1

———————————————————————— ———————————
—————————————
———————————————————————— ———————————
—————————————
List of Figures

1.1 A 3-regular graph. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2

1.2 A graph with α = 4. . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
1.3 A graph with ω=3. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
1.4 A graph and its induced subgraph. . . . . . . . . . . . . . . . . . . . 4
1.5 A connected graph. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
1.6 A disconnected graph with 2 components. . . . . . . . . . . . . . . . 5
1.7 A graph with radius 2 and diameter 3. . . . . . . . . . . . . . . . . . 5
1.8 A graph with cut vertices. . . . . . . . . . . . . . . . . . . . . . . . . 6
1.9 A complete graph with 4 vertices. . . . . . . . . . . . . . . . . . . . . 6
1.10 A complete bipartite graph. . . . . . . . . . . . . . . . . . . . . . . . 7
1.11 S9 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
1.12 B(7,7) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
1.13 S14,9 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
1.14 1-ethyl-2-methylcyclobutane and the corresponding hydrogen-depleted
molecular graph. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
1.15 A molecular graph and its hydrogen depicted representation. . . . . 10
1.16 A grid type graph. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
1.17 Hexagonal graph. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
1.18 Triangular ladder. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
1.19 A graph G and its double graph D[G]. . . . . . . . . . . . . . . . . . 26

2
List of Tables

1.1 Order and size of the graph. . . . . . . . . . . . . . . . . . . . . . . . 16

1.2 Degree pattern and its count of the graph. . . . . . . . . . . . . . . . 16
1.3 Edge partition and its count of the graph. . . . . . . . . . . . . . . . 16
1.4 M-polynomials . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22

3
Chapter 1

Basic Terminology and Graph

Indices

1.1 Introduction
In this chapter, we recall some basic definitions, notations and results from graph
theory which will be used throughout this dissertation. In this dissertation, we deal
with simple, finite and connected graphs.

1.2 Preliminaries
A graph G = (V (G), E(G), ∼) is a triple, where V (G) and E(G) is the set of vertices
and the set of edges, respectively, and ∼ is the relation that associates each edge with
two vertices. The number of elements in V (G), |V (G)|, is called the order of G and
the number of elements in E(G), |E(G)|, is called the size of G. In this dissertation,
we use m and n for the size and order, respectively. The order of a graph G is at least
1 and the graph G with order 1 is called a trivial graph. A graph with order 0 (and
size 0) is called the null graph. A graph G is said to be finite if its order and size are
finite. In a graph G two vertices y and z are adjacent if yz ∈ E(G), we also say y and
z are neighbors of each other. An edge e is a loop if e = zz and edges e1 , e2 are parallel
edges if e1 = e2 = yz. A graph G without loops and parallel edges is called a simple
graph. The set N (z) of the neighbors of vertex z, is called the neighborhood of z. In
a graph G the number of vertices adjacent to a vertex z is called the degree of z in
G, denoted as dG (z), also d(z) when the graph G is obvious. Thus d(z) = |N (z)| and

1
 2

Figure 1.1: A 3-regular graph.

P
Sz = y∈N (z) d(y). A vertex z is said to be an isolated vertex if d(z) = 0 and pendant
vertex if d(z) = 1. We denote the maximum and minimum degrees of a graph G by
△(G) and δ(G), respectively. If for a graph G, δ(G) = △(G) = t, then G is called a
t-regular graph. Figure 1.1 shows a 3-regular graph. For a graph G of order n and
for any vertex z of G, the following inequalities hold:

0 ≤ δ(G) ≤ d(z) ≤ △(G) ≤ n − 1.

The following theorem relates the degrees of the vertices and the number of edges
of a graph G.

Theorem 1.2.1. [9] For any graph G

X
d(z) = 2m.
z∈V (G)

Corollary 1.2.1. In any graph G, there is an even number of vertices of odd degree.

For a graph G, a subset I of V (G) is called an independent set of G if every two

vertices of I are non-adjacent. The maximum number of vertices in I is called the
independence number of G, denoted by α(G). In figure 1.2 white vertices form the
independence set and these are maximum in numbers so α(G) = 4. The maximum
size of a set of pairwise adjacent vertices in a graph G is called the clique number of
G, written as ω(G). In figure 1.3 the pairwise adjacent vertices are shown in white
and these are maximum in numbers so ω(G) = 3.
A graph K is called a subgraph of a graph G, denoted K ⊆ G, if V (K) ⊆ V (G)
and E(K) ⊆ E(G). The subgraph K is said to be a proper subgraph of a graph
 3

Figure 1.2: A graph with α = 4.

Figure 1.3: A graph with ω=3.

G, if either V (K) is a proper subset of V (G) or E(K) is a proper subset of E(G)

or both. If V (K) = V (G) then the subgraph K is called spanning subgraph of the
graph G. A subgraph K is said to be an induced subgraph, if whenever y, z ∈ V (K)
and yz ∈ E(G), then yz ∈ E(K) as well. Figure 1.4 shows a graph and its induced
subgraph. Consider K be a nonempty set of vertices of a graph G, then the subgraph
of G induced by K is the induced subgraph with vertex set K. This type of subgraph
is denoted as ⟨K⟩. One can easily obtain a proper subgraph of a graph G by deleting
some vertices and all edges incident to these vertices. The graphs G − e and G − v are
the proper subgraphs of the graphs G, named as edge-deleted subgraph and vertex-
deleted subgraph, respectively.

1.3 Connectedness of Graphs

A walk (of length w) in a graph G is a sequence S with alternating terms of vertices
and edges of G (not necessarily different), S = z0 e1 z1 . . . zw−1 ew zw , such that ei =
zi−1 zi , 1 ≤ i ≤ w. If z0 = y and zw = z, then this walk S connects y to z and refer
 4

Figure 1.4: A graph and its induced subgraph.

Figure 1.5: A connected graph.

to S as an yz−walk. A walk S is said to be closed walk if z0 = zw , i.e. its terminal

vertices are identical. A walk is a trail if all its edge terms are different. If the vertices
in a walk S are all different, then this walk is called a path in G. A cycle is a path
with same initial and final vertex, i.e. z0 = zw . A path and a cycle of order n is
denoted by Pn and Cn , respectively.

Theorem 1.3.1. [10] If a graph G contains a yz-walk of length w, then G contains

a yz-path of length at most w.

Theorem 1.3.2. [9] Let G be a graph in which all vertices have degree at least two.
Then G contains a cycle.

A graph G is said to be connected if each pair of distinct vertices in G belongs to

a path, otherwise graph G is disconnected. Let K be a connected subgraph of the
graph G such that no other connected subgraph of the graph G properly contains
K, then the subgraph K is called a component of the graph G. So, we can say that
a graph G is connected if and only if it has exactly one component. Figures 1.5,1.6
shows a connected and a disconnected graph, respectively.
 5

Figure 1.6: A disconnected graph with 2 components.

Figure 1.7: A graph with radius 2 and diameter 3.

Proposition 1.3.1. [67] Every graph with n vertices and m edges has at least n − m
components.

Let G be a connected graph, the distance between two vertices y, z of G is the

shortest length of any y − z path in G and is denoted by dG (y, z) or simply d(y, z),
when the graph G is obvious. By definition, d(x, x) = 0. If there is no path containing
y and z, i.e., if y and z lie in different components of G, in this case d(y, z) is not
defined. For every two vertices y and z of a connected graph G of order n

0 ≤ d(y, z) ≤ n − 1.

For a vertex z in a connected graph G, the eccentricity of z, ecc(z), is the dis-

tance between z and a vertex farthest from z in G. The minimum and maximum
eccentricities among the vertices of the graph G are called the radius and diameter,
respectively, of the graph G. The graph in figure 1.7 has radius 2, diameter 3 and
ecc(z) = 2.
A vertex z in a connected graph G is a cut-vertex of the graph G if G − z is
disconnected. If the graph G has more than one component, then the vertex z is a
cut-vertex of G if z is a cut-vertex of a component of G. In figure 1.8 vertices a, b
and c represent the cut vertices. A connected graph G without cut-vertices is called
nonseparable graph. A maximal nonseparable subgraph of a graph G is called a block.
 6

a b

Figure 1.8: A graph with cut vertices.

1.4 Some Families of Graphs

A graph G is called a complete graph if any two vertices of G are adjacent, see figure
1.9. A complete graph of order n is denoted by Kn . It has maximum size, i.e., n(n−1)
2
.
A graph G is bipartite if its vertex set V (G) can be partitioned into two subsets Y
and Z such that every edge has one end in Y and one end in Z. If every vertex in Y
is joined to every vertex in Z, then the graph G is called a complete bipartite graph.
A complete bipartite graph with |Y | = a and |Z| = b is denoted by Ka,b . Figure 1.10
shows a complete bipartite graph.

Figure 1.9: A complete graph with 4 vertices.

Theorem 1.4.1. [10] A graph G is bipartite if and only if there is no odd cycle in G.

A graph G without cycles is an acyclic graph. A connected acyclic graph is known

as a tree. We can say that each component of an acyclic graph is a tree. For this
reason, acyclic graphs are called forests. A tree with n vertices is usually denoted as
Tn .
Paths are trees. A tree is a path if and only if △(T ) = 2. A star graph is a
tree consisting of one vertex adjacent to all other vertices. A star graph with order
 7

Figure 1.10: A complete bipartite graph.

n is denoted by Sn and it has a vertex z such that d(z) = n − 1. Figure 1.11 shows
a star graph. A star is a bipartite graph so it can also be represent as K1,n−1 . A
bistar of order n, BS(p, q), consists of two vertex disjoint stars, K1,p and K1,q , where
p + q = n − 2, and a new edge joining the centers of these stars, see figure 1.12. For
every n ≥ 2 and n/2 ≤ s ≤ n − 1, the spur [17] Sn,s is a tree consisting of 2s − n + 1
edges and n − s − 1 paths of length 2 having a common end vertex; in other words it
is obtained from a star K1,s by attaching a pendant edge to n − s − 1 pendant vertices
of K1,s . Figure 1.13 illustrates the spur.

Figure 1.11: S9

Theorem 1.4.2. [10] A graph G is a tree if and only if every two distinct vertices of
G are connected by a unique path.

Theorem 1.4.3. [5] The following statements are equivalent for a graph G:
i. G is a tree.
ii. G is a minimal connected graph, that is G is connected and if yz ∈ E(G) then
 8

Figure 1.12: B(7,7)

Figure 1.13: S14,9

G − yz is disconnected.
iii. G is a maximal acyclic graph, that is G is acyclic and if x and y are non-adjacent
vertices of G, then G + xy contains a cycle.

Corollary 1.4.1. [5] A tree of order at least 2 contains at least 2 pendant vertices.

Theorem 1.4.4. [9] Let G be a graph of order n and size m. If G satisfies any two
of the properties:
i. G is connected
ii. G is acyclic
iii. m = n − 1,
then G is a tree.

The cyclomatic number γ(G) of a graph G is the number of independent cycles

existing in G. For a connected graph G of order n and size m we have the relation
 9

γ(G) = m − n + 1. If G has p connected components, then γ(G) = m − n + p. A

graph G with one cycle is called a unicyclic graph and a graph with two independent
cycles called a bicyclic graph. Unicyclic and bicyclic graphs of order n has sizes n
and n + 1, respectively. If γ = 0, then G is a tree.
An isomorphism from a graph G = (V (G), E(G)) to a graph K = (V (K), E(K))
is a bijection h : V (G) → V (K) such that yz ∈ E(G) if and only if h(y)h(z) ∈ E(K).
We write G ∼ = K if there is an isomorphism from G to K and we say that G is
isomorphic to K. Clearly, isomorphic graphs have same order and size. Figure ??
shows two isomorphic graphs.

Theorem 1.4.5. [67] The isomorphism relation is an equivalence relation on the set
of graphs.

1.5 Chemical Graph Theory

Graph theory applied in the study of molecular structure represents an interdisci-
plinary science, called chemical graph theory or molecular topology. By using tools
taken from graph theory, set theory and statistics it attempt to identify structural
features involved in structure-property activity relationship. The partitioning of a
molecule and recombining its fragmental values by additive models is one of its main
tasks. Topological characterization of chemical structures allows the classification of
molecules and modeling unknown structures with desired properties.
A chemical graph is a model of a chemical system, used to characterize the interac-
tions among its components: atoms, bonds, groups of atoms or molecules. Chemical
graphs were first introduced in the latter half of the eighteenth century. Chemical
graphs are now being used for many different purposes in all the major branches of
chemistry. A structural formula of a chemical compound can be represented by a
molecular graph, its vertices being atoms and edges corresponding to covalent bonds.
Usually hydrogen atoms are not depicted in which case we speak of hydrogen depleted
molecular graphs. A hydrogen depleted molecular graphs are shown in Figures 1.14
and 1.15. Similarly, a transform of a molecule (e.g. a chemical reaction) can be vi-
sualized by a reaction graph, whose vertices are chemical species and edges reaction
pathways.
 10

Figure 1.14: 1-ethyl-2-methylcyclobutane and the corresponding hydrogen-depleted

molecular graph.

H H
H c H
c
H
c H
H H
H
c c c

H c H
H
H H

Figure 1.15: A molecular graph and its hydrogen depicted representation.

1.6 Topological Indices in Graph Theory

A parameter or an invariant of a graph G is a number associated with G which has
the same value for any graph isomorphic to G [7]. These invariants may help us
to distinguish non-isomorphic graphs. Some of very common graph invariants are
the number of vertices, edges, components, radius, diameter, independence number,
clique number, maximum or minimum degree etc.
A topological index is a numerical descriptor of the molecular structure derived
from the corresponding molecular graph. Many topological indices are widely used
 11

for quantitative structure-property relationship (QSPR) and quantitative structure-

activity relationship (QSAR) studies [13, 15, 45, 64].
“ The boiling points of organic compounds, as well as all their physical properties,
depend functionally upon the number, kind and structural arrangement of the atoms
in the molecule. Within a group of isomers, both the number and the kind of atoms
are constant, and variations in physical properties are due to changes in structureal
interrelationships alone. The boiling points of the paraffins are given by the linear
formula:
tB = aw + b · p + c
where a, b and c are constants for a given isomeric group, and p(Wiener polarity
index) and w(Wiener index) are structural variables (topological indices).”

Harry Wiener (1924-1998) was an American Chemist of Brooklyn College, and

made important and fundamental contributions to the study of topological indices
and established a correlation between the Wiener index and boiling points (hence
viscosity and surface tension) of the paraffins. In fact, he made the foundation of
chemical graph theory, and made a bridge between group theory, graph theory, and
chemistry.
He introduced the Wiener index; this index and its generalization, the Hyper-Wiener
index, have been used in predicting antibacterial activity in drugs to modeling crys-
talline phenomena. The Wiener index plays an important role in sociometry and the
theory of social networks which was developed by Christophe Soulé et al, as well as
Gromov-Hyperbolicity Parameters. The Wiener index also plays an important role
in Nanotechnology. https://en.wikipedia.org/wiki/Harry Wiener
One of the oldest and widely used indices is the Wiener index, named after
Harry Wiener, who introduced it in 1947 [34, 35]. It has been used to model var-
ious properties of chemical species [68]. The Wiener index of a graph G having
V (G) = {z1 , z2 , . . . , zn } is defined as the sum of distances between all unordered pairs
of vertices of a graph G, i.e.,
X
W (G) = dG (zi , zj ).
1≤i<j≤n

Klein et. al. [47] introduced the hyper-Wiener index , as a generalization of the
 12

Wiener index. It is defined as

1 1 X
W W (G) = W (G) + d2 (y, z).
2 2
{y,z}⊆V (G)

Many of mathematical properties of the hyper-Wiener index and its applications in

chemistry are discussed in [28, 47, 50]. Tomescu and Jamil introduced the general
hyper-Wiener index [62], denoted by W Wα (G), for any real α by
1 X
W Wα (G) = (d(y, z)α + d(y, z)2α ).
2
{y,z}⊆V (G)

Sharma, Goswami and Madan [58] introduced an adjacency-cum-distance based

topological index, called the eccentric connectivity index of the graph G, defined as
X
ζ c (G) = d(z)ecc(z).
z∈V (G)

The total eccentric connectivity index is defined as

X
ζ(G) = ecc(z).
z∈V (G)

Another distance based topological index is the Harary index. The Harary index
of a graph G, written as H(G), is defined as follows [40, 55]:
X 1
H(G) = .
1≤i<j≤n
dG (zi , zj )

Mathematical properties and applications of this index can be seen in [14, 22].
Iranmmanesh et. al. [1] introduced a modification of the Harary index, called
additively Harary index, defined by
X d(zi ) + d(zj )
HA (G) = .
1≤i<j≤n
dG (zi , zj )

The multiplicative Harary index was introduced by Su et. al. [59], which can be seen
as a product-degree-weight version of the Harary index:
X d(zi )d(zj )
HM (G) = .
d (z , z )
1≤i<j≤n G i j
 13

A large number of topological indices have been derived depending on vertex de-
grees. We discussed some of them. Randić (product-connectivity) index was proposed
by the chemist Milan Randić [56] in 1975. The Randić index R(G) of a graph G is
defined as X 1
R(G) = p .
yz∈E(G)
d(y)d(z)

Erdös and Bollobás [24] proved that the star minimizes the Randić index for all
graphs without isolated vertices and on a fixed number of vertices. Some properties
of the Randić index concerning diameter, radius, chromatic number, average distance
and eigenvalues of the adjacency matrix were discussed in [2, 3]. In 1998 Bollobás
and Erdös [6] replaced − 21 by any real number α to generalize this index, which is
known as the general Randić index. Yuan, Li and Hu [71, 72] characterized the trees
with extremal general Randić index. General Randić index is defined as:
X
Rα (G) = (d(y)d(z))α .
yz∈E(G)

The zeroth-order general Randić index, denoted by 0 Rα (G) was defined in [38] and
[46] as: X
0
Rα (G) = d(z)α .
z∈V (G)

For α = 2 this index is also known as first Zagreb index and denoted by M1 (G).
The first and second Zagreb indices are among the oldest and most famous topological
indices. These indices were introduced by Gutman and Trinajstić [37]. These are
defined as follows: X
M1 (G) = d(z)2 ,
z∈V (G)
X
M2 (G) = d(y)d(z).
yz∈E(G)

The first Zagreb index can also be expressed as:

X
M1 (G) = (d(y) + d(z)).
yz∈E(G)

For a historical background and properties of Zagreb indices see [8, 30, 54, 73, 74].
Recently, Todeschini et. al. [63, 65] proposed the multiplicative variants of ordi-
nary Zagreb indices, defined as:
 14

d(z)2 ,
Q Q
1 (G) =
z∈V (G)
Q Q
2 (G) = d(y)d(z).
yz∈E(G)

Various properties of multiplicative Zagreb indices were discussed in [25, 29, 44].
Narumi and Katayama [53] established an index by taking the product of degrees
of all vertices of the graph G, which is known as Narumi-Katayama index, as follows:
Y
N K(G) = d(z).
z∈V (G)

Clearly, 1 (G) = N K(G)2 . So, the graphs for which N K assumes an extremal value,
Q
Q
1 also assumes.
Miliçević et. al. [51] in 2004 reformulated the Zagreb indices by replacing vertex
degrees by edge degrees as
X
EM1 (G) = d(e)2 ,
e∈E(G)
X
EM2 (G) = d(e)d(f ),
e∼f

where d(e) is the degree of the edge e in G, defined by d(e) = d(y) + d(z) − 2 with
e = yz and e ∼ f represents that the edges e and f are adjacent.
Zhou et. al. [76] proposed the sum-connectivity index. The sum-connectivity
index is obtained from Randić index by replacing the term d(y)d(z) by d(y) + d(z).
X 1
χ(G) = p .
yz∈E(G)
d(y) + d(z)

Du et. al. [22] found the minimum sum-connectivity indices of trees and uni-
cyclic graphs. Some other properties can be seen in [70, 23]. The concept of sum-
connectivity index was extend to the general sum-connectivity index in [77]:
X
χα (G) = (d(y) + d(z))α ,
yz∈E(G)

where α ∈ R. Several properties of the general sum-connectivity index were given in

[19, 43, 61, 66].
Randić index has another variant, called harmonic index [26], defined as
X 2
H(G) = = 2χ−1 (G).
d(y) + d(z)
yz∈E(G)
 15

By the inequality between the arithmetic and geometric means we have inequality
H(G) ≤ R(G) and equality holds if and only if the graph G is a regular graph.

Question 1.6.1. Find the followings

1. M1 (Kn ) =

2. M2 (Kn ) =
Q
3. 1 (Ka,b ) =
Q
4. 2 (Ka,b ) =

5. ζ(P10 ) =

Question 1.6.2. Find the harmonic and Randić indices of the following graph

Figure 1.16: A grid type graph.

Solution. To find the the required topological indices accurately and efficiently we
need three tables. These tables includes the information about the order, size, degree
sequence and their counts, and edge partitioning based on the end degrees of vertices.
For graph shown in Figure 1.16, these tables are 1.1, 1.2 and 1.3.
From these tables we can find/verify the followings of the graph

i. number of vertices

ii. number of edges

 16

Order Size
3n 7 + (n − 2)5 = 5n − 3

Table 1.1: Order and size of the graph.

deg(v) Count/Frequency/Number of edges

2 4
3 2(n − 2) + 2 = 2(n − 2)
4 n−2

Table 1.2: Degree pattern and its count of the graph.

(deg(u), deg(v)) Count/Frequency/Number of edges

(2, 3) 8
(3, 3) 2(n − 3)
(3, 4) 2(n − 1)
(4, 4) n−3

Table 1.3: Edge partition and its count of the graph.

 17

iii. edge partition count

iv. topological indices (degree based)

v. M-polynomial

vi. Entropy

Now first we find the harmonic index of the given graph:

X 2
H(G) =
d(u) + d(v)
uv∈E(G)
 2   2   2   2 
=8 + 2(n − 3) + 2(n − 1) + (n − 3)
2+3 3+3 4+3 4+4
16 2(n − 3) 4(n − 1) (n − 3)
= + + +
5 3 7 4
Now, we work for the Randić index of the graph:
X 1
R(G) = p
uv∈E(G)
d(u) · d(v)
 1   1   1   1 
=8 √ + 2(n − 3) √ + 2(n − 1) √ + (n − 3) √
2·3 3·3 4·3 4·4
8 2(n − 3) (n − 1) (n − 3)
=√ + + √ +
6 3 3 4
■

1.6.1 Edge Weight Based Entropy of Graphs

The entropy of a probability distribution known as a measure of the unpredictability
of information content or a measure of the uncertainty of a system.” This quotation
was the foundation, described in [?] as a seminal theory for the idea of entropy.
Due to this concept has strongly based on statistical methodology, it became
well-known for chemical structures and their corresponding graphs. This parameter
provides a piece of extensive information about graphs, structures, and chemical
topologies. In 1955, the notion and its idea were used first time for graphs. In
sociology, ecology, biology, chemistry, and in a variety of other technical fields, graph-
based entropy or simply entropy has applications, [?, ?].
 18

Definition 1.6.1. In 2014 entropy for an edge weighted graph F is introduced in

[?]:
 
X
′ ′ ψ (u′ v′ )  ψ (u′ v′ ) 
Ωψ (F ) = − u v ∈ E (F ) X log  X . (1.6.1)
 
ψ (uv)  ψ (uv) 
uv∈E(F ) uv∈E(F )

Where ψ (uv) is a weight for an edge uv.

By letting the edge of weight equal to the main part of the topological index, [?, ?]
introduced the following entropies for an edge-weighted-based graph. The follow-
ings are some important formula formulas for this research work and all these are
based on the Equation (1.6.1).

Definition 1.6.2. The first and second Zagreb entropies are defined as following
[?, ?] ;
1 hY i
ΩM1 (G) = − log uv ∈ E (G) [d(u) + d(v)]d(u)+d(v)
M1 (G)
+ log (M1 (G)) . (1.6.2)

1 hY i
ΩM2 (G) = − log uv ∈ E (G) [d(u) · d(v)]d(u)·d(v)
M2 (G)
+ log (M2 (G)) . (1.6.3)

1.7 Mathematical Models for Prediction of Vari-

ous Physical-Chemical Properties using Topo-
logical Indices
ˆ Boiling point of alcohols (any of a class of organic compounds characterized by
one or more hydroxyl (—OH) groups attached to a carbon atom of an alkyl
group (hydrocarbon chain).)

ln(bp) = 4.279 + 0.181 ln (W )

 19

ˆ Boiling points of primary amines ( An amine in which the amino group is directly
bonded to one carbon of any hybridization which cannot be a carbonyl group
carbon.)
ln(bp) = 1.275 + ln (W ) − 0.006
• S. Gupta, M. Singh, A. K. Madan, Eccentric distance sum: A novel graph invariant for predicting biological and
physical properties , J. Math. Anal. Appl. 275 (2002) 386-401.

ˆ Boiling points of alkanes

bp = 181W 0.1775

• C. Adams, R. Frazosa, Introduction to topology pure and applied. Pearson

ˆ Toxicity (pc) of aliphatic ethers

pc = 0.061χ + 1.195

ˆ Boiling point of alkanes

bp = 67.47 (±1.71) R − 130.44 (±5.44)
bp = 67.560(±1.554)χ − 113.96 (±4.563)
• B. Lucic, I. Sovic, J. batista, K. Skala, D. Plavsic, D. Vikic, D. Beslo, S. Nikolic, N. Trinajstic, The sum-connectivity
index - An additive variant of the Randic̀ connectivity index, Current Computer-Aided Drug Design 9 (2013) 184-194.

ˆ Boiling point of 35 nonanes

bp = 94.23 (±6.68) + 25.58 (±3.97) R−1.25

ˆ Molar volume alkanes

mv = 55.85 (±2.10) + 16.53 (±0.32) R−0.07

• S. C. Basak, S. Nikolic, N. Trinajstic, D. Amic, D. Beslo, QSPR Modeling: Graph connectivity indices versus line
graph connectivity indices , J. Chem. Inf. Comput. Sci. 40 (2000) 927-933.

ˆ Cross sectional area of alcohols

ln csa = 4.8927 + 0.238 ln ξ c

ˆ Molar refractions of heterogenous compounds

ln mr = 1.4409 + 0.54 ln ξ c

• V. Sharma, R. Goswami, A. K. Madan, Eccentric connectivity index: A novel highly discriminating topological
descriptor for structure property and structure activity studies, J. Chem. Inf. Comput. Sci. 37 (1997) 273-282.
 20

1.8 Polynomials
Definition 1.8.1. Polynomials in two variables are algebraic expressions consisting
of terms in the form axn y m . The degree of each term in a polynomial in two variables
is the sum of the exponents in each term and the degree of the polynomial is the largest
such sum.

In mathematics, a graph polynomial is a graph invariant whose values are polyno-

mials. Invariants of this type are studied in algebraic graph theory. Important graph
polynomials include:

i. The characteristic polynomial, based on the graph’s adjacency matrix.

ii. The chromatic polynomial, a polynomial whose values at integer arguments give
the number of colorings of the graph with that many colors.

iii. The dichromatic polynomial, a 2-variable generalization of the chromatic poly-

nomial.

iv. The flow polynomial, a polynomial whose values at integer arguments give the
number of nowhere-zero flows with integer flow amounts modulo the argument.

v. The (inverse of the) Ihara zeta function, defined as a product of binomial terms
corresponding to certain closed walks in a graph.

vi. The Martin polynomial, used by Pierre Martin to study Euler tours

vii. The matching polynomials, several different polynomials defined as the generating
function of the matchings of a graph.

viii. The reliability polynomial, a polynomial that describes the probability of remain-
ing connected after independent edge failures

ix. The Tutte polynomial, a polynomial in two variables that can be defined (after
a small change of variables) as the generating function of the numbers of con-
nected components of induced subgraphs of the given graph, parameterized by
the number of vertices in the subgraph.

x. M-Polynomial, to determine closed form of many degree-based topological in-

dices.
 21

1.8.1 M-Polynomial of Graphs

For a graph G = (V, E), a degree-based topological index is a graph invariant of the
form X
I(G) = f (du , dv ) (1.8.1)
uv∈E(G)

where f = f (x, y) is a function appropriately selected for possible chemical applica-

tions [15]. For instance, the generalized Randić index Rα (G), α = 0, is defined with
1.8.1 by setting f (x, y) = xy [3]. Collecting edges with the same set of end-degrees
we can rewrite 1.8.1 as X
I(G) = mij (G)f (i, j).
i≤j

If G = (V, E) is a graph and v ∈ V , then dG (v) (or d(v) for short if G is clear
from the context) denotes the degree of v. Let G be a graph and let mij (G), i, j ≥ 1,
be the number of edges e = uv of G such that {dG (v), dG (u)} = {i, j}. As far as we
know, the quantities mij were first introduced and applied in [14]. In [16] authors
introduced the M -polynomial of G as
X
M (G; x, y) = mij (G)xi y j
i≤j

.
Followings are operators used to obtained topological index from the M -polynomial

Z x
∂f f (z, y)
Dx = x Sx = dz Qa = xa f (x, y)
∂x z
Z0 y
∂f f (x, z)
Dy = x Sy = dz J = f (x, x) (1.8.2)
∂y 0 z
In the following table 1.4, the topological index and the derivation formula is given

Question 1.8.1. Find the M-polynomial of the graph given in Figure 1.16 and using
the polynomial verify the values of the harmonic and Randić indices found in the
question 1.6.2.
 22

Topological index Derivation from M (G; x, y)

First Zagreb (Dx + Dy ) (M (ℵ; x, y))
Second Zagreb (Dx Dy ) (M (ℵ; x, y))
Second modified Zagreb (Sx Sy ) (M (ℵ; x, y))


General Randić Dxα Dyα (M (ℵ; x, y))


General inverse Randić Sxα Syα (M (ℵ; x, y))
Symmetric division index (Dx Sy + Dy Sx ) (M (ℵ; x, y))
Harmonic index 2Sx J (M (ℵ; x, y))
Inverse sum index Sx JDx Dy (M (ℵ; x, y))
Augmented Zagreb Sx3 Q−2 JDx3 Dy3 (M (ℵ; x, y))

Table 1.4: M-polynomials

Solution. By using the definition of the M-polynomial and the table 1.3, we have the
following
X
M (G; x, y) = mij (G)xi y j
i≤j

= 8 · x2 y 3 + 2(n − 3)x3 y 3 + 2(n − 1)x3 y 4 + (n − 3)x4 y 4

 23

For verification of the harmonic index

X 
i j
H(G) = 2Sx J(M (G; x, y)) = 2Sx J mij (G)x y
i≤j
 
2 3 3 3 3 4 4 4
= 2Sx J 8 · x y + 2(n − 3)x y + 2(n − 1)x y + (n − 3)x y
 
= 2Sx 8x2 x3 + 2(n − 3)x3 x3 + 2(n − 1)x3 x4 + (n − 3)x4 x4
 
= 2Sx 8x5 + 2(n − 3)x6 + 2(n − 1)x7 + (n − 3)x8
 z6 z7 z8 z 9 x
= 2 8 + 2(n − 3) + 2(n − 1) + (n − 3)
5 6 7 8 0

 x6 x7 x8 x9 
= 2 8 + 2(n − 3) + 2(n − 1) + (n − 3)
5 6 7 8

Now, by putting x = 1, we get the result

16 2(n − 3) 4(n − 1) (n − 3)
H(G) = + + + .
5 3 7 4

Similarly, for the Randić index we will do the steps.

X 
R(G) = Dxα Dyα (M (G; x, y)) = Dxα Dyα mij (G)xi y j
i≤j
 
= Dxα Dyα 8 · x2 y 3 + 2(n − 3)x3 y 3 + 2(n − 1)x3 y 4 + (n − 3)x4 y 4

= 8 · 2α · 3α x2 y 3 + 2(n − 3)3α · 3α x3 y 3 + 2(n − 1)3α · 4α x3 y 4

+ (n − 3)4α · 4α x4 y 4
8 2(n − 3) (n − 1) (n − 3)
=√ + + √ +
6 3 3 4

■
P
Theorem 1.8.1. Let G = (V, E) be a graph and let I(G) = uv∈E(G) f (d(u), d(v),
where f (x, y) is a polynomial in x and y. Then

I(G) = f (Dx , Dy )(M (G : x, y))|x=y=1 .

 24

Proof. Let r ≥ 0 and s ≥ 0 be fixed. Then

X  X 
Dxr Dys (M (G : x, y)) = Dxr Dys mij (G)xi y j = Dxr mij (G)j s xi y j
i≤j i≤j
X
= mij (G)ir j s xi y j
i≤j

αr,s xr y s . Then f (Dx , Dy )(M (G; x, y)) is equal to

P
Suppose that f (x, y) = r,s

X X X X X
Dxr Dys (M (G; x, y)) = αr,s mij (G)ir j s xi y j = mij (G) αr,s ir j s xi y j
r,s r,s i≤j i≤j r,s

it follows that
X X X
f (Dx , DY )(M (G; x, y))|x=y=1 = mij (G) αr,s er j s = mij (G)f (i, j) = I(G)
i≤j r,s i≤j

Hence, the result.

Exercise 1.8.1. Find the first and second Zagreb indices and inverse sum index by
using the M-polynomial of the hexagonal graph given in Figure 1.17

Figure 1.17: Hexagonal graph.

Exercise 1.8.2. Investigate the M-polynomial of the following graph. Further using
the M-polynomial find the Augmented Zagreb and second modified Zagreb and sym-
metric division idices of the graph.
 25

Figure 1.18: Triangular ladder.

1.9 Graph Operations

In this section, we discuss some graph operations. We can construct many interesting
classes of graphs from simpler graphs by using graph operations also known as graph
products [41].
The cartesian product, G1 □G2 , of graphs G1 and G2 has the vertex set V (G1 □G2 ) =
V (G1 ) × V (G2 ) and (y1 , y2 )(z1 , z2 ) is an edge of G1 □G2 if y1 = z1 and y2 z2 ∈
E(G2 ), or y1 z1 ∈ E(G1 ) and y2 = z2 . If G1 , G2 , . . . , Gn are graphs, then we de-
n n
note G1 □G2 □ · · · □Gn by □ Gi . If G1 = G2 = · · · = Gn = G, we denote □ Gi by
1 1
Gn .
The composition G1 [G2 ] (or lexicographic product) of graphs G1 and G2 with
disjoint vertex sets and edge sets is again a graph on vertex set V (G1 ) × V (G2 )
in which (y1 , y2 ) is adjacent with (z1 , z2 ) whenever y1 z1 ∈ E(G1 ) or y1 = z1 and
y2 z2 ∈ E(G2 ).
The disjunction G1 ∨ G2 of graphs G1 and G2 is the graph with vertex set V (G1 ) ×
V (G2 ) and (y1 , z1 ) is adjacent with (y2 , z2 ) whenever y1 y2 ∈ E(G1 ) or z1 z2 ∈ E(G2 ).
The symmetric difference G1 ⊕ G2 of two graphs G1 and G2 is the graph with
vertex set V (G1 ) × V (G2 ) and E(G1 ⊕ G2 ) = {(y1 , y2 )(z1 , z2 )|y1 z1 ∈ E(G1 ) or y2 z2 ∈
E(G2 ) but not both}.
The direct product of two graphs G1 and G2 is the graph G1 ×G2 with V (G1 ×G2 ) =
V (G1 ) × V (G2 ) and the vertices (y1 , y2 ) and (z1 , z2 ) are adjacent if y1 z1 ∈ E(G1 ) and
y2 z2 ∈ E(G2 )
Let G1 and G2 be two graphs on disjoint vertex sets. Their sum is the graph
G1 + G2 on the vertex set V (G1 ) ∪ V (G2 ) and the edge set E(G1 + G2 ) = E(G1 ) ∪
E(G2 ) ∪ {yz : y ∈ V (G1 ), z ∈ V (G2 )}. The sum of two graphs is also known as join.
For G1 and G2 we define their corona product G1 ◦ G2 as the graph obtained
 26

by taking |V (G1 )| copies of G2 and joining each vertex of the i-th copy with vertex
yi ∈ V (G1 ).
Suppose that G1 and G2 are graphs with disjoint vertex sets. For given vertices
y ∈ V (G1 ) and z ∈ V (G2 ) a splice (G1 .G2 ; y, z) of G1 and G2 by vertices y and z is
defined by identifying the vertices y and z in the union of G1 and G2 [11]. Similarly, a
link (G1 ∼ G2 ; y, z) of G1 and G2 by vertices y and z is defined as the graph obtained
by joining y and z by an edge in the union of these graphs.
Let Gi , 1 ≤ i ≤ n, be some graphs and yi ∈ V (Gi ). A chain graph is obtained
from the union of the graphs Gi , i = 1, . . . , n, by adding the edges yi yi+1 , where
yi ∈ V (Gi ) for 1 ≤ i ≤ n − 1, and denoted by G = G(G1 , . . . , Gn ; y1 , . . . , yn ). Note
that G(G1 , G2 ; y1 , y2 ) ∼
= (G1 ∼ G2 ; y1 , y2 ).
Sagan et. al. [60] computed some exact formulas for the Wiener polynomial of
various graph operations. In [48, 49] Ashrafi et. al. computed index PI and Zagreb
indices respectively, of some graph operations.
Munarini et. al. [52] defined the double graph of a simple graph denoted as D[G].
The double graph of a simple graph G can be build up by taking two distinct copies
of the graph G and joining every vertex y in one copy to every vertex z ′ in the other
copy corresponding to a vertex z adjacent to y in the first copy.
By adding a loop to every vertex of K2 we obtained the graph K2s . The double
graph of a simple graph G can be expressed as D[G] = G□K2s . Since the direct
product of a simple graph with any graph is always a simple graph, it follows that
the double of a simple graph is still a simple graph. Some of its elementary properties
are discussed in [52]. If G has n vertices and m edges then D[G] has 2n vertices and
4m edges. For an illustration see Fig 1.19.

x3 x5
x1
x2
G
x4
x6

y5
y3
G'
y1 y2
y6
y4
G D[G]

Figure 1.19: A graph G and its double graph D[G].

Chapter 2

Extremal Graph Theory w.r.t

Topological Indices

2.1 Extremal Graphs for Some Topological Indices

Question 2.1.1. Investigate the behavior of the first Zagreb when an edge is added
between two non adjacent vertices of G.

Solution. Let G be a graph such that uv ∈ / E(G). Now we construct a new graph
G from G by adding edge between u and v, i.e. uv ∈ E(G∗ ). By inserting edge
∗

between the degree of u and v increases by one, and the degrees of all other vertices
remain same, i.e. dG∗ (v) = dG (v) + 1, dG∗ (u) = dG (u) + 1, and dG∗ (x) = dG (x) when
u ̸= x ̸= v. Now we investigate the behavior of M1 ,
X X
M1 (G) − M1 (G∗ ) = dG (y 2 ) − dG∗ (y 2 )
y∈V (G) y∈V (G∗ )
X X
= dG (y 2 ) + dG (u)2 + dG (u)2 − dG (y 2 ) − dG∗ (u)2 − dG∗ (u)2
y∈V (G),u̸=y̸=v y∈V (G),u̸=y̸=v

= dG (u) + dG (u) − (dG (u) + 1) − (dG (u) + 1)2 < 0

2 2 2

From above we have M1 (G) < M1 (G∗ ), this shows that adding an edge between two
non adjacent vertices increases the first Zagreb index. ■

Exercise 2.1.1. Check the above question for the second Zagreb, inverse degree sum
and Randić indices.

Think: What question 2.1.1 tells about the extremal graphs w.r.t the first Zagreb
index???

27
 28

From question 2.1.1, we get that by increasing the edges the first Zagreb index
increases and as we remove an edge between two vertices decreases the first Zagreb
index. From this we can notice that a graph with maximum number of edges gives
the maximum first Zagreb index and a graph with minimum number of edges gives
the minimum first Zagreb index, i.e.

M1 (T ) ≤ M1 (G) ≤ M1 (Kn ).

where T is a tree on n vertices.

But there are many trees on n vertices, so we need to find the extremal trees w.r.t
topological indices. We will discuss a little later.
Question 2.1.2. Check the behavior of the distances between two vertices of G when
we add an edge between two non adjacent vertices u and v of G.
Solution. Clearly, adding an edge between u and v reduces the distance at least
between u and v. In general, we can say adding an edge may reduce the distance
between two vertices of G. ■

2.1.1 Extremal Trees

Theorem 2.1.1. Among trees of order n ≥ 5, Pn has the minimum second Zagreb
index, i.e.
M2 (Pn ) ≤ 2 + (n − 3)4 = 4n − 10.
Proof. Suppose that T is a tree, but not a path, which has the minimum value
of the second Zagreb index. We will probe by contradiction. Suppose that P =
v1 v2 · · · vk−1 vk is a longest path of T . Then d(v1 ) = d(vk ) = 1. Considering the
degrees of v2 and vK−1 , we have following two cases:
Case I. d(v2 ) > 2 or d(vk−1 ) > 2.
Without loss of generality, we assume that d(vk−1 ) = d > 2. Since P is a longest
path of T , all the vertices adjacent to vk−1 m other than Vk−2 , must be leaves. Let
u1 , u2 , · · · , ud−2 be the neighbors of vk−1 , other than vk−2 and vk . By deleting the
edges vk−1 u1 , vk−1 , · · · , vk−1 ud−2 and adding the edges vk u1 , vk u2 , · · · , vk ud−2 , we get
a new tree T ′ , as shown in Figure 1. If d(vk−1 ) = 1, T is a star Sn . For n ≥ 5, we
have

M2 (Sn ) − M2 (Pn ) = (n − 1) · ((n − 1) · 1) − 2 · (2 · 1) − (n − 3) · (2 · 2)

= (n − 1)2 − 22 − (n − 3)4
= n2 − 2n + 1 − 4 − 4n + 12 > 0
= n2 − 6n + 9
 29

Now we can assume that d(vk−1 ) = a ≥ 2. So we have

M2 (T ) − M2 (T ′ ) = a · d + (d − 2) · (d · 1) + d · 1 − a · 2 − 2 · (d − 1) − (d − 2) · ((d − 1) · 1)
= ad + d2 − 2d + d − 2a − 2d + 2 − d2 + 3d − 2
= ad − 2a
= a (d − 2) > 0

Case 2. d(v2 ) = d(vk−1 ) = 2.

Since the tree T is not a path, there exist a vertex vi such that d(vi ) = d ≥ 3 for
some i = 3, 4, · · · k − 2. Let u1 , u2 , · · · ud−2 be the neighbors of vi , other than vi−1 and
vi+1 . By deleting edges u1 vi , u2 vi , · · · , ud vi and adding edges u1 vk , u2 vk , · · · , ud−2 vk ,
we get a new tree T ′ as shown in Figure. Suppose d(vi−1 ) = a, d(vi+1 ) = b.

M2 (T ) − M2 (T ′ ) = a · d + (d − 2) · (d · 1) + (d · b) + 2 · 1
− a · 2 − b · 2 − 2 · (d − 1) − (d − 2) · ((d − 1) · 1)
= ad + d2 − 2d + db + 2 − 2a − 2b − 2d + 2 − d2 + 3d − 2
= ad + bd − d − 2a − 2b + 2
= d (a + b − 1) − 2(a + b − 1) > 0

Therefore, in both cases we get a tree T ′ with smaller value of M2 , which contradicts
to the choice of T , and the proof is complete.

Exercise 2.1.2. Investigate the tree(s) with the minimum first Zagreb index, forgotten
topological index and for the first reformulated Zagreb index.

Theorem 2.1.2. Let T (n) be a set of tree on n vertices, then the star graph maximize
the second Zagreb index among the set of trees of order n.

Proof. Let T be a tree but not a star with maximum second Zagreb index and v1 , v2
and v3 be three vertices of T such that v1 v2 ∈ E(T ) and v2 v3 ∈ E(T ). Sup-
pose that d(v1 ) = i, d(v2 ) = j and d(V3 ) = q, where i, q ≥ 2. Let N (v1 ) − v2 =
{w1 , w2 . · · · , wq−1 }. By deleting the edges v3 w1 , v3 w2 , · · · v3 wq−1 and adding the new
edges v1 w1 , v1 w2 , · · · v1 wq−1 , we get a new tree T ′ , as shown in Figure.
 30

Now comparing the values of the second Zagreb index of T and T ′ , we have
i−1
X i−1
X
′
M2 (T ) − M2 (T ) = dT (ua ) · dT (v1 ) + dT (v1 ) · dT (v2 ) + dT (v2 ) · dT (v3 ) + dT (wb ) · dT (v3 )
a=1 a=1
i−1 q−1
X X
− dT ′ (ua ) · dT ′ (v1 ) − dT ′ (wb ) · dT ′ (v1 ) − dT ′ (v1 ) · dT ′ (v2 ) − dT ′ (v2 ) · dT ′ (v3 )
a=1 b=1
i−1 q−1
X X
=i dT (ua ) + i · j + j · q + q dT (wb )
a=1 b=1
i+q−1 q−1
X X
(i + q − 1) dT ′ (ua ) − (i + q − 1) dT ′ (wb ) − (i + q − 1) · j − j · 1 < 0.
a=1 b=1

It contradicts the fact that T has the maximum second Zagreb index.

Exercise 2.1.3. From the set T (n), investigate the tree(s) with the maximum first
and second Zagreb indices, forgotten index, hyper Zagreb index and the inverse degree
sum index.

Transformation 1. Suppose that G is a connected graph and A = y1 y2 · · · ya

and B = z1 z2 · · · zb be two paths of lengths a and b, respectively. Let v be a vertex in
G, then H and K are two graphs derived from G such that H = G + {vy1 y2 · · · ya } +
{vz1 z2 · · · zb } and K = G + {vy1 y2 · · · ya } + {ya z1 z2 · · · zb }, respectively. The graphs
H and K are shown in Fig. ??.

Lemma 2.1.1. Let H and K be the above defined graphs, then for γ > 0 we have

(i) γ1 (K) > γ1 (H),

Q Q

(ii) γ2 (K) < γ2 (H).

Q Q

Proof. Suppose that dG (v) = p ≥ 1. Then by the definition of γ1 we have

Q

Qγ
(K) (p + 1)γ · 2γ(a+b−1) (p + 1)γ · 2γ
Q1γ = = >1
1 (H) (p + 2)γ · 2γ(a+b−2) (p + 2)γ

Similarly, from the definition of γ2 and Lemma ?? we have

Q

Qγ
(K) (p + 1)γ(p+1) · 2γ2(a+b−1) (p + 1)γ(p+1) · 4γ  16 γ
Q2γ = = ≤ < 1.
2 (H) (p + 2)γ(p+2) · 2γ2(a+b−2) (p + 2)γ(p+2) 27

Hence, (i) and (ii) are true.

 31

Transformation 2. Let G be a connected graph. Suppose that uv, uzi are edges
of G, where 1 ≤ i ≤ a, such that dG (v) = 2 and dG (zi ) = 1 for each i. We derived a
graph H from G as H = G − uzi + vzi for 1 ≤ i ≤ a. The construction of H from G
is shown in Fig. ??.

Lemma 2.1.2. Let G and H be two graphs as shown in Fig. ??. For γ > 0, we have

(i) γ1 (H) < γ1 (G),

Q Q

(ii) γ1 (H) > γ1 (G).

Q Q

Proof. Assume that dG (v) = b + 1 ≥ 2, then dH (v) = a + b + 1, where a, b ≥ 1. Now

from the definitions of general multiplicative Zagreb indices we have
Qγ
(G) (b + 1)γ (a + 1)γ  ab + a + b + 1 γ
Qγ1 = = > 1,
1 (H) (a + b + 1)γ a+b+1
Qγ
(G) (b + 1)γ(b+1) (a + 1)γ(a+1)  22 (a + 1)a+1 γ
Qγ2 = ≤ < 1,
2 (H) (a + b + 1)γ(a+b+1) (a + 2)(a+2)

the last inequality in the second expression is due to the Lemma ??.

Transformation 3. Let G be a connected graph and u, v be two vertices of G.

Also assume that uyi , vzj , where 1 ≤ i ≤ a and 1 ≤ j ≤ b, are edges in G with
dG (yi ) = 1 = dG (zj ) for each i and j. Now we derive two graphs H and K from G as:
H = G − uyi + vyi for 1 ≤ i ≤ a and K = G − vzj + uzj for 1 ≤ j ≤ b. The graphs
H and K are shown in Fig. ??.

Lemma 2.1.3. Let G, H and K are the graphs discussed in Transformation 3. For
γ > 0, we have

(i) either γ1 (G) > γ1 (H) or γ1 (G) > γ1 (K),

Q Q Q Q

(ii) either γ2 (G) < γ1 (H) or γ2 (G) < γ1 (K).

Q Q Q Q

Proof. Assume that dG (v) = c + a and dG (u) = d + b, then for γ > 0 we have
Qγ
1 (G) (c + b)γ (d + a)γ  cd + db + ac + ab γ
Qγ = γ = > 1, if c ≥ d,
1 (H) d (a + b + c)γ cd + ad + bd
Qγ
(G) (c + b)γ (a + d)γ  cd + ac + db + ab γ
Qγ1 = γ = > 1, if c < d.
1 (K) c (a + b + d)γ cd + ac + bc

Now we will prove (ii).

 32

Let f (x) = xhγx and ϕ(x)

i = logf (x) − logf (p) − logf (x − p + 1), where 1 < p < x.
Then ϕ′ (x) = γ log x−p+1
x
> 0 since γ > 0. Hence, ϕ(x) is increasing for x > 0. Also,
′
ϕ (p) < ϕ(x), which is equivalent toQf (x) > f (p)f (x − p + 1).
From the definitions of 1 and γ2 and construction of graphs H and K, we have
Qγ

γ γ
Y Y
(H) − (G) ≥ (c + a + b)γ(c+a+b) − (c + a)γ(c+a) − (b + d)γ(b+d)
2 2
γ γ
Y Y
(K) − (G) ≥ (d + a + b)γ(d+a+b) − (c + a)γ(c+a) − (b + d)γ(b+d)
2 2

So,
γ γ γ
Y Y Y
(H) + (K) − 2 (G) ≥ (c + a + b)γ(c+a+b) − 2(c + a)γ(c+a) −
2 2 2

2(b + d)γ(b+d) + (d + a + b)γ(d+a+b)

> (c + a)γ(c+a) (b + 1)γ(b+1) + (d + b)γ(d+b) (a + 1)γ(a+1)
− 2(c + a)γ(c+a) − 2(b + d)γ(b+d) > 0.

which implies that γ2 (G) < γ1 (H) or γ2 (G) < γ1 (K).

Q Q Q Q

Transformation 4. Let P = y1 y2 · · · ya be a pendant path of G such that z1 and

z2 are two neighbors of y1 other than y2 . We derive a new graph H = G − z1 y1 + z1 ya .
This transformation is illustrated in Fig. ??.
Lemma 2.1.4. Let G and H be two graphs from the Fig. ??. For γ > 0, we have
(i) γ1 (G) < γ1 (H),
Q Q

(ii) γ2 (G) > γ2 (H).

Q Q

Proof. From the definitions of general multiplicative Zagreb indices, we have

Qγ
(G) 3γ
Qγ1 = γ γ <1
(H) 2 ·2
Q1γ
(G) 33γ
Qγ2 = 2γ 2γ > 1
2 (H) 2 ·2

Hence, we complete the proof of Lemma 2.1.4.

 33

2.1.2 Extremal (n, m)-graphs w.r.t general multiplicative Za-

greb indices
In this subsection, we present the extremal (n, m)-graphs for general multiplicative
Zagreb indices when γ > 0, here we consider only the graphs for m = n − 1 or m = n.
Theorem 2.1.3. Let T be any (n, n − 1)-graph and n ≥ 5, then for γ > 0 we have
(i) γ1 (Sn ) < γ1 (T ) < γ1 (Pn ),
Q Q Q

Qγ Qγ Qγ
(ii) 2 (Pn ) < 2 (T ) < 2 (Sn ).

Proof. Let T be a tree of order n and T ∗ be another tree of order a, attached to a

vertex of T . By repeating the Transformation 1, we can convert the T ∗ into a path
Pa+1 and Lemma 2.1.1 guarantees that during this process the first multiplicative Za-
greb index increases, while the second general multiplicative Zagreb index decreases.
Those implies that the first and second general multiplicative Zagreb index are max-
imum and minimum for Pn , respectively.
Now, by repeating the Transformation 2 on T ∗ we obtained Sa+1 as an attached
tree. Lemma 2.1.2 ensures that applying the Transformation 2 on T decreases the first
general multiplicative Zagreb index and increases the second general multiplicative
Zagreb index, respectively. This implies that Sn attains the minimum for the first gen-
eral multiplicative Zagreb index and maximum for the second general multiplicative
Zagreb index.

Let Cna be a unicyclic graph of order n obtained by attaching n−a pendant vertices
to a vertex of Ca .
Theorem 2.1.4. Let G be an (n, n)-graph of order n and girth a. Then for γ > 0,
we have
(i) γ1 (Cn3 ) ≥ γ1 (G),
Q Q

Qγ 3
Qγ
(ii) 2 (Cn ) ≤ 2 (G).

Both equalities hold when G = Cn3 .

Proof. Let G be an (n, n)-graph of order n and girth a. One can notice that by
repeating Transformations 2 and 3, G can be converted into a unicyclic graph such
that all the pendant vertices are adjacent to a unique vertex. Lemmas 2.1.2
Qγ and 2.1.3
imply
Qthat during the process ofQapplying theQ Transformations 2 and
Q 3, 1 decreases
and γ2 increases. So, we have γ1 (Cna ) ≥ γ1 (G) and γ2 (Cna ) ≤ γ2 (G).
Q
 34

Moreover, we have
Qγ a
(C ) (n − l + 2)γ · 2γ(a−1)
Q1γ n3 = > 1,
1 (Cn ) (n − 1)γ · 4γ

since the function f (x) = (n − x + 2)γ · 2γ(x−1) isQan increasing function as f ′ (x) =
γ
−1 (C n−1 ) n−4
2γ(x−1) γ n−x+2 + log(2) > 0 for 3 < x < n − 1. Q1 γ (Cn 3 ) = 3·2
n−1
> 1 when n ≥ 5.
1 n

Qγ a
(C ) (n − l + 2)γ(n−l+2) · 2γ(a−1)
Q2γ n3 = < 1,
2 (Cn ) (n − 1)γ(n−1) · 16γ
γ(n−x+2)
since the function g(x) = (n − x +2) · 4γ(x−1)
 is a decreasing function as
′ γ(x−1) γ(n−x+2) 4
g (x) = 4 γ n − x + 2) − 1 + log n−x+2 < 0 for 3 < x < n. Hence,
both (i) and (ii) are true.

Theorem 2.1.5. Let G be any (n, n)-graph of order n and γ > 0, then
(i) γ1 (G) ≤ γ1 (Cn ),
Q Q

Qγ Qγ
(ii) 2 (G) ≥ 2 (Cn ).

Proof. Transformation 1 can convert any unicyclic graph G into a sun graph and
Lemma 2.1.1 ensures that every time applying the Transformation 1 produces a uni-
cyclic graph with greater first general multiplicative Zagreb index. Applying repeat-
edly Transformation 4 on the sun graph eventually gives Cn and Lemma 2.1.4 implies
that Cn attains the maximum first general multiplicative Zagreb index.
(ii) can be similarly proved.

Theorem 2.1.4 and 2.1.5 completely characterize the extremal graphs among (n, n)-
graphs.

2.1.3 Extremal (n, n + 1)-graphs w.r.t general multiplicative

Zagreb indices
In this subsection we first introduce some notations to be used in this section.
For any (n, n + 1)-graph, there are at least two cycles in G. Depending on the
nature of the two cycles we have following cases:
(1) the two cycles Ca and Cb have only one common vertex, such a graph will be
denoted by Ca,b ;
 35

(2) the two cycles Ca and Cb are connected by a path of length c > 0, such a graph
will be denoted by Ca,c,b ;

(3) the two cycles Ca and Cb have a common path of length c > 0, such a graph will
c
be denoted by Ca,b .
c
Ca,b , Ca,c,b and Ca,b will be considered as the main subgraphs of G.
∗
We use Bn to denote a graph obtained by joining two pairs of pendant vertices
among three selected pendant vertices in Sn . Graphs Ca,b , Ca,c,b and Ca,b c
and Bn∗ are
shown in the Fig. ??.
The following transformation can be easily deduce from Transformations 2 and 4.
Transformation 5. Let P = xz1 z2 · · · za y be an internal path of G and dG (x), dG (y) ≥
2. We construct a graph H = G−{z2 z3 , z3 z4 , · · · , za−1 za , za zy }+{z1 z3 , z1 z4 , · · · , z1 za , z1 zy }
as shown in Fig. ??.

Lemma 2.1.5. Let G and H be two graphs from Fig. ?? and γ > 0, then

(i) γ1 (G) > γ1 (H),

Q Q

Qγ Qγ
(ii) 2 (G) < 1 (H).

Proof. (i) and (ii) follows immediately from Lemmas 2.1.2 and 2.1.4.

Theorem 2.1.6. Let G be an (n, n + 1)-graph and n ≥ 4. For γ > 0, we have

(i) γ1 (Bn∗ ) < γ1 (G),

Q Q

(ii) γ2 (Bn∗ ) > γ2 (G).

Q Q

Proof. For n = 4 and n = 5, any (n; n + 1)-graph contains one and five graphs,
respectively, and results can be seen easily. In the following we will consider the case
of n ≥ 6.
(i). Let G be an (n, n+1)-graph with minimum first general multiplicative Zagreb
index having BG as its main subgraph. Then BG is one of types (1), (2) or (3). By
Transformations 2 and 3, we have that G must be a graph with some pendant vertices
adjacent to one vertex of the subgraph BG .
Claim. Length of any cycle in BG is at most 4.
Proof. Suppose, to the contrary, that BG has a cycle of length 5. If BG is of type
(1) or (2), then by applying Transformation 5 and Lemma 2.1.5 we can construct a
′
new graph BG with smaller first general multiplicative Zagreb index contradicting to
the minimality of the first general multiplicative Zagreb index of G.
 36

c
Now, consider that G is of type (3), i.e., BG = Ca,b with 1 ≤ a ≤ min{a, b} and
a+b > 5. This implies that one of two integers a and b is not less than 3, so G contains
an internal path of length greater or equal to 2 in BG . By applying Q Lemmas Q 2.1.2
and 2.1.5, we can convert G into a bicyclic graph G∗ such that γ1 (G∗ ) < γ1 (G).
Hence, again we have a contradiction.
So the length of any cycle in BG is 3 or 4. Since we have BG = C3,3 or BG = C3,a,3
a
or BG = C2,2 . For BG = C3,a,3 , we must have a = 1 otherwise Lemmas 2.1.2 and
2.1.4 ensure that we can construct a new bicyclic graph of order n with smaller first
general multiplicative Zagreb index.
Now we construct some new bicyclic graphs with the existing ones.
∗
1. C3,3 obtained from C3,3 by attaching n−5 pendant vertices to a vertex of degree
2,
∗∗
2. C3,3 obtained from C3,3 by attaching n−5 pendant vertices to a vertex of degree
4,
∗
3. C3,1,3 obtained from C3,1,3 by attaching n − 6 pendant vertices to a vertex of
degree 2,
∗∗
4. C3,1,3 obtained from C3,1,3 by attaching n − 6 pendant vertices to a vertex of
degree 3,
1∗ 1
5. C3,3 obtained from C3,3 by attaching n−4 pendant vertices to a vertex of degree
2.

The values Qγ of ∗the first general multiplicative

Qγ ∗∗ Zagreb indices of
Qγthose graphs are as
γ γ γ ∗
follows:Q 1 (C3,3 ) = (n − 3) · 8 · 4 , 1 (C = (n − 1) · 16 , 1 (C3,1,3 ) = (n − 4)γ ·
Qγ3,3 ) 1∗
γ γ
γ ∗∗
8Q · 9 , 1 (C3,1,3 ) = (n − 3) · 3 · 16 , 1 (C2,2 ) = (n − 2)γ · 9γ · 2γ . Also, we have
γ γ γ γ γ
γ ∗ γ γ γ
1 (Bn ) = (n − 1) · 3 · 4 .
Obviously, (i) is true.
Now, we consider forQ(ii). From the definition of the second Qγ general multiplicative
γ ∗ γ(n−3) ∗∗
Zagreb index, we have 2 (C3,3 ) = (n − 3) · 2 · 4 , 2 (C3,3 ) = (n − 1)γ(n−1) ·
6γ 4γ
γ γ
28γ , 2 (C3,1,3∗
) = (n−4)γ(n−4) ·26γ ·36γ , 2 (C3,1,3 ∗∗
) = (n−3)γ(n−3) ·28γ ·33γ , γ2 (C2,2
1∗
Q Q Q
)=
γ(n−2) 3γ 2γ
Q γ ∗ γ(n−1) 2γ 3γ
(n − 2) · 3 · 2 . Also, we have 2 (Bn ) = (n − 1) ·2 ·3 .
(n−3)(n−3)
The function ϕ(x) = (n−1)(n−1) is a decreasing function as ϕ′ (x) = (n − 3)(n−3) (n −
h i
n−3
1)(1−n) log n−1 < 0. This completes the proof of Theorem 10:
 37

Qγ ∗  (n − 3)(n−3) · 256 γ
(C ) 33 · 256
Q2γ 3,3 = ≤ < 1,
(B ∗ ) (n − 1)(n−1) · 27 55 · 27
Qγ 2 ∗ n 4
(C3,3 ) (n − 3)(n−3)·4
Qγ2 ∗∗ = > 1,
2 (C3,1,3 ) (n − 4)(n−4) · 36
Qγ ∗
(Bn ) (n − 1)(n−1) · 27
Qγ2 ∗∗ = > 1,
(C3,3 ) (n − 1)(n−1) · 26
Q2γ 1∗
(C ) (n − 2)(n−2) · 27 44 · 27
Q2γ 2,2 = ≤ < 1.
(B ∗ ) (n − 1)(n−1) · 4 55 · 4
Qγ 2 ∗ n
(C ) (n − 3)(n−3)·16 33 · 16
Q2 γ 3,1,3
∗
= ≤ < 1.
2 (Bn ) (n − 1)(n−1) 55

This completes the proof of Theorem 10.

Let B1 (n) = {Ca,b : a + b − 1 = n}, B2 (n) = {Ca,c,b : a + b + c − 1 = n} and
c
B3 (n) = {Ca,b : a + b − 1 = n}, and clearly all these sets contain (n, n + 1)-graphs. If
Gi ∈ Bi (n), for i = 1, 2, 3, then γ1 (G1 ) = 4γ ·2γ(n−1) , γ1 (G2 ) = 32γ ·2γ(n−2) = γ1 (G3 )
Q Q Q

and γ2 (G1 ) = 44γ · 22γ(n−1) , γ2 (G2 ) = 36γ · 22γ(n−2) = γ2 (G3 ).

Q Q Q

Theorem 2.1.7. Let G be an (n, n + 1)-graph such that G ∈

/ B1 (n) ∪ B3 (n) and H
be a graph in B2 (n) ∪ B3 (n). Then for γ > 0,
1. γ1 (G) < γ1 (H);
Q Q

2. γ2 (H) < γ1 (G).

Q Q

Proof. By applying Transformation 1 any tree T of size d attached to a bicyclic

graph G can be changed into a path Pd+1 , and by applying Transformation 4 any
pendant path can be converted into an internal path of G. Lemmas 2.1.1 and 2.1.4
guarantee that both transformations increase the first general multiplicative Zagreb
index and decrease the second general multiplicative Zagreb index. Thus the extremal
(n.n+1)-graphs with maximum first multiplicative Zagreb index and minimum second
multiplicative Zagreb index must be from the set B1 (n) ∪ B2 (n) ∪ B3 (n). Further, we
have
Qγ
(G ) 4 · 2n−1
Qγ1 1 = < 1,
1 (G2 ) 9 · 2n−2
Qγ
(G ) 256 · 2 · 4
Qγ2 1 = > 1.
2 (G2 ) 9
This completes the proof of Theorem 11.
 38

2.2 Graph Theoretic Parameters

A parameter of a graph G, is a numerical value (usually non-negative) associated with
a graph. Examples of parameter are order of G, size of G, maximum or minimum
degree, radius, diameter, girth, circumference, chromatic number, clique number,
independent number, number of pendant vertices etc.

Some examples

2.2.1 Classes of Graphs

As we have different classes of number such as class of natural numebrs, class of odd
numbers, class of even number, class of positive numbers, class of number greater
than 5 but less than 50 etc. In graph theory we have classes of graphs with respect
to different parameters

i. class of graphs of order 4

ii. class of trees of order 6

iii. class of trees of order 10 and 4 pendant vertices

 39

iv. class of trees of order 8 with maximum degree 3

in generalized way we can say

i. class of graph of order n

ii. class of trees of order n

iii. class of trees of order n and p pendant vertices

iv. class of trees of order n with maximum degree △

v. class of bipartite graphs of order n

vi. class of graphs of complete bipartite graphs of order n

Some Examples

2.2.2 Extremal Graphs in Different Classes of Graphs w.r.t.

Topological Indices
It is an interesting discussion to investigate the extremal graphs in a given class of
graphs w.r.t to a topological index. The results obtained from these can help us to
 40

investigate the molecules with maximum or minimum given property.

Let Gn be the class of graphs of order n.

Question 2.2.1. Find the extremal graphs in Gn for the forgotten topological index.
 41

Question 2.2.2. Find the extremal tress of order 12 and 5 pendant vertices for the
second Zagreb index.
 42

Let Un△ be the class of unicyclic graphs of order 10 with maximum degree 4.

Question 2.2.3. Find the extremal graphs in Un△ with respect to the first Zagreb
index.
Bibliography

[1] Y. Alizadeh, A. Iranmmanesh, T. Dos̆lić, Additively weighted Harary index of

some composite graphs, Discrete Mathematics, 313 (2013) 26–34.

[2] O. Araujo, J. A de la Peña, The connectivity index of a weighted graph, Linear

Algebra Appl., 283 (1998) 171–177.

[3] O. Araujo, J.A de la Peña, Some bounds for the connectivity index of a chemical
graph, J. Chem. Inf. Comput. Sci., 38 (1998) 827–831.

[4] A. R. Ashrafi, H. Yousefi-Azari, M. H. Khalifeh, The Hyper-Wiener index of

graph operations, Comput. Math. Appl., 56 (2008) 1402–1407.

[5] B. Bollobás, Graph theory–An introductory course, Springer-Verlag, 1979.

[6] B. Bollobás, P. Erdős, Graphs of extremal weights, Ars Combin., 50 (1998)

225–233.

[7] F. Buckley, F. Harary, Distances in graphs, Addison-Wesley, 1989.

[8] Braun, A. Kerber, M. Meringer, C. Rucker, Similarity of molecular descriptors:

the equivalence of Zagreb indices and walk counts, MATCH Commun. Math.
Comput. Chem., 54 (2005) 163–176.

[9] J. A. Bondy, U. S. R. Murty, Graph theory, Springer, 2008.

[10] G. Chartrand, P. Zhang, Intoroduction to graph theory, Mc Graw Hill, 2005.

43
 44

[11] T. Došlić, Splices, links and their degree-weighted Wiener polynomials, Graph
Theory Notes, New York, 48 (2005) 47–55.

[12] H. Deng, A unified approach to the extremal Zagreb indices of trees, unicyclic
graphs and bicyclic graphs, MATCH Commun. Math. Comput. Chem., 57
(2007) 597–616.

[13] M. V. Diudea, QSPR/QSAR Studies by molecular descriptors, Nova Sci. Publ.

Huntington, NY, 2000.

[14] M.V. Diudea, Indices of reciprocal properties or Harary indices, J. Chem. Inf.
Comput. Sci., 37 (1997) 292–299.

[15] J. Devillers, A.T. Balaban (Eds.), Topological indices and related descriptors in
QSAR and QSPR, Gordon and Breach, Amsterdam, 1999.

[16] E. Deutsch, S. Klavzar, M-polynomial and degree-based topological indices,

IJMC, 6(2), (2015), 93-102.

[17] K. Ch. Das, K. Xu, I. Gutman, On Zagreb and Harary indices, MATCH Com-
mun. Math. Comput. Chem., 70 (2013) 301–314.

[18] K. C. Das, A. Yurttas, M. Togan, A. S. Cevik, N. Cangul, The multiplicative

Zagreb indices of graph operations, Journal of Inequalities and Applications,
2013, 2013:90.

[19] Z. Du, Bo Zhou, N. Trinajstic, On the general sum-connectivity index of trees,

Applied Mathematics Letters, 24 (2011) 402–405.

[20] Z. Du, B. Zhou, N. Trinajstić, Sum-connectivity indices of trees and unicyclic

graphs of fixed maximum degree, arXiv.org/1210.5043.

[21] K.C. Das, B. Zhou, N. Trinajstić, Bounds on Harary index, J. Math. Chem.,
46 (2009) 1369–1376.
 45

[22] Z. Du, B. Zhou, N. Trinajstić, Minimum sum-connectivity indices of trees and

unicyclic graphs of a given matching number, J. Math. Chem., 47 (2010), 842-
855.

[23] Z. Du, B. Zhou, On sum-connectivity index of bicyclic graphs, Bull. Malays.

Math. Sci. Soc., in press.

[24] P. Erdős, B. Bollobás, Graphs of extremal weights, Ars Combin., 50 (1998)

225–233.

[25] M. Eliasi, A. Iranmanesh, I. Gutman, Multiplicative versions of first Zagreb

index, MATCH Commun. Math. Comput. Chem., 68 (2012) 217–230.

[26] S. Fajtlowicz, On conjectures of Graffiti - II, Congr. Numer., 60 (1987) 187–197.

[27] M. Goubko, Minimizing degree–based topological indices for trees with given
number of pendent vertices, MATCH Commun. Math. Comput. Chem., 71
(2014) 33–46.

[28] I. Gutman, Relation between hyper-Wiener and Wiener index, Chem. Phys.
Lett., 364 (2002) 352–356.

[29] I. Gutman, Multiplicative Zagreb indices of trees, Bull. Soc. Math. Banja Luka,
18 (2011) 17–23.

[30] I. Gutman, K.C. Das, The first Zagreb index 30 years after, MATCH Commun.
Math. Comput. Chem., 50 (2004) 83–92.

[31] I. Gutman, M. Goubko, Trees with fixed number of pendent vertices with mini-
mal first Zagreb index, Bull. Int. Math. Virt. Inst., 3 (2013) 161–164.

[32] M. Goubko, I. Gutman, Degree–based topological indices: Optimal trees with

given number of pendents, Appl. Math. Comput., 240 (2014) 387–398.
 46

[33] I. Gutman, M. K. Jamil, N. Akhter, Graphs with fixed number of pendent ver-
tices and minimal first Zagreb index, Transactions on Combinatorics, 4 (2015)
29–34.

[34] I. Gutman, S. Klavzar, B. Mohar (Eds.), Fifty years of the Wiener index,
MATCH Commun. Math. Comput. Chem., 35 (1997), 12–59.

[35] I. Gutman, S. Klavzar, B. Mohar (Eds.), Fiftieth anniversary of the Wiener

index, Discrete Appl. Math., 80 (1) (1997), 11–13.

[36] M. Goubko, T. Réti, Note on minimizing degree–based topological indices of

trees with given number of pendent vertices, MATCH Commun. Math. Comput.
Chem., 72 (2014) 633–639.

[37] I. Gutman, N. Trinajstić, Graph theory and molecular orbitals, Total π-electron
energy of alternant hydrocarbons, Chem. Phys. Lett., 17 (1972) 535–538.

[38] H. Hua, H. Deng, Unicycle graphs with maximum and minimum zeroth-order
general Randić indices, J. Math. Chem., 41 (2007) 173–181.

[39] Y. Hu, X. Li, Y. Yuan, Trees with maximum general Randić index, MATCH
Commun. Math. Comput. Chem., 52 (2004) 129–146.

[40] O. Ivanciuc, T.S. Balaban, A.T. Balaban, Reciprocal distance matrix, related
local vertex invariants and topological indices, J. Math. Chem., 12 (1993) 309–
318.

[41] W. Imrich, S. Klavžar, Product graphs: Structure and recognition, Wiley, New
York, 2000.

[42] M. K. Jamil, Distance based topological indices and double graph, Iranian Jour-
nal of Mathematical Chemistry, accepted.
 47

[43] M. K. Jamil, I. Tomescu, General sum-connectitvity index of trees and unicyclic

graphs with fixed maximum degree, submitted.

[44] M. K. Jamil, I. Tomescu, N. Akhter, Extremal degree-product indices of graphs

with fixed number of pendant vertices and cyclomatic number, International
Letters of Chemistry, physics and Astronomy, 59 (2015) 53–61.

[45] M. Karelson, Molecular Descriptors in QSAR/QSPR, Wiley-Interscience, New

York, (2000).

[46] L. B. Kier, L. H. Hall, Molecular connectivity in structure - activity analysis,

Wiley, New York, 1986.

[47] D.J. Klein, I. Lukovits, I. Gutman, On the definition of the hyper-Wiener index
for cycle-containing structures, J. Chem. Inf. Comput. Sci., 35 (1995) 50–52.

[48] M.H. Khalifeh, H. Yousefi-Azari, A.R. Ashrafi, A matrix method for computing
Szeged and vertex PI indices of join and composition of graphs, Linear Algebra
Appl., (2008) doi:10.1016/j.laa.2008.01.015.

[49] M.H. Khalifeh, H. Yousefi-Azari, A.R. Ashrafi, The first and second Zagreb
indices of some graph operations, Discrete Applied Mathematics, 157 (2009)
804–811.

[50] S. Klavzar, P. Zigert, I. Gutman, An algorithm for the calculation of the hyper-
Wiener index of benzenoid hydrocarbons, Comput. Chem., 24 (2000) 229–233.

[51] A. Miliçević, S. Nikolić, N. Trinjstić, On reformulated Zagreb indices, Mol.

Divers., 8 (2004) 393–399.

[52] E. Munarini, C. Perelli Cippo, A. Scagliola, N. Zagaglia Salvi, Double graphs,

Discrete Math., 308 (2008) 242–254.
 48

[53] H. Narumi, M. Katayama, Simple topological index. A newly devised index char-
acterizing the topological nature of structural isomers of saturated hydrocarbons,
Memoirs of the Faculty of Engineering, Hokkaido University, 16 (1984) 209–214.

[54] S. Nikolić, G. Kovacevic, A. Miliçević, N. Trinajstić, The Zagreb indices 30

years after, Croat. Chem. Acta, 76 (2003) 113–124.

[55] D. Plavs̆ić, S. Nikolić, N. Trinajstić, Z. Mihalić, On the Harary index for the
characterization of chemical graphs, J. Math. Chem., 12 (1993) 235–250.

[56] M. Randić, On characterization of molecular branching, J. Am. Chem. Soc., 97

(1975) 6609–6615.

[57] M. Saheli, T. Dos̆lić, Eccentric connectivity index of composite graphs, Utilitas

Mathematica, 95 (2014) 3–22.

[58] V. Sharma, R. Goswami, A.K. Madan, Eccentric connectivity index: A novel

highly discriminating topological descriptor for structure-property and structure-
activity studies, J. Chem. Inf. Comput. Sci., 37 (1997) 273–282.

[59] G. Su, I. Gutman, L. Xiong, L. Xu, Reciprocal product degree distance of graphs,
Manuscript.

[60] B. E. Sagan, Y. N. Yeh, P. Zhang, The Wiener polynomial of a graph, Int. J.

Quantum. Chem., 60 (5) (1996) 959–969.

[61] I. Tomescu, M. Arshad, On the general sum-connectivity index of connected

unicyclic graphs with k pendant vertices, Discr. Appl. Math., 181 (2015) 306–
309.

[62] I. Tomescu, M. Arshad, M. K. Jamil, Extremal topological indices for graphs of

given connectivity, Filomat, 29:7 (2015) 1639–1643.
 49

[63] R. Todeschini, D. Ballabio, V. Consonni, Novel molecular descriptors based

on functions of new vertex degrees. In: Gutman, I., Furtula, B. (eds.), Novel
Molecular Structure Descriptors - Theory and Applications I, 73–100, Univ.
Kragujevac, Kragujevac, 2010.

[64] R. Todeschini, V. Consonni, Handbook of Molecular Descriptors, Wiley–VCH

Meinheim, 2000.

[65] R. Todeschini, V. Consonni, New local vertex invariants and molecular descrip-
tors based on functions of the vertex degrees, MATCH Commun. Math. Comput.
Chem., 64 (2010) 359–372.

[66] I. Tomescu, M. K. Jamil, Maximum general sum-connectivity index for trees

with given independence number, MATCH Commun. Math. Comput. Chem.,
72 (2014) 715–722.

[67] D. B. West, Introduction to graph theory, Prentice Hall, 1996.

[68] H. Wiener, Structural determination of paraffin boiling points, J. Amer. Chem.

Soc., 69 (1947) 17.

[69] http:www.wolframalpha.com

[70] R. Xing, B. Zhou, N. Trinajstić, Sum-connectivity index of molecular trees,

J.Math. Chem., 48 (2010) 583–591.

[71] Y. Yuan, X. Li, Y. Hu, Trees with minimum general Randić index, MATCH
Comun. Math. Comput. Chem., 52 (2004) 119–128.

[72] Y. Yuan, X. Li, Y. Hu, Trees with maximum general Randić index, MATCH
Commun. Math. Comput. Chem., 52 (2004) 129–146.

[73] B. Zhou, I. Gutman, Relations between Wiener, hyper-Wiener and Zagreb in-
dices, Chem. Phys. Lett., 394 (2004) 93–95.
 50

[74] B. Zhou, I. Gutman, Further properties of Zagreb indices, MATCH Commun.

Math. Comput. Chem., 54 (2005) 233–239.

[75] Q. Zhao, S. Li, Sharp bounds for the Zagreb indices of bicyclic graphs with
k-pendent vertices, Discr. Appl. Math., 158 (2010) 1953–1962.

[76] B. Zhou, N. Trinajstić, On a novel connectivity index, J. Math. Chem., 46

(2009) 1252–1270.

[77] B. Zhou, N. Trinajstić, On general sum-connectivity index, J. Math. Chem., 47

(2010) 210–218.