacian eigenvalues ofG. Then by () and some

computation, we get

LELT(G)=

n–

i=

r+  + μi+ 

μ

i + (r+ )μi+

r+  +n(r– ) 

r+ . ()

On the other hand, by using the Cauchy-Schwarz inequality and the fact thatni=–μi=nr

andni=–μi =nr+nr, we have

n–

i=

r+  + μi+ 

μ

i+ (r+ )μi

(n– )n–

i=

r+  + μi+ 

μi + (r+ )μi

≤(n– )

r+  + nr

n– + 

n– 

n–

i=

μi + (r+ )μi

= (n– )

r+  + nr

n– +  nr n– (r+ )

= (n– )√r+  +(n– )nr,

which, together with (), yields the required upper bound. For the sharpness of this bound, it is not difficult to check that if G∼=Kn then the equality holds. Conversely,

we assume that the equality holds. Then the above two inequalities should be equali-ties. Thus, it follows from the second inequality that, for any ≤i,jn–  andi=j,

μi+ (r+ )μi=

μj + (r+ )μj, that is, (μiμj)(μi+μj+r+ ) = , which implies that

μi=μj. In other words, we haveμ=μ=· · ·=μn–. Note that in this case, the first

in-equality is also an in-equality. Eventually, by Lemma ., we getG∼=Kn.

We next prove (ii). Sinceμi≥,i= , , . . . ,n– , by () we get

LELT(G)≥

n–

i=

r+  + μi+ 

(r+ )μi+

r+  +n(r– ) 

(9)

We now take ai =

r+  + μi+ 

(r+ )μi and bi = , i = , , . . . ,n– . Set A =

r+  + √(r+ )r, a=√r+ , and B=b= . It is easy to see that  <aaiA,

and  <bbiB. Furthermore, ifr≥, thenA=

r+  + √(r+ )r≤√r+  and hence,

(ABab)(r+  –r+ )= (r)

r+  + √(r+ )(r+ )

< r

r = r;

ifr= , then by a direct calculation, we also obtain (ABab)< r. Thus by Lemma .

and (), we get

n–

i=

r+  + μi+ 

(r+ )μi

(n– )n–

i=

r+  + μi+ 

(r+ )μi

– (n– )

(ABab)

> (n– )

nr n– +

√r+ 

n–  LEL(G) + 

> (n– ) r+ 

r+ + 

nr n– + 

> (n– )√r+ ,

which, together with (), would yield the required lower bound.

This completes the proof.

Remark  It was proved in [] that ifGis a regular graph ofnvertices and of degreer, then

LELT(G)>n(r– ) 

r+  +nr+ , and ()

LELT(G)≤(n– )√r+(nr– ) 

r+  +√r+ , ()

with the equality if and only ifG∼=K. Obviously, the lower bound in Theorem . is better

than that in (). For the upper bound, it is easy to check that, forn≥ andr> ,

r+  + nr

n– ≤

r+  + √r≤√r+√r+ ,

implying that

(n– )√r+  +(n– )nr≤(n– )(√r+√r+ ),

(10)

4 The incidence energy

In this section, we investigateIEof line graph, subdivision graph, para-line graph and total graph of a regular graph.

Theorem . Let G be a regular graph of degree r with n vertices.Then

(i) (see[])IE(L(G))≤(n– )

n–

n–r–  + √

r–  +n(r–)√r– ,with the equality if and only ifG∼=Kn;

(ii) IE(L(G)) > (n– )

n– n–r–  +

r–  +n(r–)√r– .

Proof Suppose thatλ≥λ≥ · · · ≥λnare the eigenvalues ofGand thatq≥q≥ · · · ≥qn

are the signless Laplacian eigenvalues ofG. Noticing thatGisr-regular, we haveqi=r+λi,

i= , , . . . ,n. Moreover, sinceL(G) is (r– )-regular, from Lemma ., it follows that the signless Laplacian eigenvalues ofL(G) are r–  +q, r–  +q, . . . , r–  +qn, and r– 

(with multiplicityn(r– )/). Thus, from () and noting thatq= r, it follows that

IEL(G)=

n

i=

r–  +qi+

r–  +n(r– ) 

r– . ()

Remark that (i) has been proved in [] by Gutmanet al.Hence, we here only give a proof for (ii). Takeai=

r–  +qiandbi= ,i= , , . . . ,n. SetA=

r– ,a=√r– , and

B=b= . It is clear that  <aaiA(as ≤qn≤ · · · ≤q≤q= r), and  <bbiB.

Furthermore, we have

(ABab)= r

r–  + √(r– )(r– ) < r

r–  ≤r.

Thus by Lemma . and the fact thatni=qi= (n– )r, it follows from () that

IEL(G)≥

(n– )n

i=

(r–  +qi) –

 (n– )

(ABab)

+√r–  +n(r– ) 

r– 

> (n– ) n–  n– r–  +

r–  +n(r– ) 

r– ,

as desired, completing the proof.

Remark  In [], Gutmanet al.gave the following lower bound:

IEL(G)>n(r– ) 

r–  +√r–  + (n– )√r– . ()

Obviously, the lower bound in Theorem . is better than that in (), sincenn–– >  always holds forn≥.

Since the subdivision graphS(G) of any graphGis bipartite, by Theorem . we have

(11)

Theorem . Let G be a regular graph of degree r with n vertices.Then

(i) IE(S(G))≤(n– )

r+  + (

r++√(n–)(nrr–))

n– + √

r+  +

n(r–)

 ,with the equality

if and only ifG∼=Kn;

(ii) IE(S(G)) > (n– )r++ nr

(n–)√r++ √

r+  +

n(r–)  .

Theorem . Let G be a regular graph of degree r with n vertices.Then

(i) IE(C(G))≤(n– ) r–  + r nr n–+

r+√r–  +n(r–)(√r+√r– ),with

the equality if and only ifG∼=Kn;

(ii) IE(C(G)) > (n– ) r–  –r+ 

r nr n––

 +

r+√r–  +n(r–)(√r+

r– ).

Proof SinceC(G) is a regular graph of degreer, it follows from Lemma . that the signless Laplacian eigenvalues ofC(G) arer–  (with multiplicityn(r– )/),r(with multiplicity

n(r– )/), and

r– ±(r– )+ q

i

 (i= , , . . . ,n),

where r=q≥q≥ · · · ≥qnare the signless Laplacian eigenvalues ofG. Now, by () and

a little calculation, we get

IEC(G)=

n

i=

r–  + (r– )r+qi+

r+√r– 

+n(r– )

 (

r+√r– ). ()

On the other hand, by the Cauchy-Schwarz inequality and the fact thatni=qi= (n– )r,

we have

n

i=

r–  + (r– )r+qi

(n– )n

i=

r–  + (r– )r+qi

≤(n– )

r–  + 

n– 

n

i=

(r– )r+qi

= (n– )

r–  +  r nr n– ,

which, together with (), yields the desired upper bound. Moreover, it is easily seen that the equality holds if and only ifq= randq=q=· · ·=qn, which, in turn, implies that

μ=μ=· · ·=μn–(sinceGisr-regular and henceμi= rqni+,i= , , . . . ,n). Thus,

by Lemma ., we haveG∼=Kn. Hence, (i) follows.

We next prove (ii). Takeai=

r–  + (r– )r+qiandbi= ,i= , , . . . ,n. SetA=

( + √)r– ,a=

(12)

 <bbiB. Furthermore, we get

(ABab)= 

(

( + √)r–  +

( + √)r– (√ + ))

< 

( + √)r– (√ + )< 

r.

Then by Lemma ., it follows from () that

IEC(G)≥

(n– )n

i=

r–  + (r– )r+qi

– (n– )

(ABab)

+√r+√r–  +n(r– )

 (

r+√r– )

> (n– )

r–  –  r+

n– 

n

i=

(r– )r+qi

+√r+√r–  +n(r– )

 (

r+√r– ). ()

Now, takeai=

(r– )r+qiandbi= ,i= , , . . . ,n, and setA=

r,a=√(r– )r, andB=b= . It is easy to check that  <aaiA,  <bbiB, and (ABab)< .

Thus, recalling thatni=qi= (n– )r, and again by Lemma . we have

n

i=

(r– )r+qi≥(n– ) r–

nr n– –

 ,

which, together with (), yields the required result.

The proof is completed.

Theorem . Let G be a regular graph of degree r with n vertices.Then

(i) IE(T(G))≤n√r–  +√(n– )(n– )r+√r+n(r–)√r– ,with the equality if and only ifG∼=Kn;

(ii) IE(T(G)) > (n– )

r–  + √(r– ) +√r+√r–  +n(r–)√r– .

Proof SinceT(G) is a regular graph of degree r, it follows from Lemma . that the sign-less Laplacian eigenvalues ofT(G) are r–  (with multiplicityn(r– )/), and

r–  + qi±

(r– )+ q

i

 (i= , , . . . ,n),

where r=q≥q≥ · · · ≥qn≥ are the signless Laplacian eigenvalues ofG. Thus, from

() and by a little computation, we have

IET(G)=

n

i=

r–  + qi+ 

q

i + (r– )qi+ (r– )r

+√r+√r–  +n(r– ) 

(13)

On the other hand, by using the Cauchy-Schwarz inequality and the fact thatni=qi=

(n– )randni=q

i = (n– )r+nr, we obtain

n

i=

r–  + qi+ 

q

i + (r– )qi+ (r– )r

(n– )n

i=

r–  + qi+ 

q

i + (r– )qi+ (r– )r

≤(n– )

r–  +(n– )r

n–  + 

n– 

n

i=

qi + (r– )qi+ (r– )r

= (n– )

r–  +(n– )r

n–  + 

(n– )r n–  (r– )

= (n– )

r–  + (n– )r

n– 

,

which, together with (), would yield the desired upper bound. For the sharpness of this bound, it is easy to check that ifG∼=Knthen the equality holds. Conversely, we assume that

the equality holds. Then the above two inequalities should be equalities. Thus, it follows from the second inequality that, for any ≤i,jnandi=j,

q

i+ (r– )qi+ (r– )r=

q

j + (r– )qj+ (r– )r, that is, (qiqj)[qi+qj+ (r– )] = , which implies thatqi=qj.

In other words, we haveq=q=· · ·=qn. Note also that in this case, the first inequality

is also an equality. Now, using the same argument as in the proof of Theorem ., we eventually arrive atG∼=Kn. Hence, (i) follows.

We next prove (ii). From (), we have

IET(G)>

n

i=

r–  + qi+ 

q

i + 

(r– )qi+

(r– )

+√r+√r–  +n(r– ) 

r– 

=

n

i=

r–  + √(r– ) + qi

+√r+√r–  +n(r– ) 

r– . ()

Now take ai =

r–  + √(r– ) + qi and bi = , i = , , . . . ,n. Set A =

r–  + √(r– ),a=

r–  + √(r– ), andB=b= . Evidently,  <aaiA

(as ≤qn≤ · · · ≤q≤q= r), and  <bbiB. Furthermore, we get

(ABab)< (r)

r–  + √(r– ) + √(r– )(r– )< r

r =

(14)

Thus, noting thatni=qi= (n– )rand from Lemma ., it follows that

n

i=

r–  + √(r– ) + qi

(n– )n

i=

r–  + √(r– ) + qi

–  (n– )

(ABab)

> (n– ) r–  + √(r– ) +(n– )

n–  r–  r

> (n– )

r–  + √(r– ) (forn≥),

which, together with (), yields the desired lower bound.

This completes the proof.

Competing interests

The authors declare that they have no competing interests.

Authors’ contributions

All authors contributed equally and significantly in writing this article. All authors read and approved the final manuscript.

Acknowledgements

The authors would like to thank the anonymous referees for their helpful comments and suggestions toward improving the original version of this paper. This work was supported in part by National Natural Science Foundation of China (Nos. 11501133, 11571101, 11461004), China Postdoctoral Science Foundation (No. 2015M572252), Guangxi Natural Science Foundation (No. 2014GXNSFBA118008) and the Scientific Research Fund of Guangxi Provincial Education Department (No. YB2014007).

Received: 20 September 2015 Accepted: 1 February 2016 References

1. Gutman, I: The energy of a graph. Ber. Math. Statist. Sekt. Forschungsz. Graz.103, 1-22 (1978) 2. Nikiforov, V: The energy of graphs and matrices. J. Math. Anal. Appl.326, 1472-1475 (2007) 3. Li, X, Shi, Y, Gutman, I: Graph Energy. Springer, New York (2012)

4. Liu, J, Liu, B: A Laplacian-energy-like invariant of a graph. MATCH Commun. Math. Comput. Chem.59, 397-419 (2008) 5. Gutman, I, Zhou, B, Furtula, B: The Laplacian-energy-like invariant is an energy like invariant. MATCH Commun. Math.

Comput. Chem.64, 85-96 (2010)

6. Stevanovi´c, D, Ili´c, A, Onisor, C, Diudea, M: LEL - a newly designed molecular descriptor. Acta Chim. Slov.56, 410-417 (2009)

7. Liu, B, Huang, Y, You, Z: A survey on the Laplacian-energy-like invariant. MATCH Commun. Math. Comput. Chem.66, 713-730 (2011)

8. Stevanovi´c, D, de Abreu, NMM, de Freitas, MAA, Vinagre, C, Vecchio, RD: On the oriented incidence energy and decomposable graphs. Filomat23, 243-249 (2009)

9. Jooyandeh, MR, Kiani, D, Mirzakhah, M: Incidence energy of a graph. MATCH Commun. Math. Comput. Chem.62, 561-572 (2009)

10. Akbari, S, Ghorbani, E, Koolen, JH, Oboudi, MR: A relation between the Laplacian and signless Laplacian eigenvalues of a graph. J. Algebraic Comb.32, 459-464 (2010)

11. Gutman, I, Kiani, D, Mirzakhah, M: On incidence energy of graphs. MATCH Commun. Math. Comput. Chem.62, 573-580 (2009)

12. Gutman, I, Kiani, D, Mirzakhah, M, Zhou, B: On incidence energy of a graph. Linear Algebra Appl.431, 1223-1233 (2009)

13. Zhou, B: More upper bounds for the incidence energy. MATCH Commun. Math. Comput. Chem.64, 123-128 (2010) 14. Ramane, HS, Walikar, HB, Rao, SB, Acharya, BD, Hampiholi, PR, Jog, SR, Gutman, I: Spectra and energies of iterated line

graphs of regular graphs. Appl. Math. Lett.18, 679-682 (2005)

15. Gao, X, Liu, Y, Liu, W: Kirchhoff index in line, subdivision and total graphs of a regular graph. Discrete Appl. Math.160, 560-565 (2012)

16. Yan, W, Yeh, Y, Zhang, F: The asymptotic behavior of some indices of iterated line graphs of regular graphs. Discrete Appl. Math.160, 1232-1239 (2012)

17. Wang, W, Luo, Y: On Laplacian-energy-like invariant of a graph. Linear Algebra Appl.437, 713-721 (2012) 18. Cvetkovi´c, D, Rowlinson, P, Simi´c, S: An Introduction to the Theory of Graph Spectra. Cambridge University Press,

Cambridge (2010)

(15)

20. Shirai, T: The spectrum of infinite regular line graphs. Trans. Am. Math. Soc.352, 115-132 (1999)

21. Zhang, F, Chen, Y, Chen, Z: Clique-inserted graphs and spectral dynamics of clique-inserting. J. Math. Anal. Appl.157, 211-225 (2009)

References

Related documents

Whether an individual plans to go to college is influenced both by how highly an actor values academic advancement and his or her expectations for success (Eccles 1994).

(2) differentiation of the types of innovative software that law firms may choose to adopt, (3) a brief review of classic diffusion theory, (4) the implications of opinion leaders

those surveyed on the island, (N=50) 75 percent repor{having either direct or indirect access to fr·esh produce, fish and shellfish through this informal network. These

In this study, the expression of endogenous KSP at mRNA and protein levels were determined in both HCC cell line Hep3B, and normal liver cell line THLE-3 by RT-PCR

Shariah audit process is the steps involve in performing shariah audit function which is done by auditors which basically consist of i) preliminary review ii) establish materiality

In this study, riverbed sediments taken from the Yangtze mainstream and its major tributaries were analyzed in terms of their ESR signal intensity and CI values of quartz in

Merris, Laplacian matrices of graphs: a survey, Linear Algebra Appl. Rada, Energy ordering of catacondensed hexagonal systems,

• Do not remove the paper source tray that you selected in either the print driver or the control panel, when the printer is printing.. • Do not open the doors when the printer