Chapter 1

Introduction

Graph theory is a branch of mathematics that deals with mathematical structures which
represent the interrelation between objects. There are various techniques which have been
employed in the study of graphs. One of the effective ways is to use matrices. The spec-
tral graph theory is the study of properties of graphs using matrix theory. It has numerous
applications in several fields such as quantum chemistry, computer science, physics, biol-
ogy, engineering and economics [1]. There are several areas in spectral graph theory that
has been studied extensively due to its significance, both theoretical and practical. Since
the time the concept of graph energy was introduced by Gutman [2] in 1978, this topic had
been of interest for many researchers. There are several variations of graph energy found
in the literature [3, 4, 5, 6]. One of its variations which has been gaining the attention of
numerous researchers due to its combined perspective towards graph coloring and graph
energy is the color energy of graphs, initiated by Adiga et al. [7]. In the present study, we
extend the study of color energy of graphs by exploring its variations and by presenting a
generalized approach to the concept of color energy. First, we present the basic concepts
and results required for a better understanding of the matter presented in the subsequent
chapters.
 Chapter 1. Introduction

1.1 Basic terminology

Unless otherwise mentioned, for graph theoretic concepts we refer to West [8] and Char-
trand and Zhang [9]. For spectral theoretic concepts, we refer to Cvetković et al. [10] and
Gutman et al. [5].
A graph G = (V, E) is an ordered pair consisting of a finite nonempty set V of objects
called vertices together with a set E (possibly empty) of 2-element subsets of V called
edges [9]. A loop is an edge which joins a vertex with itself. Multiple edges or parallel
edges are two or more edges which joins the same pair of distinct vertices. A simple graph
is a graph which has neither loops nor multiple edges. A graph is finite if its vertex and
edge sets are finite [9]. The order of G is the number of vertices in G and the size of G
is the number of edges in G. Throughout the dissertation, we consider simple and finite
graphs of order n and size m.
If uw is an edge for u, w ∈ V , then u and w are adjacent vertices and they are called
neighbors of each other. The set of neighbors of a vertex v ∈ V is the open neighborhood
N(v) of v and the set N(v) along with the vertex v is the closed neighborhood N[v] of v.
The number of vertices in the open neighborhood of a vertex v is the degree deg(v) of the
vertex v in G. The largest and the smallest degree among vertices of G are the maximum
degree ∆(G) and minimum degree δ (G) respectively [9].
For u, v ∈ V (G), a u − v walk W in G is a sequence of vertices in G, beginning with u
and ending at v such that consecutive vertices in W are adjacent in G [9]. A walk in G that
has no repeated edge is a trail in G and a u − v walk in G that has no repeated vertex is
called a u − v path in G [9]. Two vertices u and v are connected vertices in G if there is a
u − v path in G. A connected graph G is a graph such that every two vertices u, v ∈ V are
connected in G. A graph is disconnected if it is not connected [9]. In the present work,
we consider connected as well as disconnected graphs.

2
 Chapter 1. Introduction

The distance d(u, v) between the two vertices u and v in a connected graph G is the
length of the shortest u − v path. The detour distance D(u, v) is the length of the longest
u − v path in G. The diameter of a connected graph G is diam(G) = max{d(u, v) : u, v ∈
V (G)}. The eccentricity of a vertex u in a connected graph G is e(u) = max{d(u, v) : v ∈
V (G)}. The radius of G is min{e(u) : u ∈ V (G)} [8].
A graph H is called a subgraph of G if the vertex and edge sets of H are contained
in the vertex and edge sets of G. A subgraph H is called a spanning subgraph of G if
V (H) = V (G). Let S be a subset of the vertex set V of G. Then the subgraph induced by
S is the graph G[S] where two vertices u and v are adjacent in G[S] if and only if they are
adjacent in G. A subgraph H of G is called induced subgraph of G if H = G[S] for some
S ⊆ V (G) [9].
An empty or null graph is a graph with no edges and a non-empty graph is a graph
with one or more edges. A complete graph Kn of order n is a graph in which all the distinct
vertices are adjacent. A path Pn is a graph of order n and size n − 1 that joins a sequence
of distinct vertices u1 , u2 , . . . , un such that the edges are ui ui+1 , for i = 1, 2, . . . , n − 1. A
cycle Cn of order n is a closed path on n vertices [9].
An r-regular graph is a graph with all the vertices having the same degree r. A
complete graph Kn is an (n − 1)-regular graph whereas a cycle Cn is a 2-regular graph. A
cubic graph is a 3-regular graph [9].
The generalized Petersen graph G(p, k) is a connected cubic graph consisting of an
inner star polygon with the vertices u0 , u1 , . . . , u p−1 and an outer regular polygon with
p−1
S
the vertices v0 , v1 , . . . ,v p−1 . The edges ui vi are called spokes [11]. G(p, k) is a graph
i=0
of order n = 2p and p ≥ 3 with vertex and edge sets as follows: For 1 ≤ k ≤ p − 1 and
2k ̸= p,

(i) V [G(p, k)] = {u0 , u1 , . . . , u p−1 , v0 , v1 , . . . , v p−1 },

3
 Chapter 1. Introduction

(ii) E[G(p, k)] = {ui vi , ui ui+1 , vi vi+k : i ∈ Z p and Z p is group of integer modulo p}.

Figure 1.1 depicts G(5, 1) and G(5, 2), where G(5, 2) is the Petersen graph.

v1 v1

u1
u1
v5 v2 v5 u5 u2 v2
u5 u2
u4 u3
u4 u3

v4 v3 v4 v3
( A ) G(5, 1) ( B ) G(5, 2)

F IGURE 1.1: Generalized Petersen graphs

A bipartite graph G is a graph whose vertex set V can be partitioned into two disjoint
independent non-empty sets U and V , such that U and V are partite sets. A complete
bipartite graph Kr,s is a bipartite graph whose vertex set can be partitioned into two partite
sets V and W such that vw is an edge of G if and only if v ∈ V and w ∈ W [9]. A graph
obtained by removing a perfect matching or one factor from Kr,r is the graph Kr,r − F1
[12]. It is also called crown graph and denoted by Sn0 [7]. K3,3 and K3,3 − F1 are illustrated
in Figure 1.2.

v3 v6 v3 v6

v2 v5 v2 v5

v1 v4 v1 v4
( A ) K3,3 ( B ) K3,3 − F1

F IGURE 1.2: Complete bipartite and crown graph on 6 vertices

4
 Chapter 1. Introduction

A graph is k-partite if its vertex set can be partitioned into k subsets V1 ,V2 , . . . ,Vk and
there is an edge between two vertices in two different partite sets. A 2-partite graph is
called bipartite graph. A complete multipartite graph is a k-partite graph for some integer
k ≥ 2 such that any two vertices are adjacent if and only if they are in different partite sets
[9]. It is denoted by Kr×s [13]. One of its specific case is a cocktail party graph Kr×2 such
that there are r partite sets wherein each partite set has 2 vertices [7].
A tree is a connected acyclic graph. A star K1,n−1 is a tree with one central vertex of
degree n − 1 connected with n − 1 pendant vertices. It is also a complete bipartite graph.
A double star Br,s is a tree obtained by joining r and s pendent vertices at the end points
of an edge respectively. It has exactly two non-pendant vertices that are adjacent [9]. That
is, it is a graph obtained from two star graphs K1,r−1 and K1,s−1 by joining their central
vertices. Figure 1.3 gives K1,7 and B4,4 .

u3

u2 u4
u1 w2 u2
w1 u1
u5 w3 u3
u8
u7 u6 w4 u4
( A ) K1,7 ( B ) B4,4

F IGURE 1.3: Star and double star graph on 8 vertices

The complement G of G is a graph with the same vertex set as that of G and two
vertices are adjacent in G if and only if they are non-adjacent in G. Let P = {V1 ,V2 , . . . ,Vk }
be a vertex partition of G. Then for all Vi , V j in P and i ̸= j, the k-complement (G)k of G is
obtained by removing edges between the vertices of Vi and V j and adding edges between
the vertices of Vi and V j which are not in G [14]. The k(i)-complement (G)k(i) of G for

5
 Chapter 1. Introduction

the vertex partition is obtained by removing edges of G within Vr and adding the edges
from G between the vertices of Vr [15]. Figure 1.4 illustrates these concepts.

u1 u1

u5 u2 u5 u2

u4 u3 u4 u3
(A) H (B) H
u1 u1

u5 u2 u5 u2

u4 u3 u4 u3
( C ) (H)2 ( D ) (H)2(i)

F IGURE 1.4: Complement, 2-complement and 2(i)-complement of graph

H corresponding to the partition {{u1 , u3 }, {u2 , u4 , u5 }}

The union G1 ∪G2 of two graphs G1 and G2 has vertex set V (G1 )∪V (G2 ) and edge set
E(G1 ) ∪ E(G2 ). The join of two graphs G1 and G2 has vertex set V (G1 ) ∪V (G2 ) and edge
set E(G1 ) ∪ E(G2 ) ∪ {ui vi |ui ∈ V (G1 ) and vi ∈ V (G2 )}. For example, a wheel graph Wn
of order n + 1 is a graph obtained by join of Cn and K1 [9]. In the present work, we follow
the notation G1 ▽G2 for join of two graphs G1 and G2 which was used by Cvetković et al.
[10] and Sampathkumar et al. [16].
A clique is a set of pairwise adjacent vertices and the clique number ω(G) is the order
of the largest clique. An independent set is a set of pairwise non-adjacent vertices and
the maximum cardinality of an independent set is the independence number α(G).
Domination is one of the rich area of research in graph theory. It has attracted many
researchers due to its various applications in several technical fields. Now, we state some
definitions related to domination that are required for the further discussion.

6
 Chapter 1. Introduction

A subset S of vertex set V of a graph G is called a dominating set if every vertex in

V is either in S or is a neighbor of a vertex in S. The minimum order of S is called the
domination number γ(G) [17].
The D-partition or domatic partition is a partitioning of vertex set V of G into domi-
nating sets. The maximum order of domatic partitions of V is called the domatic number
d(G) [18].
A subset D of the vertex set V of G is a total dominating set if every vertex of V is
adjacent to some vertex in D. The cardinality of the smallest total dominating set is the
total domination number γt (G). A partitioning of V into total dominating sets is total
domatic partition. The maximum number of such partitions of V is total domatic number
t(G) [19].
A k-ply dominating set is a subset D of the vertex set V of G if for each vertex u in
V −D there exist distinct k vertices of D which are adjacent to u. A k-ply domatic partition
is a partition of V into k-ply dominating sets. Maximum number of such partitions of V
is k-ply domatic number and denoted by d k (G) [20].
We need few results concerned with domination number and related concepts for the
further discussion that are mentioned in the following theorems:

Theorem 1.1.1. [9] If G is a graph of order n, then

(i) ∆(G) ≤ n − γ(G),

(ii) γ(G) ≤ 2n , provided G has no isolated vertices.

Theorem 1.1.2. [21] Let G(p, k) be a generalized Petersen graph. Then

(i) d(G(p, k)) = 4 if and only if p ≡ 0( mod 4),

(ii) t(G(p, k)) = 3 if and only if p ≡ 0( mod 4),

(iii) d 2 (G(p, k)) = 2,

7
 Chapter 1. Introduction

(iv) d 3 (G(p, k)) = 2 if and only if p is even.

Now we state some of the basic results from linear algebra that are required for the
present study:

Theorem 1.1.3. [22] If A is a real or complex matrix of order n ×n with the characteristic
polynomial φ (G, λ ) and eigenvalues λ1 , λ2 , . . . , λn , then

(i) A principal sub-matrix of order r × r of A is a sub-matrix consisting of the same set

of r rows and r columns and a principal minor of order r × r of A is the determinant
of a principal sub-matrix of order r × r.

(ii) If a0 , a1 , a2 , . . . , an are the coefficients of φ (G, λ ), then (−1)r ar is the sum of prin-
cipal minors of order r × r.

(iii) If the rth symmetric function Sr (A) is the sum of the product of the eigenvalues of A
taken r at a time, then it is the sum of r × r principal minors of A.

(iv) Trace of A is the sum of diagonal entries of A and it can also be represented as
tr(A) = S1 (A) = −a1 .
n
(v) ∏ λi = |A|.
i=1

Definition 1.1.4. [23] A matrix A is symmetric if A = AT and normal if A commutes with

its transpose or conjugate transpose.

Theorem 1.1.5. [24] If λ is an eigenvalue of the matrix A = (ai j )n×n and R = max Ri ;
i
where Ri is the sum of absolute values of the entries of A in the ith row, then

(i) |λ | ≤ n max |ai j |,

i, j

(ii) |λ | ≤ R,
n n
(iii) ∑ |λi |2 ≤ ∑ |ai j |2 and equality holds if A is a normal matrix.
i=1 i, j

8
 Chapter 1. Introduction

A B


Lemma 1.1.6. [25] If A, B,C, D are matrices where A is invertible and S = CD , then
the determinant of S, |S| = |A||D −CA−1 B|. If A and B commute, then |S| = |AD − BC|.

AB


Lemma 1.1.7. [10] If C = BA is a symmetric block matrix of order 2 × 2, then the
spectrum of C is the union of the spectra of A + B and A − B.

Lemma 1.1.8 (Cauchy–Schwartz inequality). [24] If ai and bi are the real numbers for
1 ≤ i ≤ n, then
n n n
2 2
( ∑ ai bi ) ≤ ∑ (ai ) ∑ (bi)2. (1.1)
i=1 i=1 i=1

Lemma 1.1.9 (Hölder’s inequality). [26] If xi , yi ≥ 0, p > 1, 1p + 1q = 1, then

n n
1/p n
1/q
∑ xiyi ≤ ∑ xip ∑ yqi . (1.2)
i=1 i=1 i=1

Lemma 1.1.10. [27] Let A be a matrix and λ1 , λ2 , λ3 , . . . , λn be its eigenvalues. Then

s
n n
∑ |λi |2 ≤ ∑ |λi |.
i=1 i=1

Lemma 1.1.11. [26, 28] Let ai and bi , for 1 ≤ i ≤ n, be positive real numbers. If M1 =
max max (b ), m = min (a ) and m = min (b ), then
1≤i≤n(ai ), M2 = 1≤i≤n i 1 1≤i≤n i 2 1≤i≤n i
q q 
2
1 M1 M2 m1 m2
(i) ∑ni=1 (ai )2 ∑ni=1 (bi )2 ≤ 4 m1 m2 + M1 M2 ∑ni=1 ai bi ,

2 n2

2
(ii) ∑ni=1 (ai )2 ∑ni=1 (bi )2 − ∑ni=1 ai bi ≤ 4 m1 m2 − M1 M2 .

Lemma 1.1.12. [29] Let the real numbers ai ̸= 0 and bi , for 1 ≤ i ≤ n such that a ≤ abii ≤ b.
Then
n n n
∑ (ai)2 + ab ∑ (bi)2 ≤ (a + b) ∑ aibi.
i=1 i=1 i=1

Lemma 1.1.13. [30] Let ai be a real number such that 0 ≤ a ≤ ai ≤ b for 1 ≤ i ≤ n. Then

1 n

1 n 1

(a+b)2
(i) n ∑i=1 ai n ∑i=1 ai ≤ 4ab ,

9
 Chapter 1. Introduction

1 n

1 n 1

[ 2n ]b+[ n+1
2 ]a [ n+1 n
2 ]b+[ 2 ]a
(ii) n ∑i=1 ai n ∑i=1 ai ≤ ab , where [x] represents the integral
part the real number x.

Definition 1.1.14. [31] If M is a matrix associated with a graph G of order n, the M-

coronal ΓM (λ ) of G is the sum of elements of the matrix (λ In − M)−1 . That is, ΓM (λ ) =
1n T (λ In − M)−1 1n , where 1n is a column vector of order n × 1 and 1Tn is its transpose and
In is an identity matrix of order n × n.

This definition is a generalization of the coronal of a graph introduced by McLeman

and McNicholas for the adjacency matrix of a graph [32]. We would be referring to the
Lemma 1.1.15 proved by Liu and Zhang [33] relating ΓM (λ ) with a real matrix A, the
identity matrix In and a matrix Jn of order n × n in which each element ai j = 1.

Lemma 1.1.15. [33] Let A be an n × n real matrix. Then

|λ In − A − αJn | = [1 − αΓM (λ )]|λ In − A| (1.3)

where α is a real number and λ is an eigenvalue of A.

In the following theorem, BC (m, n) represents the class of matrices such that they
are symmetric block circulant matrices having m blocks each of order n.

Theorem 1.1.16. [34] Let A be a symmetric block circulant matrix such that A ∈ BC (m, n)
and it has 2 or more blocks such that each block Ak is a circulant matrix having the en-
(k)
tries ar for r = 0, 1, . . . , n − 1 and k = 0, 1, . . . , m − 1. Let m be an even integer such that
m ≥ 2. Then for the eigenvalues λ j,s of A where j = 0, 1, . . . , m − 1 and s = 0, 1, . . . , n − 1,
we have the following cases:

10
 Chapter 1. Introduction

(i) If n is even and greater than or equal to two, then

n/2−1  
(0) (0) (0) 2srπ
λ j,s = a0 + (−1)s an/2 + 2 ∑ ar cos
r=1 n
m/2−1 n−1    
k j rs
+2 ∑ ∑ cos 2 m n π
k=1 r=0
 n/2−1  
(m/2)
j s (m/2) (m/2) 2srπ
+ (−1) a0 + (−1) an/2 + 2 ∑ ar cos . (1.4)
r=1 n

(ii) If n is odd and greater than or equal to three, then

n−1  
(0) (0)2srπ
λ j,s = a0 + 2 ∑ ar cos
r=1 n
m/2−1 n−1    
k j rs
+ 2 ∑ ∑ cos 2 π
n=1 r=0 m n
 n−1  
j (m/2) (m/2) 2srπ
+ (−1) a0 + 2 ∑ ar cos . (1.5)
r=1 n

Theorem 1.1.17. [23] Let A = [ai j ] ∈ Mn , Ri (A) = ∑ |ai j | for all i = 1, 2, . . . , n which
i̸= j
denotes the deleted absolute row sums of A, and consider the Geršgorin discs {z ∈ C :
|z − aii | ≤ Ri (A)}, i = 1, 2, . . . , n. The eigenvalues of A are in the union of Geršgorin
n
{z ∈ C : |z − aii | ≤ Ri (A)}. Furthermore, if the union of k of the n discs
S
discs G(A) =
i=1
that comprise G(A) forms a set Gk (A) that is disjoint from the remaining n − k discs,
then Gk (A) contains exactly k eigenvalues of A, counted according to their algebraic
multiplicities.

Definition 1.1.18. [23] A Hermitian matrix A ∈ Mn is positive definite if x∗ Ax > 0 for all
nonzero x ∈ Cn . It is positive semidefinite if x∗ Ax ≥ 0 for all nonzero x ∈ Cn .

11
 Chapter 1. Introduction

Theorem 1.1.19. [23] A Hermitian matrix is positive semidefinite if and only if all of its
eigenvalues are nonnegative. It is positive definite if and only if all of its eigenvalues are
positive.

1.2 Graph coloring

Graph coloring is one of the sought after areas for research in graph theory. The concept
of graph coloring is widely used in computer science, engineering, scheduling problems,
etc. There is a vast literature available on the same [2, 35]. The graph coloring problems
include coloring of vertices or edges or both. For the present study, we consider vertex
coloring which is an assignment of integers to the vertices of a graph under consideration.
The assignment of colors to the vertices can be done in several ways, one of the well-
studied method is proper vertex coloring.

Definition 1.2.1. [35] A proper vertex coloring is a function c : V (G) → {0} ∪ N such
that c(vi ) ̸= c(v j ) if vi and v j are adjacent in G, for 1 ≤ i, j ≤ n. The minimum number of
colors required to color a graph G is the chromatic number χ(G) of G.

If V1 ,V2 , . . . ,Vk are the set of vertices in a graph G such that vertices in Vi are colored
with i for 1 ≤ i ≤ k, then each nonempty set Vi for 1 ≤ i ≤ k is called a color class and
the color classes produce a partition of V (G) [35]. Therefore, a proper vertex coloring of
a graph G leads to partitioning of V (G) into distinct independent sets.
During past few decades, various type of proper vertex colorings have been intro-
duced into literature. For more details about it, we refer to [36]. Now we present some
definitions and related parameters of some coloring schemes which would be used in
subsequent chapters.

Definition 1.2.2. [35] For a given finite set T of non-negative integers containing 0, a
T -coloring of a graph G is an assignment c of colors (positive integers) to the vertices of

12
 Chapter 1. Introduction

G such that if uw is an edge in G, then |c(u) − c(w)| ∈

/ T . The T -chromatic number χT (G)
is the minimum number of colors used in a T -coloring of G.

Definition 1.2.3. [35] For non-negative integers h and k, an L(h, k)-coloring c of a graph
G is an assignment of colors (non-negative integers) to the vertices of G such that if
u and w are adjacent vertices of G, then |c(u) − c(w)| ≥ h while if d(u, w) = 2, then
|c(u) − c(w)| ≥ k.

A T -coloring and an L(2, 1)-coloring is depicted in Figure 1.5.

1 0

4 2

1 1 5 4

7 7
( A ) T -coloring for the set T = {0, 1, 2} ( B ) L(2, 1)-coloring

F IGURE 1.5: T and L(2, 1)-coloring of a graph

We would be using the results related to the chromatic number of a graph that are
stated in the following theorem:

Theorem 1.2.4. [9] If G is a graph of order n, then

(i) 1 ≤ χ(G) ≤ n,

n
(ii) ≤ χ(G) ≤ n − α(G) + 1,
α(G)

(iii) n ≤ χ(G)α(G),

(iv) χ(G) ≤ 1 + ∆(G),

13
 Chapter 1. Introduction

(v) for each finite set T of non-negative integers containing 0,

χT (G) = χ(G).

Definition 1.2.5. [35] Let G be a connected graph of order n and diameter d. For an
integer k with 1 ≤ k ≤ d, a radio k-coloring of G is an assignment c of colors (positive
integers) to the vertices of G such that d(u, v) + |c(u) − c(v)| ≥ 1 + k for every two distinct
vertices u and v of G. The value rck (c) of a radio k-coloring c of G is the maximum color
assigned to a vertex of G, and the radio k-chromatic number rck (G) of G is min{rck (c)}
over all radio k-colorings c of G. Figure 1.6 illustrates a radio 2-coloring of a graph.

5 5

1 3
7 7
F IGURE 1.6: Radio 2-coloring of a graph

Now we consider few improper colorings, in which the adjacent vertices may receive
same colors.

Definition 1.2.6. [35] For distinct vertices u and v in a connected graph G of order n,
let D(u, v) denote the length of a longest u − v path in G. A coloring c of G is called
a Hamiltonian coloring if D(u, v) + |c(u) − c(v)| ≥ n − 1 for every two distinct vertices
u and v of G. The value hc(c) of a Hamiltonian coloring c of G is the maximum color
assigned to a vertex of G. The Hamiltonian chromatic number hc(G) of G is min{hc(c)}
over all Hamiltonian colorings c of G. Figure 1.7 gives a hamiltonian coloring of a graph.

14
 Chapter 1. Introduction

5 2

4 3
F IGURE 1.7: Hamiltonian coloring of a graph

A Hamiltonian u − v path in a graph G is a u − v path containing every vertex of G [9].

A graph G is called Hamiltonian-connected graph if for every pair of distinct vertices u, v
of G, there is a Hamiltonian u − v path in G [37].
The results pertaining to Hamiltonian coloring stated in the following theorem are
required for the present study.

Theorem 1.2.7. [37, 38]

(i) Let G be a connected graph. Then hc(G) = 1 if and only if G is Hamiltonian-

connected.

(ii) For n ≥ 5, let Hn be the graph obtained from the complete graph Kn−2 , where
V (Kn−2 ) = {v1 , v2 , v3 , . . . , vn−2 }, by adding the two pendant edges v1 vn−1 and vn−2 vn .
Then Hn is a graph order n and diameter 3 such that hc (Hn ) = 2n − 5.

(iii) For a complete bipartite graph, hc (Kr,s ) = χ(Kr,s ) = r.

(iv) For a star graph of order n ≥ 3, hc (K1,n−1 ) = (n − 2)2 + 1.

Definition 1.2.8. [39] The open neighborhood coloring of a graph G = (V, E) is a col-
oring f : V (G) → N, such that for each w ∈ V (G) and ∀u, v ∈ N(w), f (u) ̸= f (v). The
maximum value of f (w), ∀w ∈ V (G) is called the span of the open neighborhood color-
ing. The minimum value of span of f over all open neighborhood colorings f is called
open neighborhood chromatic number χonc (G).

15
 Chapter 1. Introduction

Definition 1.2.9. [40] For an integer k ≥ 1, a k-improper C-coloring of a graph G = (V, E)

is a coloring φ : V (G) → N such that for each vertex v at most k vertices in N(v) receive
colors different from φ (v). The k-improper upper chromatic number χ k−imp (G) is the
maximum number of colors permitted in such a coloring.

Figure 1.8 depicts an open neighborhood coloring and a 2-improper coloring of a

graph.

1 1

1 1
2 3 1 2
2 3 2 3

4 2
( A ) Open neighborhood coloring ( B ) 2-Improper C-coloring

F IGURE 1.8: Improper colorings of a graph

We need a few results for k-improper upper chromatic number χ k−imp (G) from [40]
that are given in the next theorem.

Theorem 1.2.10. [40]


 n +1
 

2 if k = 1,
(i) χ k−imp (Pn ) =

n
 if k ≥ 2.

 n
 

2 if k = 1,
(ii) χ k−imp (Cn ) =

n
 if k ≥ 2.

16
 Chapter 1. Introduction






1 if k = 1 and n ≥ 4,




2
 if k = 2 and n ≥ 4,
(iii) χ k−imp (Wn ) =




k+1 if n ≥ k ≥ 1,




n + 1
 if k ≥ n ≥ 3.

n
 
(iv) χ k−imp (Kn ) = holds for all1 ≤ k ≤ n − 1.
n−k





2 if k = 1,




(v) χ k−imp G(5, 2) = 5 if k = 2,





10 if k ≥ 3.


1.3 Color energy of a graph

The concept of color energy of a graph was introduced by Adiga et al. [7] as a variation
of energy of a graph. It connects two major areas of graph theory, viz. graph coloring and
graph energy.
Earlier to the nineteenth century, energy was a concept related to molecular graphs.
Molecular graphs are the models of molecular structures in terms of vertices and edges
wherein vertices represent atoms and edges represent the bond between them [5]. For
certain scientific reasons, the molecular graphs are restricted up-to degree 3. Energy of a
molecular graph represents the total π-electron energy of a molecule [5]. Hückel gave a
detailed mathematical derivation of this quantity. It was clearly in terms of eigenvalues
of the corresponding adjacency matrix of a graph with some constant values. Gutman
[2] observed that the mathematical entity can be generalized to energy of an arbitrary
graph and it may have some mathematical implications. So, he put forward the concept
of energy in the realm of graph theory. He defined energy of a graph E(G) as the sum of
absolute values of eigenvalues of adjacency matrix of the graph [5].

17
 Chapter 1. Introduction

The bounds of energy of a graph in terms of its order and size have been studied by
McClelland [5]. These kind of bounds are known as McClelland type of bounds. One of
the bounds is given in the following theorem:
√
Theorem 1.3.1. [5] For a graph G of order n and size m, E(G) ≤ 2mn.

Many new variations in graph energy have been proposed based on different matrices
other than adjacency matrix. The details can be found in some recent survey articles [3, 6].
The first few variations of energy are extended energy, Laplacian energy, distance energy
and Randić energy [41, 42]. These energies are based on extended adjacency matrix,
a derived matrix from Laplacian matrix and degree matrix, distance matrix and Randić
matrix respectively.
A special type of matrix, namely L-matrix was introduced by Sampathkumar and
Sriraj [43], by using the idea of partitioning the vertex set along with the adjacencies of
vertices. The authors defined L-matrix as M = (ai j ) whose elements are





2 if vi and v j are adjacent such that c(vi ) = c(v j ),




1
 if vi and v j are adjacent such that c(vi ) ̸= c(v j ),
ai j =




−1 if vi and v j are non-adjacent such that c(vi ) = c(v j ),




0
 otherwise.

They have proved that there is one to one correspondence between the adjacency ma-
trix of a signed graph and L-matrix of its underlined graph. Another useful property of
L-matrix is mentioned in the next theorem.

Theorem 1.3.2. [43] A n × n matrix M = (ai j ) with zero diagonal whose entries belong
to the set {2, 1, 0, −1} is the L-matrix of a labeled graph G = (V, E) of order n if and only
if

18
 Chapter 1. Introduction

(i) ai j , a jk ∈ {2, −1} =⇒ aik ∈ {2, −1},

(ii) ai j ∈ {2, −1} and a jk ∈ {0, 1} =⇒ aik ∈ {0, 1}.

The concept of L-matrix was an inspiration for Adiga et al. [7] to introduce the con-
cept of color matrix Ac (G) of a graph G. Based on Ac (G), they initiated the study of color
energy of a graph G in 2013.

Definition 1.3.3. [7] The color energy of a graph G is the sum of the absolute values of
the eigenvalues of color matrix of G. It can be written as

n
Ec (G) = ∑ |λi |
i=1

where λi ’s are eigenvalues of the color matrix Ac (G) whose entries are





1 if vi and v j are adjacent such that c(vi ) ̸= c(v j ),


ai j = −1 if vi and v j are non-adjacent such that c(vi ) = c(v j ),





0
 otherwise,

where c(vi ) is the color of a vertex vi in G.

The characteristic polynomial of Ac (G) is

Pc (G, λ ) = |λ In − Ac (G)| = a0 λ n + a1 λ n−1 + . . . + an−1 λ + an ,

where In is an identity matrix and the coefficients a0 , a1 , . . . , an are the principal minors of
Ac (G). The spectrum of Ac (G) is

 
λ1 λ2 λ3 . . . λr 
Specc (G) =  
p1 p2 p3 . . . pr

19
 Chapter 1. Introduction

where p1 , p2 , . . . , pr are the multiplicities of distinct eigenvalues λ1 > λ2 > . . . > λr of

Ac (G) respectively, for r ≤ n.
In Example 1.3.4, the computation of color energy of a graph G is described. The
color energy of a graph G corresponding to minimum number of colors χ(G) is denoted
by Eχ (G) and, color matrix and spectrum of G are denoted by Aχ (G) and Specχ (G)
respectively [7].

Example 1.3.4. Consider a graph G having the chromatic number χ(G) = 3 as shown in
Figure 1.9.

1 1

1 3
F IGURE 1.9: A graph G with χ(G) = 3

The color matrix of G is

 
1 1 −1 −1
0
 
 
1 0 1 1 1
 
 
Aχ (G) = 
 1 1 0 0 0 .

 
−1 1 0 0 −1
 
 
−1 1 0 −1 0

By obtaining the characteristic polynomial of Aχ (G) and computing further we get the
eigenvalues of Aχ (G) as −3, −1, 1, 1, 2. Therefore, Eχ (G) = 8.

In addition to the concept of color matrix, Adiga et al. [7] introduced another concept
that is color complement of a graph. It is a specific case of k-complement (G)k of G.

20
 Chapter 1. Introduction

Definition 1.3.5. [7] If G is graph with vertex coloring, then its color complement is a
graph (G)c such that
(i) ui and u j are adjacent in (G)c , if they are non-adjacent in G with c(ui ) ̸= c(u j ),

(ii) ui and u j are non-adjacent in (G)c , if they are adjacent in G or non-adjacent with
c(ui ) = c(u j ) in G.

Adiga et al. [7] determined the color energy Eχ (G) of some families of graphs cor-
responding to minimum number of colors χ(G) required to color a graph G. They also
obtained color energy of color complements of a few classes of graphs. Some of the re-
quired results for the characteristic polynomials Pχ (G, λ ) and Eχ (G) are mentioned in the
following theorem.

Theorem 1.3.6. [7] Let G be a graph and Eχ (G), Eχ (G) are the corresponding color
energies of G and (G)c with respect to minimum number of colors χ(G). Then
(i) Pχ (Kn , λ ) = (λ + 1)(n−1) [λ − (n − 1)], Eχ (Kn ) = 2(n − 1), and Eχ (Kn )c = 0

(ii) Pχ (K1,n−1 , λ ) = [λ − 1]n−1 [(−1)2n+1 ][λ + n − 1],

Pχ ((K1,n−1 )c , λ ) = λ [λ − 1]n−2 [λ + n − 2],
Eχ (K1,n−1 ) = 2(n − 1) and, Eχ (K1,n−1 )c = 2(n − 2),

(iii) Eχ (Kr,s ) = 2(r + s − 1) and Eχ (Kr,s )c = 2(r + s − 2),

(iv) Eχ (Kr,s − F1 ) = Eχ (Kr,s − F1 )c = 4(n − 1),

(v) Eχ (Kn×2 ) = 6(n − 1) and Eχ (Kn×2 )c = 2n.

(vi) If Nn is a null graph of order n, then Pχ (Nn , λ ) = [λ + (n − 1)][λ − 1]n−1 and

Eχ (Nn ) = Eχ (Nn )c = 2(n − 1).

Along with these results, Adiga et al. [7] have examined some properties of color
matrix and its eigenvalues. We need the following property of eigenvalues of Ac (G) for
the further investigation.

21
 Chapter 1. Introduction

Proposition 1.3.7. [7] Let G be a graph of order n and size m such that λ1 , λ2 , λ3 , . . . , λn
are the eigenvalues of the color matrix Ac (G) and m′c is the number of pairs of non-
n
adjacent vertices with same colors. Then ∑ λi2 = 2(m + m′c ).
i=1

Further, the authors derived few bounds for color energy of a graph in terms of its
order n, size m and the number of pairs of non-adjacent vertices m′c . The following result
containing McClelland type of bound of color energy would be useful for the further
discussion.

Theorem 1.3.8. [7] Let G be a graph of order n and size m. Then

q
2 p
2(m + m′c ) + n(n − 1)D n ≤ Ec (G) ≤ 2n(m + m′c ) (1.6)

where m′c is the number of pairs of non-adjacent vertices having same color in G and D
is the determinant of Ac (G).

Some more bounds for color energy have been established in terms of eigenvalues
of color matrix of a graph G, its order and size [44, 45, 46, 47]. Joshi and Joseph [44]
obtained some bounds for color energy of a graph G in terms of m and m′c . One of their
result which is useful for the present work, is mentioned in the next theorem.

Theorem 1.3.9. [44] Let G be a graph of size m and m′c be the number of pairs of non-
adjacent vertices receiving same color in G. Then

m + m′c ≤ Ec (G) ≤ 2(m + m′c ).

p
2 (1.7)

Several researchers explored the concept of color energy and obtained color energy of
different graph classes and a few derived graphs with respect to proper coloring [44, 45,
46, 48, 49]. Recent studies on color energy [50, 51] that includes the determination of the

22
 Chapter 1. Introduction

color energy of generalized color complements of some common classes of graphs and
cluster graphs.
Apart from finding color energies of different graph classes and obtaining bounds for
color energy, some new variations in color energy of graphs have been introduced into
the literature. One of the first few variations is color Laplacian energy of a graph. This
concept was introduced independently by two groups of researchers, namely Bhat and
D’souza [52] as well as Shigehalli and Betagiri [53] in 2015. Later, Bhat and D’Souza
[54] introduced the color signless Laplacian energy of a graph G. Another variations in
color energy introduced by researchers have taken into consideration vertex subsets of the
underlying graph. A few such variations are minimum covering color energy, minimum
dominating color energy, color Harary energy, Rándic color energy etc [55, 56, 57, 58,
59].
No studies have been found in the literature concerning variations in color energy
corresponding to various types of coloring. Therefore, in the present work we make an
attempt to study the color energy of graphs corresponding to different coloring protocols
and, explore and analyze the values obtained from color energies with respect to different
coloring. In this view, a part of the present study deals with the study of variation of color
energy corresponding to various coloring protocols; both proper and improper coloring.
Sampathkumar et al. [16] have introduced k-partition energy EPk (G) by considering
partitions of V (G). It is defined as the sum of the absolute values of eigenvalues of L-
matrix Pk (G). The authors obtained bounds and values of EPk (G) for various classes of
graphs and their generalized complements. Some of the required results are mentioned in
the following theorem.

Theorem 1.3.10. [16] Let G be a graph and, EPk (G) and EPk (G)1(i) be the corresponding
k-partition energy of G and (G)1(i) with respect to partition Pk for k = 1, 2. Then

(i) EP1 (Kn ) = 4(n − 1) and EP1 (Kn )1(i) = 2(n − 1),

23
 Chapter 1. Introduction

√
(ii) EP1 (K1,n−1 ) = n − 2 + n2 + 12n − 12,
√
EP1 (K1,n−1 )1(i) = 2[(n − 2) + n2 − 3n + 3],
EP2 (K1,n−1 ) = 2(n − 1) and
p
EP2 (K1,n−1 )1(i) = 2[(n − 2) + (n − 2)2 + (n − 1)],
n−1
(iii) EP1 (Cn ) = |7 − n| + ∑ |1 + 6cos( 2mπ
n )| and
m=1
n−1
EP1 (Cn )1(i) = |2n − 8| + ∑ |1 + 6cos( 2mπ
n )|,
m=1

(iv) EP2 (Kr,s ) = 2(r + s − 2) and EP2 (Kr,s )1(i) = 4(r + s − 2).

In the concept of k-partition energy, where vertex partitioning is considered, the prop-
erties of elements of partition are not taken into consideration. In this direction, one of the
problem we examine in the present study is whether we can generalize of color energy
which can accommodate all the type of colorings? Responding to this question in the
affirmative, we propose a generalization to the concept of color energy of a graph.

1.4 An overview of the thesis

The work in this thesis includes the study on the concept of color energy of a graph
and its variations. The study on variations of the color energy using various coloring
protocols has been done. Further, the conceptualization of color energy in terms of P-
energy is introduced and the work is extended by determining the P-energy of several
known classes of graphs. Along with this, a few Python programs have been presented to
compute various types of color energy. The thesis is organized into six chapters.
Chapter 1 consists of prerequisites for the present study and a brief review of litera-
ture on color energy of graphs. It is a collection of basic terminologies related to graphs
and color energy of graphs. A few graph and spectral theoretic results that are required
for subsequent chapters are also presented. Moreover, the concepts related with graph
coloring and color energy are provided in detail.

24
 Chapter 1. Introduction

Chapter 2 includes the derivation of bounds for the color energy of a graph in terms
of various graph parameters such as χ(G), γ(G), ∆(G), etc. Along with this, the results
for bounds of the color energy in terms of extreme absolute values of eigenvalues of color
matrix of the graph under consideration are appended.
Chapter 3 comprises of the investigation of the color energy with respect to proper
coloring such as T -coloring, L(h, k)-coloring, radio coloring have been done. The deter-
mination of the color energy of a few improper colorings such as Hamiltonian coloring,
open neighborhood coloring and improper C-coloring also have been included in this
chapter. We also examine color equi-energetic graphs.
Chapter 4 deals with the concept of P-energy as a generalization of color energy of
a graph. The results include bounds and the derivations of the exact values of P-energy
of some classes of graphs like a complete graph, star graph, complete bipartite graph, etc
corresponding to different vertex partitions.
Chapter 5 investigates the P-energy of generalized Petersen graphs G(p, k) with
different vertex partitions. Examination of the inter-relation between the P-energy of
G(p, k) with respect to various vertex partitions and formation of a chain of inequalities
for G(p, 1) also have been done.
Chapter 6 contains the results pertaining to the P-energy of the join of graphs. The
value of P-energy is presented in terms of characteristic polynomials of the component
graphs. In particular, explicit expressions for characteristic polynomial of the join of
regular and complete bipartite graphs are determined and the P-energy of join of regular
and complete bipartite graphs are obtained.
Chapter 7 incorporates Python programs for the computation of P-energy, color
energy and several other types of energies. Further, a Python program to compare the
P-energy of generalized Petersen graphs corresponding to different vertex partitions is
also presented.

25