1,262
Views
1
CrossRef citations to date
0
Altmetric
Articles

On the topological matrix and topological indices

Abstract

The topological matrix M of a graph G are indexed by V(G). If i,jL then the (i,j)entry of M(G) is 1 and the (i,j)entry of M(G) is 0 for otherwise. In this paper, an upper bound for this topological matrix is found and the weighted topological matrix is defined. Also, some inequalities for the topological energy and the weighted Laplacian topological energy and of M(G) are obtained.

The Topological index TI(G) of a graph G is qualified by TI=|A+D|. Some relations are found for basic mathematical operations of TI(G) and some bounds are reported for some topological indices in this paper.

1 Introduction

Let G be a simple connected graph on the vertex set V(G) and the edge set E(G). If vi and vj are adjacent, then we use the notation vivj. For viV(G), the degree of vi, denoted by di is the number of the vertices adjacent to vi. Let A(G) be adjacency matrix of G and λ1, λ2, …, λn be its eigenvalues (λ1 is the greatest eigenvalue). The graph eigenvalues provide these well-known results: i=1nλi=0,i=1nλi2=2m,λ12mn Citation[1,2]. In this paper, the reciprocals of the graph indices are examined under these basic concepts. The graph indices which is one of the topics involved in the studies of graph theory. An important part of these graph indices is topological indices that is used in chemical graph theory, particularly. Chemical graph theory can make topological representation of a molecule. The numerical values of this chemical structure are descriptors. A topological index is calculated from a graph representing a molecule and this index is also invariant number of graphs. Most recommended topological indices consists of vertex, edge, degree relationship. This, describes the atomic relationship in chemical graph theory. This means that, graph theory contributes organic compounds with calculations and creating diagrams. This paper indicates special bounds for some topological indices by the help of different mathematical formulas and series. These topological indices are described as follows:

Let A be an adjacency matrix and D be a graph distance matrix. The topological index defined by TI=|A+D|where |A+D| denotes the determinant of the matrix addition.

The atom-bond connectivity index ABC is one of the popular degree-based topological indices in chemical graph theory Citation[3] such that ABC=ABC(G)=vivjE(G)di+dj2didj.

Randić index is a well known topological structure such that R(G)=vivjE(G)1didj.It is referred that the articles Citation[4] and Citation[5] for the various properties on the Randić index. Another remarkable topological descriptor is the Harmonic index, characterized in Citation[6] as H(G)=vivjE(G)2di+dj.

Recently Citation[7], a graph invariant came into the focus of attention, defined as F=F(G)=i=1ndi3which for historical reasons Citation[8] was named forgotten topological index. F satisfies the identities F(G)=ijdi2+dj2.

One of the aim of this paper is to define the weighted topological matrix of a graph and to provide upper bounds on the weighted topological matrix Mij and the eigenvalues of the graph G. It is needed that the following results in order to effect these bounds:

The Topological matrix Mij of a graph G is defined by (1.1) Mijw=1;i,jL0;otherwise.(1.1) where L is a link.

Let λ1(M) be greatest eigenvalue and M=(mij) be an nxn irreducible nonnegative matrix. Let Ri(M)=j=1mmij Citation[9]. Then, (1.2) (minRi(M):1in)λ1(M)(maxRi(M):1in).(1.2)

Let V be a vertex set, viV and ei be an average degree . If G is a simple, connected graph then [Citation10] (1.3) λ1(G)max(eiej:1i,jn,vi,vjE)(1.3)

The energy of a graph G is described as E(G)=i=1n|λi| where λi, i=1,2,,n are the eigenvalues of graph G. The Laplacian energy of a graph G is qualified as LE(G)=i=1n|μi2mn| where μi, i=1,2,,n are the laplacian eigenvalues of graph G Citation[2,11].

Lemma 1.1

[Citation11] Let ai,biR and a,b,A,B be real constants such that i=1,2,,n , 0<aaiA and 0<bbiB . Then, |ni=1naibii=1naii=1nbi|α(n)(Aa)(Bb)

where α(n)=n[n2]([11n][n2]) .

See Citation[1,12–16] for details about graph theory and topological structures.

The scheme of the paper is as follows: In Section 1, a list of some previously known definitions and results are introduced. In Section 2, a new topological matrix and topological energy are defined for the weighted graphs. Also, some relations are established in terms of edges, vertices and degrees. In addition, different relationships are obtained for topological indices.

2 Main results

2.1 On the topological matrix

In this section, the weighted Topological matrix and the weighted Topological energy are defined. Furthermore, a relationship is found between the weighted Topological energy and the eigenvalues of the weighted Topological matrix.

Definition 2.1

Let G be a simple, connected, edge weighted graph. The weighted Topological matrix Mw=Mw(G) of G defined by, (2.1) Mijw=wij;i,jL0;otherwise.(2.1)

where w(u)=iuti, w(v)=jvcj.

The weighted Topological eigenvalues ρ1w, ρ2w, …, ρnw are the eigenvalues of its weighted Topological matrix Mw. These eigenvalues can be put in order such that ρ1wρ2wρnw.

Definition 2.2

Let G be a simple, connected weighted graph with n vertex. Let G be edge weighted graph and these weights be positive real numbers. The weighted Topological energy ME=ME(Gw) of Gw is defined as follows: (2.2) ME=ME(Gw)=i=1n|ρiw|(2.2) where ρiw is the eigenvalue of weighted Topological matrix Mw.

Definition 2.3

Let G be a simple, connected weighted graph. The weighted Laplacian Topological energy LME=LME(Gw) of Gw is defined as follows: (2.3) ME=ME(Gw)=i=1n|ρiw2mn|(2.3) where ρiw is the laplacian eigenvalue of weighted Topological matrix Mw.

Theorem 2.4

If G is a simple, connected graph of M , then

λ1(G)2ed1 where d1 is the degree of G .

Proof

Let multiply the topological matrix with the diagonal matrix and the inverse of diagonal matrix. Let show this multiplication by D(G)1M(G)D(G). Let consider an eigenvector of D(G)1M(G)D(G) and this eigenvector be X=(x1,x2,,xn)T. Let one eigencomponent xi=1 and the other eigencomponent 0<xk1 for every k. Let xj=maxk(xk:vivkE,ik). It is known that

(D(G)1M(G)D(G))X=λ1(G)X. It is implies that λ1(G)xi=k(m1kdkd1)xk. Since (1,k)L then, λ1(G)xi=1d1k(dk)xk. It is well known that i=1ndi=2e. Thus, λ1(G)2ed1. From (1.3), the inequality results that λ1(G)2ed1. □

Corollary 2.5

Let G has n vertices and m edges. Let Ḡ be the complement of a graph G . If G and Ḡ be connected non-singular graphs of M then, λ1(G)λ1(Ḡ)(n1)(2end1)(n1d1)d1.

Proof

Using Theorem 2.4, λ1(G)λ1(Ḡ)2ed12ēn1d1.Since, 2e+2ē=n(n1), then λ1(G)λ1(Ḡ)(n1)(2end1)(n1d1)d1.

Theorem 2.6

Let G be a simple, connected graph with m edges and ME(G) be a Topological energy of G then, |ME(G)+ME(Ge)|4m+4m(mδ).

Proof

Let λi be ith eigenvalue of M(G). Define λi=λi(Ge). It is known that ME(G)=i=1n|λi|. Since i=1n(λi)2=2m and i=1n(λi)2=2(mδ) then, (2.4) i=2n(|λi|+|λi|)22m(λ1)2+2m2δ(λi)2(2.4) (2.5) +2(2m(λ1)2)(2m2δ(λi)2)(2.5) (2.6) 4m4m2n2+4m24mδ4δ2n22n+1+2(2m4m2n2)(2m2δ4m2n2).(2.6) This means that, (2.7) ME(G)+ME(Ge)|λ1|+|λ1|+i=2n|λi|+i=2n|λi|(2.7) (2.8) |λ1|+|λ1|(2.8) (2.9) +4m4m2n2+4m24mδ4δ2n22n+1+2(2m4m2n2)(2m2δ4m2n2).(2.9) Since |λ1|2mn and |λ1|2(mδ)n1, the inequality turns into (2.10) ME(G)+ME(Ge)2mn+2(mδ)n1(2.10) (2.11) +4m4m2n2+4m24mδ4δ2n22n+1+2(2m4m2n2)(2m2δ4m2n2).(2.11) It is constructed a sequence Gn(n2) of graphs to complete the argument such that . This completes argument. □

Theorem 2.7

Let Gw be a weighted graph with n vertices and m edges. Let ρ1wρ2wρk1wρkw(kn) be k non-zero laplacian eigenvalues of the weighted Topological matrix and MLE(Gw) be a Topological Laplacian energy of Gw then, |MLE(Gw)|(2m(n1)n)2kα(k)|ρ1wρkw|2.

where α(k)=k[k2]([11k][k2]) while [x] denotes integer part of a real number x .

Proof

If MLE(Gw) is a weighted Topological Laplacian energy of Gw then, MLE(Gw)=i=1k|ρiw2mn| and (2.12) i=1k|ρiw2mn|2=i=1n|ρiw2mn|2=i=1n(ρiw2mn)2.(2.12) Seeing that, (2.13) i=1n(ρiw2mn)2=i=1n((ρiw)24mnρiw+4m2n2)(2.13) (2.14) =(2m)28m2n+4m2n2.(2.14) It requires that, (2.15) i=1k|ρiw2mn|2=4m28m2n+4m2n2.(2.15)

Setting ai=|ρiw2mn|, bi=|ρiw2mn|, a=b=|ρkw2mn| and A=B=|ρ1w2mn|, i=1,2,,k. Lemma 1.1 becomes (2.16) |ki=1k|ρiw2mn|2(i=1k|ρiw2mn|)2|α(k)(|ρ1w2mn||ρkw2mn|)2.(2.16) Thus, the inequality transforms into (2.17) (4m28m2n+4m2n2)k(MLE(Gw))2α(k)|ρ1wρkw|2.(2.17) Hence, |MLE(Gw)|(2m(n1)n)2kα(k)|ρ1wρkw|2.

2.2 Some notes on the topological indices

In this section, some results are given for Topological indices of graph operations. These results are obtained in terms of Topological indices and in fact, the bounds are tight. Also, upper bounds on the some topological indices are determined involving just the forgotten topological index, the Randić index, the ABC index, the Harmonic index and the edges.

Theorem 2.8

Let G and H be two simple graph and EG , EH be edge sets, VG , VH be vertex sets, respectively. Then, (1)TI(G+H)=(1)V+1(V+1)2V+.(2)TI(GH)=(1)V1(V1)2V.(3)TI(GH)=(1)V1(V1)2V. where |VG+VH|=V+ , |VGVH|=V and |VGVH|=V .

Proof

(1) Let EG and EH be edge sets, VG, VH be vertex sets, V=VGVH, E=EGEH(uG,uH)uGV(G),uHV(H). It is obvious from the definition, TI(G+H)=(1)V+1(V+1)2V+.(2) Let denote G=G×H edge sets. To define the product G×H of two graphs think any two points u=(uG,uH) and v=(vG,vH) and u,vV=VG×VH. Hence, u and v are adjacent in G×H whenever [uG=vG and uH adjacent to vH] or [uH=vH and uG adjacent to vG]. Then, TI(GH)=(1)V1(V1)2V.(3) If two graphs G and H have at least one common vertex then their intersection will be a graph such that V(GH)=V(G)V(H) and E(GH)=E(G)E(H). Hence, TI(GH)=(1)V1(V1)2V.

Theorem 2.9

Let G and H be two simple graphs. If G and H are isomorf then topological indices of G and H are equal.

Proof

Let VG and VH be vertex sets of G and H, respectively. Let EG and EH be edge sets of G and H, respectively. Since G and H are isomorf graphs then VG=VH and EG=EH. Thus, the adjacency matrix and the distance matrix of these graphs are equal that is ; A(G)=A(H) and D(G)=D(H). It is seen that TI(G)=TI(H). □

Furthermore, some formulas for Topological index of popular graphs are obtained, directly. Let Kn, Pn, Cn and Sn show the n-vertex complete graph, path, cycle and star graph, respectively.

Corollary 2.10

Let G be a simple connected graph with n vertices. Then,

(1) For 2n6 , TI(Kn)=TI(Cn)=TI(Sn).

(2) For 2n3 , TI(Kn)=TI(Cn)=TI(Sn)=TI(Pn).

(3) For n2 , TI(Kn)=TI(Sn)=(1)n1(n1)2n.

Theorem 2.11

Let G be a connected graph with m edges, then 4H1(G)F(G)2R12(G)2m(F(G)+2R12(G)).

Proof

Minkowski inequality gives (vivjE(G)(di+dj+1)2)12(vivjE(G)(di+dj)2)12+(vivjE(G)1)12.In the Bernoulli inequality, (1+x)α is greater than or equal to 1+αx for x1. It gives (1+di+dj)21+2(di+dj). Hence, vivjE(G)(1+2(di+dj))=m+2vivjE(G)(di+dj). It follows that, (m+2vivjE(G)(di+dj))12(vivjE(G)(di+dj)2)12+m.From the above inequality, m+2vivjE(G)(di+dj)vivjE(G)(di+dj)2+m+2mvivjE(G)(di+dj)2.It is known that, vivjE(G)(di2+dj2)=vivjE(G)((di+dj)22didj). By expanding the terms under summation, vivjE(G)(di+dj)2=F(G)+2R12(G).It is concluded that 2vivjE(G)(di+dj)F(G)+2R12(G)+2m(F(G)+2R12(G)).Hence, 4H1(G)F(G)2R12(G)2m(F(G)+2R12(G)).

Theorem 2.12

Let G be a connected regular graph with m edges, then ABC(G)R(G)2(H1(G)m).

Proof

By the definition of ABC index, vivjE(G)di+dj2didjvivjE(G)di+dj2didj.It follows that, ABC2(G)vivjE(G)di+dj2vivjE(G)didj.Indeed, it is seen that ABC2(G)2H1(G)2mR2(G).Hence, ABC(G)R(G)2(H1(G)m).

3 Conclusion

Topological indices is the corner stone of the theories in this paper. Therefore, new inequalities and relations are obtained for topological structures throughout this paper. Firstly, the weighted graphs are examined and some special bounds are formed. In continuation, to gain a more intuitive understanding of the topological index of composite graphs and original graphs, some relations are found. Lastly, some relationships are stated between different topological indices.

Acknowledgments

The author would like thank for the valuable suggestions of referees.

References

  • BapatR.B., Graphs and Matrices2010Indian Statistical InstituteNew Delhi 110016, India
  • DasK.C.MojalalS.A., On energy and Laplacian energy of graphs Electron. J. Linear Algebra 312016 167–186
  • GutmanI., Degree-based topological indices Croat. Chem. Acta 862013 351–361
  • Recent Results in the Theory of RandiĆ Index2008Univ.Kragujevac
  • LiX.ShiY., A survey on the Randić index MATCH Commun. Math. Comput. Chem. 592008 127–156
  • FajtlowiczS., On conjectures of Graffiti–II Congr. Number 601987 187–197
  • FurtulaB.GutmanI., A forgotten topological index J. Math. Chem. 532015 1184–1190
  • GutmanI., On the origin of two degree-based topological indices Bull. Acad. Serb. Sci. Arts 1462014 39–52
  • HornR.A.JohnsonC.R., Matrix Analysis1985Cambridge University PressNew York
  • DasK.C.KumarP., Some new bounds on the spectral radius of graphs Discrete Math. 2812004 149–161
  • MilovanovćI.Z.Milovanovć.E.I.ZakićA., A short note on graph energy MATCH Commun. Math. Comput. Chem. 722014 179–182
  • AslamA.AhmadS.BinyaminM.A.GaoW., Calculating topological indices of certain OTIS interconnection networks Open Chem. 172019 220–228
  • AslamA.NadeemM.F.ZahidZ.ZafarS.GaoW., Computing certain topological indices of the line graphs of subdivision graphs of some rooted product graphs Mathematics 7 5 2019 393
  • BüyükköseS.Kaya GökG., Graf Teoriye Giris2018Nobel Akademik Yayıncılık Egitim Danısmanlık Tic.Ltd.StiAnkara
  • GaoW.IqbalZ.IshaqM.AslamA.SarfrazR., Topological aspects of dendrimers via distance based descriptors IEEE Access 7 1 2019 35619–35630
  • VasudevC., Graph Theory with Applications, Vol. 12006New Age International PublishersNew Delhi/ Indian4–5, 21, 56–57