Edge Magic Total Labeling of Prism

A Thesis

Presented to

The Faculty of the

Department of Mathematics and Statistics

Polytechnic University of the Philippines

By:

Maebelle Rose G. Bautista

Maria Victoria C. Coronado

Jonathan C. Nava

Mary Rose A. Reyes

Adviser: Mr. Willy S. Gayo

(March 21, 2016 )

Chapter 1

Introduction

A graph G is an ordered pair < V (G), E(G) >, where V (G) is a nonempty and finite set and
E(G) may be empty or a subset of two-element subsets of V (G) that is, E(G) ⊆ V (G){2} and
V (G){2} = {{x, y}|x, y ∈ V (G)}. The set of vertices in a graph denoted by V (G) is called the
vertex set of G with elements called vertices. The set of edges in a graph denoted by E(G) is
called edge set of G with elements called edges. We may refer to these sets simply as V and
E instead of V (G) and E(G) respectively. Each edge e connects two vertices x, y ∈ V (G),
e = {x, y}, x 6= y, which can be also written as e = xy = yx. If e = xy ∈ E(G), then x and
y are the endpoints of the edge e. A vertex v is incident with an edge e if v ∈ e, then e is an
edge at v. Also, two edges are incident if they share a common endpoint. Two vertices x,
y of G are adjacent if xy is an edge of G, denoted by x ∼ y. The neighborhood of a vertex
v is the set of vertices adjacent to v, that is NG (v) = {x ∈ V (G)|vx ∈ E(G)}. The degree
of u in G is |NG (u)|, denoted by degG (u). Furthermore, δ(G) is written for the smallest of
all degrees of vertices in G, while ∆(G) indicates the largest. The order of G is defined by
|V (G)| = p and |E(G)| = q defines the size of G, we may also say G is a (p, q) − graph.

2
 3

Let G be a graph illustrated in below

a b c d

The graph G is a (4, 3)−graph having a vertex set and edge set of V (G) = {a, b, c, d}
and E(G) = {ab, bc, cd} respectively. The vertex b is incident to the edge of ab, also a and b
are adjacent. The neighborhood of b is the set NG (b) = {a, c} while degG (b) = 2.

Given a graph G with vertex set V (G) and edge set E(G), we define an edge-magic
total labeling on a graph G as a bijection ρ from V (G) ∪ E(G) onto the set {1, 2, ..., p + q}
where p = |V (G)| and q = |E(G)|, so that for each edge e = xy where e ∈ E(G) is the edge
of graph G and x, y ∈ V (G) are vertices of G, satisfies ρ(x) + ρ(xy) + ρ(y) = k for some
constant k. This constant is called magic constant or magic sum. If graph G admits an edge
magic total labeling, we may call G as an edge-magic graph. For convenience, we will use
EMTL as an abbreviation for edge-magic total labeling.

EMTL was developed from the well-known magic squares. Magic square is an n × n
array, n ≥ 3, n ∈ N and an arrangement of numbers from 1 to n2 in such a way that the
sum of the rows and columns is always equal.
In 1963, J. Sedlacĕk introduced magic labeling such that the sum of the labels of the
edges incident to a vertex is the same. In his study, he made used of simple and undirected
graph. Meanwhile, W.D. Wallis first developed the study of Edge Magic Total Labeling.
Throughout the years, many have developed further studies and conclude some theorems
and properties for EMTL and started to use it in different kinds of graph such as cycle and
complete bipartite.
4 CHAPTER 1. INTRODUCTION

The figure below illustrates the EMTL of C4 and K2,3

8 1 5

2 6

4 7 3

Figure 1.1: EMTL of C4 with k = 14

3 6 9

10 5
11 8 7 4

1 2

Figure 1.2: EMTL of k2,3 graph with k = 15

In this paper, we are going to show an algorithm of edge-magic total labeling to use in
prisms.

Let n ∈ N. A prism is a (2n, 3n) − graph, isomorphic to the cartesian product of a

complete graph of order 2 and a cycle of order n
1.1. STATEMENT OF THE PROBLEM 5

1.1 Statement of the Problem

The main objective of this paper is to develop an algorithm in labeling an edge-magic

prism and find the possible value of magic constant k.

1.2 Significance of the Study

The findings of this study will benefit other researchers who wish to extend the study of
edge magic total labeling of prism. This study is also a solution to the following:

[a] Communication network. According to Jger and Kolonko, suppose that one wants
to address all the elements in a unique way. The vertices and edges can be the nodes and
links to represent the communication network.

[b] Radar impulses. A pulse is transmitted from a source S towards a target object
O. The target reflects the pulse and the reflection is received by a detector D. From the
time span between transmission and reception of the pulse one can calculate the distance
between source and object. The problem of this measurement is that only a fraction of the
sent energy is reflected and reaches the detector. One does not know if the first received
signal was the first transmitted signal and thus the measured distance may deviate from the
real distance. One tries to minimize this inherent inaccuracy by transmitting a set of low
amplitude, narrowly defined pulses.

The timing of these pulses should be defined such that the difference in transmission time
of two signals is pairwise different. One can then match precisely individual received signals
to sent signals using the following idea. An array of detectors is distributed like a template
6 CHAPTER 1. INTRODUCTION

of the pulse. For each sent signal there is one corresponding detector placed at a distance
from the source. The distance of placement is equivalent to the timing of the corresponding
signal. Consequently, the pulse will excite all detectors at the same time when it returns. As
the difference in timing of the signals are pairwise different, at no other time more than one
signal will be received by a detector. This allows to match transmitted signals to received
signals and therefore a precise measurement. Note that a number of pulses are sent and one
needs to distinguish between the signals of different pulses.
Chapter 2

Related Studies

Sedlacĕk introduced magic labeling of graphs in 1963, in such a way that sum of the
labels of the edges incident to a vertex is the same, independent of the choice of the vertex.
Afterwards, Kotzig and Rosa formulated the study of total labeling in which the labels are
the pairwise different integers from 1 to p + q such that the sum of the labels of an edge and
its two endpoints is constant.

H. Enomoto, A. Llado, T. Nakamigawa and G. Ringel investigated whether several fam-

ilies of graphs have the property in which the vertex-labels are the integers 1,2,...,p are the
smallest possible labels that is called super edge-magic labeling, or not and give several con-
jectures.

S.M. Hedge and S. Shetty define the maximum (minimum) of all constants say M (G)
(m(G)) and introduce strong magic if m(G) = M (G), ideal magic if 1 ≤ M (G) − m(G) ≤ p
and weak magic labeling if M (G) − m(G) > p, and prove that some known classes of graphs
admit such labeling.

7
8 CHAPTER 2. RELATED STUDIES

K.A. Sugeng and M. Miller study the properties of a-vertex consecutive edge magic that
has a bijection β : V (G) ∪ E(G) → {1, 2, ..., p + q} where β is an edge-magic labeling
and β(V ) = {a + 1, ..., a + p}, 0 ≤ a ≤ q and b-edge consecutive magic graphs that has
a bijection γ : V (G) ∪ E(G) → {1, 2, ..., p + q} where γ is an edge-magic labeling and
γ(E) = {b + 1, ..., b + q}, 0 ≤ b ≤ p.

W. Wallis, E. Baskoro, M. Miller and Slamin study edge magic total labeling λ and its
dual labeling λ0 defined by λ0 (xi ) = (v + e + 1) − λ(xi ) where xi is p and for any edge xy,
λ0 (xy) = (v + e + 1) − λ(xy). Later they continue their work and study those graphs with
no edge-magic total labeling such as regular graphs.

Previous studies about edge magic labeling such as [11], [12] and [3], shown that different
cases may define if a graph is edge magic or not. These researches show results in different
types of graphs such as cycles, paths, suns, and other families of graphs are edge-magic for
odd order p and leave no results in some even order p. Thus, this paper use and develop an
algorithm that come up from previous studies and relate to another graph which is prism.
Chapter 3

Methodology and Preliminary Results

Definition 1 A graph with vertices of degree r is called an r-regular Graph.

Definition 2 A graph of order n formed by a circuit with no repeated vertices except the
initial u0 and terminal un vertices is called a Cycle, denoted by Cn , clearly, n ≥ 3.

Definition 3 A graph of order n whose vertices are adjacent to one another except the ver-
tices itself is called Complete graph, denoted by Kn .

Definition 4 Let G and H be graphs. The cartesian product of G and H is a graph denoted
by G × H such that
V (G × H) = V (G) × V (H)

and
{(u1 , v1 ), (u2 , v2 )} ∈ E(G × H)

if and only if one of the following is true:

(i) u1 = u2 and {v1 , v2 } ∈ E(H).

(ii) v1 = v2 and {u1 , u2 } ∈ E(G).

9
10 CHAPTER 3. METHODOLOGY AND PRELIMINARY RESULTS

The operation (×) motivates the researchers to use a graph generated by the cartesian
product of Cn and K2 through out this paper.

Definition 5 Let n ∈ N. A prism, denoted by Yn is a (2n, 3n)−graph isomorphic to the

graph of the cartesian product K2 × Cn , where K2 is a complete graph of order 2 and Cn is
a cycle of order n.

Definition 6 Let p = |V (G)| and q = |E(G)|. A labeling or valuation ρ of a graph is a

mapping from the graph elements to numbers. The domain of the labeling is the set of all
vertices and edges mapped to positive integers, that is, ρ : V (G) ∪ E(G) → N where N ⊆ N,
thus if N = 1, 2, . . . , 5n we call this labeling a total labeling.

Definition 7 Suppose G has a labeling ρ. The weight w of a graph element is the sum of
all labels associated with the graph elements. Thus, the weight of a vertex x is the sum of
the label of x and the labels of all incident edges, in other words

X
w(x) = ρ(x) + ρ(xy)
x∼y

and the weight of an edge xy is the sum of the label of the edge and the labels of the incident
vertices, that is,
w(xy) = ρ(x) + ρ(xy) + ρ(y)

Definition 8 Let G be a graph and let k ∈ N. We call k a magic sum or magic constant of
G if for all xi ∈ V (G), i = 1, 2, . . . , n, then k = w(xi , xi+1 ).
 11

Consider the two C3 graphs below. On the left side is a simple graph with no labeling.
Get the EMTL of C3 by labeling ρ to every vertex and edges and by the definition of EMTL
labeling must be ρ : V (C3 ) ∪ E(C3 ) −→ {1, 2, 3, 4, 5, 6}, (|V (C3 )| + |E(C3 )| = 3 + 3 = 6).
We have

x2 1
5 6

x1 x3 3 2
4

Figure 3.1: Cycle graph with 3 nodes

x1 −→ ρ(x1 ) = 3, x1 x2 −→ ρ(x1 x2 ) = 5,
x2 −→ ρ(x2 ) = 1, x2 x3 −→ ρ(x2 x3 ) = 6,
x3 −→ ρ(x3 ) = 2, x3 x1 −→ ρ(x3 x1 ) = 4,

For every edge xi xi+1 , w(xi xi+1 ) = 9 that is

ρ(x1 ) + ρ(x1 x2 ) + ρ(x2 ) = 3 + 5 + 1 = 9,

ρ(x2 ) + ρ(x2 x3 ) + ρ(x3 ) = 1 + 6 + 2 = 9,
ρ(x3 ) + ρ(x3 x1 ) + ρ(x1 ) = 2 + 4 + 3 = 9,

Since the labelings of the vertices and edges are distinct and the magic sum k=9, then
C3 is an edge-magic. Furthermore, there is no particular solution in labeling. But, there
are some theorems developed in previous researches that will ease our process in labeling
graphs.
12 CHAPTER 3. METHODOLOGY AND PRELIMINARY RESULTS

Theorem 1 Suppose a graph G is edge-magic with unknown magic constant k. Then the
complete labeling is specified by the vertex labels.
 
 
1 p+q+1 X X
k=  − ρ(x) + (ρ(x) + ρ(y)) (1)
q 2
x∈V (G) xy∈E(G)

Thus, the labeling of the vertices will be the primary step in this research.

Note that not every possible labeling is edge-magic and (1) may result in a non-integer
value of k, or in repeated labels. By restricting these cases, there are conditions that were
developed due to Theorem 1.

The necessary conditions in order that {ρ(x1 ), ρ(x2 ), . . . , ρ(xp )} = ρ(V (G)) with magic
sum k, are:

(C1) The sums ρ(xi ) + ρ(xj ), where xi xj is an edge, are all distinct.

(C2) 0 ≤ k − (ρ(xi ) + ρ(xj )) ≤ p + q when xi is adjacent to xj .

P roof.
(C1) Case 1. Let e1 = xr xs and e2 = xs xt . Then, xr ∼ xs and xs ∼ xt .
Supppose ρ(xr ) + ρ(xs ) = ρ(xs ) + ρ(xt ) =⇒ ρ(xr ) = ρ(xt ). Thus, labels of vertices are
repeated which is a contradiction.

Case 2. Let e3 = xr xs and e4 = xm xn . Then, xr ∼ xs and xm ∼ xn .

Supppose ρ(xr ) + ρ(xs ) = ρ(xm ) + ρ(xn ). Hence, if k is the magic sum,

=⇒ k − ρ(xr ) + ρ(xs )) = k − (ρ(xm ) + ρ(xn ))

=⇒ ρ(xr xs ) = ρ(xm xn ) =⇒ ρ(e3 ) = ρ(e4 ).

 13

Thus, the labels of edges are repeated which is also a contradiction.

Therefore ρ(xi ) + ρ(xj ) should be distinct.

(C2) Let xi and xj be adjacent.

The inequality k − (ρ(xi ) + ρ(xj )) > 0 is clear since k − (ρ(xi ) + ρ(xj )) > 0 =⇒ ρ(xi xj ) > 0.
Now suppose k − (ρ(xi ) + ρ(xj )) > p + q. Then, ρ(xi xj ) > p + q which is a contradiction
since p + q is the largest label. Thus, ρ(xi xj ) cannot exceed p + q.
Therefore, the inequality 0 < k − (ρ(xi ) + ρ(xj )) ≤ p + q must be satisfied when xi ∼ xj .

Definition 9 The edge-magic total labeling of Yn is a bijection

ρ : V (Yn ) ∪ E(Yn ) −→ {1, 2, . . . , 5n}

so that for each edge e = xy applies

w(xy) = ρ(x) + ρ(xy) + ρ(y) = k.

Suppose that U, V ⊆ V (Yn ) where U is the set of inner vertices and V is the set of outer
vertices. Then ui ∈ U , vi ∈ V , and ui is adjacent to vi for i = 1, 2, .., n.
Using (1) from Theorem 1, since |V (Yn )| = 2n and |E(Yn )| = 3n, (1) becomes

 
 
1  5n + 1 X X
k= − ρ(x) + (ρ(x) + ρ(y))
3n 2
x∈V (Yn ) xy∈E(Yn )
14 CHAPTER 3. METHODOLOGY AND PRELIMINARY RESULTS

Now,
X n
X n
X
ρ(x) = ρ(u1 ) + ρ(u2 ) + · · · + ρ(un ) + ρ(v1 ) + ρ(v2 ) + · · · + ρ(vn ) = ρ(ui ) + ρ(vi )
x∈V (Yn ) i=1 i=1

and

X
(ρ(x) + ρ(y)) = (ρ(u1 ) + ρ(u2 )) + (ρ(u2 ) + ρ(u3 )) + · · · + (ρ(un ) + ρ(u1 ))+
xy∈E(Yn )
(ρ(v1 ) + ρ(v2 )) + (ρ(v2 ) + ρ(v3 )) + · · · + (ρ(vn ) + ρ(v1 ))+
(ρ(u1 ) + ρ(v1 )) + (ρ(u2 ) + ρ(v2 )) + · · · + (ρ(un ) + ρ(vn ))
= (ρ(u1 ) + ρ(u1 )) + (ρ(u2 ) + ρ(u2 )) + · · · + (ρ(un ) + ρ(un ))+
(ρ(v1 ) + ρ(v1 )) + (ρ(v2 ) + ρ(v2 )) + · · · + (ρ(vn ) + ρ(vn ))+
ρ(u1 ) + ρ(u2 ) + · · · + ρ(un ) + ρ(v1 ) + ρ(v2 ) + · · · + ρ(vn )
Xn Xn X n Xn
=2 ρ(ui ) + 2 ρ(vi ) + ρ(ui ) + ρ(vi )
i=1 i=1 i=1 i=1 !
Xn Xn Xn Xn X
=3 ρ(ui ) + 3 ρ(vi ) = 3 ρ(ui ) + ρ(vi ) = 3 ρ(x)
i=1 i=1 i=1 i=1 x∈V (Yn )

Hence, for Yn we have

 
 
1  5n + 1 X X
k= − ρ(x) + 3 ρ(x)
3n 2
x∈V (Yn ) x∈V (Yn )

 
 
1  5n + 1 X
=⇒ k= +2 ρ(x) (2)
3n 2
x∈V (Yn )

Theorem 2 If G is edge-magic then,

p + q + 3 ≤ k ≤ 2(p + q)
 15

Proof. Suppose G has an edge-magic total labeling ρ. Some edge receives label p + q, so
the weight of that edge is at least 1 + (p + q) + 2. On the other hand, the edge with label 1
will have weight at most (p + q − 1) + 1 + (p + q). So,

p + q + 3 ≤ k ≤ 2(p + q)

Definition 10 The maximum magic strength of G, is defined as the maximum of all k where
the maximum is taken over all magic labelings ρ of G. M (G) = max{k : ρ is a magic labeling
of G}.

Thus, for any magic labeling ρ of a graph Yn ,

p + q + 3 ≤ m(G) ≤ k ≤ M (G) ≤ 2(p + q)

or

5n + 3 ≤ m(G) ≤ k ≤ M (G) ≤ 10n.

We may also get the possible large value of a magic constant k for each graphs using the
next definition.

Definition 11 Given a labeling ρ, the dual labeling of graph G denoted by ρ0 is defined by

ρ0 (xi ) = (p + q + 1) − ρ(xi )

and for any edge xy,

ρ0 (xy) = (p + q + 1) − ρ(xy).
16 CHAPTER 3. METHODOLOGY AND PRELIMINARY RESULTS

Observe that if ρ is a magic labeling of G with a magic sum k, then ρ0 is also a magic
labeling with magic sum k 0 = 3(p + q + 1) − k.
Chapter 4

Results and Discussions

Definition 12 Let < be a family of regular graphs such that if G is a regular graph of degree
d with vertices v, then G ∈ < if and only if there exist α, β, γ, δ ∈ N that satisfies the
following:
δ < 2α ,

d = 2α+1 β − 2α + 1,

v = 4(2α γ − δ).

Theorem 3 Suppose G is a graph with p vertices and q = 2n + 1 edges, and suppose every
vertex of G has an odd degree d. Select a positive integer δ such that for each vertex xi ,

d(xi ) = 2δ di + 1

di , define T = (2τ · Q), where τ , Q ∈ N and Q

P
for some non negative integer di . If T =
is an odd. If G has an edge-magic total labeling ρ then,

17
18 CHAPTER 4. RESULTS AND DISCUSSIONS

τ = 0 ⇒ P = 2δ P f or some P ≡ 1 mod 2

q = 2δ Qf or some Q ≡ 1 mod 2

τ = 1 ⇒ P 0 = 2δ+1 P 0 f or some P 0 ≡ 1 mod 2

q = 2δ+1 Q0 f or some Q0 ≡ 1 mod 2

τ ≥ 2 ⇒ 2δ+2 divides p and 2δ+2 divides q

Theorem 4 Suppose G is a regular graph of odd degree d with p vertices and q = 2n edges.
Write
d = 2r s + 1

where s = 2n + 1 and r ∈ N. If G is edge-magic, then 2r + 2 divides v.

Theorem 5 No member of < is edge magic.

Proof. Since Yn is a 3−regular graph having d = 3. For β ∈ N, it follows that

2 + 2α 2α 2 1 1
β= = + = +
2α+1 2α+1 2α+1 2 2α
1 1 1 1 3 1 3
But 0 < α
< 1 =⇒ < + α < =⇒ <β<
2 2 2 2 2 2 2
1 1 1 1
Then β = 1 =⇒ + α = 1 =⇒ α
= =⇒ 2 = 2α =⇒ α = 1.
2 2 2 2
Therefore α = 1, whenever β = 1. 
 19

Since δ ∈ N and α = 1, then we have δ < 2α =⇒ δ < 2 =⇒ δ = 1

Now, to find all Yn ∈ <, we have v = 2n, α = 1 and δ = 1.

(n + 2)
Then, v = 4(2α γ − δ) ⇒ 2n = 4(2γ − 1) ⇒ n = 4γ − 2 ⇒ γ = .
4

It follows that, n − 4γ = −2 −→ (∗). To find γ ∈ N, we will use the process in solving a

Linear Diophantine Equation.
• gcd(1, 4) = 1 ⇒ 1 divides 4. Hence, (∗) has a solution.
• gcd(1, 4) = 1 = 1 · 1 + 0 · 4.

(1)(−2) (0)(−2)
The particular solution is n0 = = −2 ; γ0 = = 0.
gcd(1, 4) = 1 gcd(1, 4) = 1
Thus,
 
−4
n = n0 + t ⇒ n = −2 − 4t ⇒ n = −(2 + 4t) and
 gcd(1, 4) = 1
1
γ = γ0 − t ⇒ γ = 0 − t ⇒ γ = −t, t ∈ Z.
gcd(1, 4) = 1
But note that n ≥ 3, n ∈ N.
4j + 4
Hence, n = −(2 + 4t) ⇒ n = 2 + 4j ⇒ γ = = j + 1, j ∈ N.
4

Thus, γ = j + 1 ∈ N whenever n ≡ 2(mod 4). This implies that for all n ≡ 2(mod 4),
Yn is not edge-magic. Therefore, Yn cannot have an edge-magic total labeling if and only if
p ≡ 4(mod 8), where p = |V (Yn )|.

In October 2000, K. Wijaya and E. Baskoro [9] proved that Pz × Cn is edge-magic total
for n is odd at least 3. Suppose that z = 2 and m = 2h + 1, where h ∈ N. Then P2 × Cm
is edge-magic total. Therefore, Ym admits an edge-magic total labeling. We call Ym an odd
prism whenever m is odd.
20 CHAPTER 4. RESULTS AND DISCUSSIONS

Algorithm 1 Let m be odd.

1. Draw a prism of order 2m.

2. Suppose that ui and vi be the variables for the vertices of prism where ui is the vertices
of the inner cycle, and vi is the vertices of the outer cycle.

3. Create a 2-by-m array, where the first row is the vertex labels of the inner vertices ui ,
and the 2nd row is the outer vertices vi , for i = 1, 2, . . . m.

4. Assign a label to each vertices say ρ(u1 ) = 1, ρ(u2 ) = 2, . . ., ρ(um ) = m and ρ(v1 ) =
m + 1, ρ(v2 ) = m + 2 up to ρ(vm ) = 2m.

5. Now, re-arrange the vertex labels of an array that is


 i+1

 i = 1, 3, . . . , m
 2



ρ(ui ) =



i + m + 1


 i = 2, 4, . . . , m − 1
2


 i + 3m

 i = 1, 3, . . . , m
 2



ρ(vi ) =



 i + 2m


 i = 2, 4, . . . , m − 1
2

6. Create a M3,m with the elements are the sum of the vertex labels that is,

 
 sn,1 s1,2 . . . sn−1,n 
 
M3,n =s∗
 1,1 s∗ 2,2 · · · s∗n,n  .

 
s01,2 s02,3 · · · s0n,1
 21

7. Now compute for the sum of the vertex labels as follows:

si,i+1 = ρ(ui ) + ρ(ui+1 ) , i = 1, 2 . . . , n − 1

s∗i,i = ρ(ui ) + ρ(vi ) , i = 1, 2, . . . , n

s0i,i+1 = ρ(vi ) + ρ(vi+1 ) , i = 1, 2 . . . , n − 1.

8. Check the elements of the matrix M3,m where all vertex sums are distinct.

9. If the elements of the matrix M3,m are all distinct, then this satisfies the condition
(C1). Moreover, we can now compute for the value of magic sum k and the value of
all edges using the condition (C2).

Refer to Figure 4.1.

v2

u2

u1 u3

v1 v3

Figure 4.1: Y3 graph

For n = 1, m = 3, thus Y3 is a (6, 9)−graph

 
3 4 5 
   
1 2 3 1 3 2  
 −→   −→ M3,3 = 
   
 6 7 8 

4 5 6 5 4 6  
9 10 11
22 CHAPTER 4. RESULTS AND DISCUSSIONS

Solving k through the vertex labels , then by (2) we get

  
1 5(3) + 1
k= + 2(1 + 2 + 3 + 4 + 5 + 6) = 18
3(3) 2

Observe that the elements of M3,3 are all distinct, thus the graph Y3 satisfies the C1.
Now, let us check if the graph Yn satisfies C2, for all edges of Y3 .

k − ρ(v1 ) + ρ(v2 ) = 18 − (5 + 4) = 9 =⇒ ρ(v1 v2 ) = 9

k − ρ(v2 ) + ρ(v3 ) = 18 − (4 + 6) = 8 =⇒ ρ(v2 v3 ) = 8

k − ρ(v3 ) + ρ(v1 ) = 18 − (6 + 5) = 7 =⇒ ρ(v3 v1 ) = 7

k − ρ(v1 ) + ρ(u1 ) = 18 − (5 + 1) = 12 =⇒ ρ(v1 u1 ) = 12

k − ρ(v2 ) + ρ(u2 ) = 18 − (4 + 3) = 11 =⇒ ρ(v2 u2 ) = 11

k − ρ(v3 ) + ρ(u3 ) = 18 − (6 + 2) = 10 =⇒ ρ(v3 u3 ) = 10

k − ρ(u1 ) + ρ(u2 ) = 18 − (1 + 3) = 14 =⇒ ρ(u1 u2 ) = 14

k − ρ(u2 ) + ρ(u3 ) = 18 − (3 + 2) = 13 =⇒ ρ(u2 u3 ) = 13

k − ρ(u3 ) + ρ(u1 ) = 18 − (2 + 1) = 15 =⇒ ρ(u3 u1 ) = 15

Therefore, Y3 admits an edge-magic labeling with k = 18.

11

3
9 8
14 13

1 15 2
12 7 10
5 6

Figure 4.2: Y3 edge-magic graph with k=18

 23

For n = 2, m = 5 thus Y5 is a (10, 15)−graph

 
4 5 6 7 8
   
1 2 3 4 5 1 4 2 5 3  
 (a) −→   (b) −→ M3,5 = 
 9 10 11 12 13 (c)
    

6 7 8 9 10 8 6 9 7 10  
14 15 16 17 18

Solving k through the vertex labels , then by (2) we get

  
1 5(5) + 1
k= + 2(1 + 2 + 3 + 4 + 5 + 6 + 7 + 8 + 9 + 10) = 29
3(5) 2

Since the elements of M3,5 are all distinct, it follows that the graph Y5 satisfies the C1.
Now, by checking if this satisfies the C2, we have

k − ρ(v1 ) + ρ(v2 ) = 29 − (8 + 6) = 15 =⇒ ρ(v1 v2 ) = 15

k − ρ(v2 ) + ρ(v3 ) = 29 − (6 + 9) = 14 =⇒ ρ(v2 v3 ) = 14

k − ρ(v3 ) + ρ(v4 ) = 29 − (9 + 7) = 13 =⇒ ρ(v3 v4 ) = 13

k − ρ(v4 ) + ρ(v5 ) = 29 − (7 + 10) = 12 =⇒ ρ(v4 v5 ) = 12

k − ρ(v5 ) + ρ(v1 ) = 29 − (10 + 8) = 11 =⇒ ρ(v5 v1 ) = 11

k − ρ(v1 ) + ρ(u1 ) = 29 − (8 + 1) = 20 =⇒ ρ(v1 u1 ) = 20

k − ρ(v2 ) + ρ(u2 ) = 29 − (6 + 4) = 19 =⇒ ρ(v2 u2 ) = 19

k − ρ(v3 ) + ρ(u3 ) = 29 − (9 + 2) = 18 =⇒ ρ(v3 u3 ) = 18

k − ρ(v4 ) + ρ(u4 ) = 29 − (7 + 5) = 17 =⇒ ρ(v4 u4 ) = 17

k − ρ(v5 ) + ρ(u5 ) = 29 − (10 + 3) = 16 =⇒ ρ(v5 u5 ) = 16

k − ρ(u1 ) + ρ(u2 ) = 29 − (1 + 4) = 24 =⇒ ρ(u1 u2 ) = 24

k − ρ(u2 ) + ρ(u3 ) = 29 − (4 + 2) = 23 =⇒ ρ(u2 u3 ) = 23

k − ρ(u3 ) + ρ(u4 ) = 29 − (2 + 5) = 22 =⇒ ρ(u3 u4 ) = 22

k − ρ(u4 ) + ρ(u5 ) = 29 − (5 + 3) = 21 =⇒ ρ(u4 u5 ) = 21

k − ρ(u5 ) + ρ(u1 ) = 29 − (3 + 1) = 25 =⇒ ρ(u5 u1 ) = 25

24 CHAPTER 4. RESULTS AND DISCUSSIONS

The graph Y5 satisfies the C2 for all edge labels of Y5 . Therefore, Y5 admits an edge-magic
total labeling with k = 29.

9
14 18 13
2
6 23 22 7
19 17
4 5
24 21
15 1 3 12
25
20 16
8 11 10

Figure 4.3: Y5 edge-magic labeling with k = 29.

By doing the same steps used in Y3 and Y5 , we can now generate the edge-magic labeling
in higher orders of odd prisms.
Supposed that this pattern is used in labeling a non odd prisms.

For n = 4, =⇒ Y4 is a (8, 12)−graph

 
5 4 5 6
   
1 2 3 4 1 3 2 4  
 (a) −→   (b) −→ M3,4 =   (c)
   
  8 8 10 10 
5 6 7 8 7 5 8 6  
12 13 14 13

The matrix M3,4 has an element of repeated sums that is 8, 10 and 13. Note that the
C1 requires the sums ρ(xi ) + ρ(xj ) are all distinct. Thus graph Y4 is not applicable to apply
for this pattern.
We conclude that this pattern is only applicable to odd prisms.
 25

Theorem 6 Let m be odd where m ∈ N. The Algorithm 1 can be used to get the EMTL of
prisms of order 2m with the minimum magic constant k. Thus, If ρ(ui ) = a and ρ(vi ) = b,
where a,b ∈ {1, 2, 3, . . . , 2m} then k = min(k).
26 CHAPTER 4. RESULTS AND DISCUSSIONS

OTHER EDGE-MAGIC TOTAL LABELS OF ODD PRISMS

We can generate another edge-magic labels of odd prisms by changing the function in step 5,
Algorithm 1 in two different generalizations, (B1) and (B2), then follow the same procedure.
Suppose that m = 5. Use (B1) and (B2) to generate another labeling of Y5

(B1) 
 i+1

 i = 1, 3, . . . , m
 2



ρ(ui ) =



i + m + 1


 i = 2, 4, . . . , m − 1
2

 i + 3m + 2

 i = 1, 3, . . . , m − 2
2












ρ(vi ) = i + 2m + 2 i = 2, 4, . . . , m − 1


 2










m + 1
 i=m

 
4 5 6 7 8
   
1 2 3 4 5 1 4 2 5 3  
 −→   −→ M3,5 = 
   
10 11 12 13 9 
 
6 7 8 9 10 9 7 10 8 6  
16 17 18 14 15

  
1 5(5) + 1
k= + 2(1 + 2 + 3 + 4 + 5 + 6 + 7 + 8 + 9 + 10) = 29
3(5) 2
 27

1
25 19 24
9
3 14 13 4
20 18
6 7
15 12
21 8 10 23
11
16 17
5 22 2

Figure 4.4: Y5 edge-magic labeling with k = 29.

This labeling may be the same magic sum and vertex labels to figure 4.3 but has different
edge magic labeling. Thus, the labeling of Y5 in figure 4.3 is not isomorphic to the labeling
of Y5 in figure 4.4.

(B2)

Let ρ(V (Ym )) = {m + 1, m + 2, m + 3, . . ., m + m = 2m, 3m + 1, 3m + 2, 3m + 3, . . .,

3m + m = 4m} such that,


 i + 2m + 1

 i = 1, 3, . . . , m
2




ρ(ui ) =



 i + 3m + 1


 i = 2, 4, . . . , m − 1
2

 i + 7m

 i = 1, 3, . . . , m
 2



ρ(vi ) =



 i + 6m


 i = 2, 4, . . . , m − 1
2
28 CHAPTER 4. RESULTS AND DISCUSSIONS

In finding the value of k for using (B2), from (2) we have

 
1 5m + 1
k= + 2(m + 1 + m + 2 + m + 3 + · · · + 2m + 3m + 1 + 3m + 2 + 3m + 3 + · · · + 4m)
3m 2
" m m
!#
1 5m(5m + 1) X X
= +2 (m + i) + (3m + i)
3m 2 i=1 i=1
" m m
!#
1 5m(5m + 1) X X
= + 2 m2 + i + 3m2 + i
3m 2 i=1 i=1
" m
!#
1 5m(5m + 1) 2 2
X
= + 2 m + 3m + 2 i
3m 2 i=1

   
1 5m(5m + 1) 2 2 m(m + 1)
= + 2 m + 3m + 2
3m 2 2

10m2 + 2m)
  
1 5m(5m + 1)
= +2
3m 2 2

1 25m2 + 5m + 20m2 + 4m 45m2 + 9m

 
= =
3m 2 6m

15m + 3
k= .
2

For example, for Y5

 
14 15 16 17 18
   
6 7 8 9 10 6 9 7 10 8  
 −→   −→ M3,5 = 
   
24 25 26 27 28
 
16 17 18 19 20 18 16 19 17 20  
34 35 36 37 38

  
1 5(5) + 1
k= + 2(6 + 7 + 8 + 9 + 10 + 16 + 17 + 18 + 19 + 20) = 39
3(5) 2

or simply,
15(5) + 3
k= = 39
2
 29

18
1 15 5
6
20 25 24 16
11 14
8 9
21 23
2 10 7 4
22
12 13
17 3 19

Figure 4.5: Y5 edge-magic labeling with k = 39.

11m + 3
Theorem 7 Let m ∈ N be odd. Every odd prism Ym is edge-magic with k = .
2

Proof. Let ρ(V (Ym )) = {1, 2, 3, . . . , 2m}.

Define ρ : V (G) ∪ E(G) → {1, 2, . . . , 5m} by:


 i+1

 i = 1, 3, . . . , m
 2



ρ(ui ) =



i + n + 1


 i = 2, 4, . . . , m − 1
2

 i + 3n

 i = 1, 3, . . . , m
 2



ρ(vi ) =



 i + 2n


 i = 2, 4, . . . , m − 1
2
30 CHAPTER 4. RESULTS AND DISCUSSIONS

ρ(ui ui+1 ) = 5m − i i = 1, 2, . . . , m − 1

ρ(vi vi+1 ) = 3m − i + 1 i = 1, 2, . . . , m − 1

ρ(vi vi+1 ) = 4m − i + 1 i = 1, 2, . . . , m

ρ(u1 um ) = 5m

ρ(v1 vm ) = 2m + 1

Thus, for every x, y ∈ V (Y m), and xy ∈ E(Ym ),

w(xy) = k = ρ(ui ) + ρ(ui ui+1 ) + ρ(ui+1 ) = ρ(vi ) + ρ(vi vi+1 ) + ρ(vi+1 )

= ρ(ui ) + ρ(ui vi ) + ρ(vi )

11m + 3
k = ∈N
2

Or from equation (2), we have

    2 
1 (5m + 1)! 1 5m(5m + 1) 4m + 2m
k= + 2(1 + 2 + · · · + 2m) = +2
3m (5m − 1)! 2! 3m 2 2

5m2 + 5m + 8m2 + 4m 11m + 3

= = . 
6m 2

Now, solve magic sum k for Ym , where m = 7 and m = 9.

If m = 7, for Y7 we have,

11(7) + 3 77 + 3 80
k= = = = 40.
2 2 2

Thus, the labeling of Y7 with k = 40 is shown below (Figure 4.6).

 31

14
16 22 15

10 4 11
29 35
23 28
7 1
17 21
30 34
24 3 5 27
13 31 33 8
6 32 2
18 25 26 20

9 19 12

Figure 4.6: Y7 edge-magic labeling with magic sum k = 40

If m = 9, for Y9 we have,
11(9) + 3 99 + 3 102
k= = = = 51.
2 2 2
Thus, the labeling of Y9 with k = 51 is shown below (Figure 4.7).

12
23 22
31
16 17
32 8 30
40 39
24 3 4 21
41 38
11 33 7 9 29 13
42 37
25 2 5 20
34 43 6 1 45 28
44
15 18
35 36
26 19
10 27 14

Figure 4.7: Y9 edge-magic labeling with magic sum k = 51

32 CHAPTER 4. RESULTS AND DISCUSSIONS

Theorem 8 An edge-magic regular graph of degree d, with p vertices and q edges, has magic
sum at least.
 
((d − 1)p(p + 1) + (p + q)(p + q + 1)
2q

Proof. Let G be edge-magic d − regular with p vertices and q edges. G is regular implies
that for every xi ∈ V (G), degG (xi ) = dxi = d, for 1 ≤ i ≤ p.
X
For any graph G it holds that (ρ(xy) + ρ(x) + ρ(y)) = kq. This sum contains each
xy∈E(G)
label once, and each vertex label ρ(xi ) an additional dxi − 1 times. This yields

p+q
X X
kq = i+ (dxi − 1)ρ(xi )
i=1 xi ∈V (G)

(3) .

Hence,

p+q  
X X p+q+1 X
kq = i + (d − 1) ρ(xi ) = + (d − 1) ρ(xi )
i=1
2
xi ∈V (G) xi ∈V (G)

(p + q)(p + q + 1) X
=⇒ kq = + (d − 1) ρ(xi ) (4)
2
xi ∈V (G)

1
or since q = dp,
2

X
kdp = (p + q)(p + q + 1) + 2(d − 1) ρ(xi )
xi ∈V (G)

.
p  
X X p+1
Observe that ρ(xi ) ≥ i= , then (4) becomes
i=1
2
xi ∈V (G)

 
(p + q)(p + q + 1) p+1 (p + q)(p + q + 1) + (d − 1)[p(p + 1)]
kq ≥ + (d − 1) =
2 2 2
 33

(p + q)(p + q + 1) + (d − 1)[p(p + 1)]

=⇒ k ≥ .
2q

Since prism is a 3-regular graph, then for all Yn , d=3. Also, Y11 is a (22,33)-graph where
in p = 22 and q = 33. Now, we can find the least magic sum of Y11 .

   
((3 − 1)(22)(22 + 1) + (22 + 33)(22 + 33 + 1) (1012 + 3080)
= = 62.
2(33) 66

Theorem 9 Suppose Ym has an edge-magic total labeling ρ where m ∈ N is odd, then Ym

11m + 3
has a minimum magic strength m(Ym ) = .
2

Proof. From Theorem 7

   
(d − 1)p(p + 1) + (p + q)(p + q + 1) (2)(2m)(2m + 1) + (5m)(5m + 1)
m(G) = =
2p (2)(3m)

8m2 + 4m + 25m2 + 5m 33m2 + 9m

       
3n(11m + 3 11m + 3
= = = = .
6m 6m 2(3m) 2

11m + 3
But k = and k is the magic sum whenever ρ(V (Ym )) = {1, 2, 3, . . . , 2m}.
2

11m + 3
∴ m(Ym ) =
2


34 CHAPTER 4. RESULTS AND DISCUSSIONS
19m + 3
Corollary 1 For m ∈ N where m is odd, then M (Ym ) =
2

11m + 3 19m + 3
Proof. M (Ym ) = 3(5m + 1) − = . 
2 2

Referring to the figure 4.7 where the magic sum k = 51 = m(G). Find the maximum
magic sum M (Y9 ),

19(9) + 3 174
M (Y9 ) = = = 87.
2 2

On the other hand, solving the dual labeling of Y9 gives us,

k 0 = 3(5m + 1) − k

= 3(46) − 51

= 87.

∴ The maximum magic sum M (Y9 ) and the dual label of the minimum magic sum
m(Y9 ) = 51 are both equal to k 0 = 87.
 35

34
23 24
15
30 29
14 38 16
6 7
22 43 42 25
5 8
35 13 39 37 17 33
4 9
21 44 41 26
12 3 40 45 1 18
2
31 28
11 10
20 27
36 19 32

Figure 4.8: Y9 edge-magic labeling with magic sum k = 87

36 CHAPTER 4. RESULTS AND DISCUSSIONS

FINDING THE POSSIBLE VALUES OF k

Let G be a (p, q)−graph with an EMTL ρ. Among the labels, write S as the set of vertex
labels of G where S = {ρ(xi ) : 1 ≤ i ≤ p}, and s for the sum of elements of S or simply
X
s= ρ(x). Then S can consist of the p smallest label, the p largest label, or somewhere
x∈V (G)
in between, so,
p p+q
X X
i≤s≤ i
i=1 i=1+q
   
p+1 p+1
≤ s ≤ pq + .
2 2

Hence, for m ∈ N is odd, Ym ,

   
2n + 1 2 2n + 1
=⇒ ≤ s ≤ 6n + .
2 2
Now, for the set of the vertex labels of Yn ,
write S = {(ρ(ui )), (ρ(vi ))} = {(ρ(u1 ), ρ(u2 ), . . . , ρ(un )), (ρ(v1 ), ρ(v2 ), . . . , ρ(vn ))}.
1
Note that for Yn , k = [5n(5n + 1) + 4s].
6n

• For Y3 , a (6, 9) − graph.

1
We have 21 ≤ s ≤ 75 and k = (120 + 2s). For k ∈ N, s ≡ 3(mod9).
9
Thus the possibilities are s = 21, 30, 39, 48, 57, 66, 75
with k = 18, 20, 22, 24, 26, 28, 30 respectively. k = 18, s = 21, S = {(1, 3, 2), (5, 4, 6)}.

k = 20, s = 30, no solutions.

k = 22, s = 39, S = {(1, 6, 11), (7, 12, 2)}.
k = 24, s = 48, S = {(1, 8, 12), (9, 13, 5)}.
k = 26, s = 57, S = {(4, 9, 14), (15, 5, 10)}.
k = 28, s = 66, no solutions.
k = 30, s = 75, S = {(15, 13, 14), (11, 12, 10)}.
 37

• For Y5 , a (10, 15) − graph.

1
15 ≤ s ≤ 205 and k = (325 + 2s) ⇒ s ≡ 10( mod 15).
15
k = 29, s = 55, S = {(1, 4, 2, 5, 3), (8, 6, 9, 7, 10)}.
k = 31, s = 70, S = {(1, 5, 2, 7, 10), (19, 3, 13, 6, 4}.
k = 33, s = 85, S = {(1, 7, 2, 11, 18), (13, 3, 9, 16, 5)}.
k = 35, s = 100, S = {(1, 9, 2, 12, 15), (14, 3, 22, 6, 16)}.
k = 37, s = 115, S = {}.
k = 39, s = 130, S = {(1, 14, 4, 12, 25), (22, 7, 15, 19, 11)}.
k = 41, s = 145, S = {}.
k = 43, s = 160, S = {(25, 17, 24, 14, 11), (12, 23, 4, 20, 10)}.
k = 45, s = 175, S = {(25, 19, 24, 15, 8), (13, 23, 17, 10, 21)}.
k = 47, s = 190, S = {(25, 21, 24, 19, 16), (7, 23, 13, 20, 22)}.
k = 49, s = 205, S = {(25, 22, 24, 21, 23), (18, 20, 17, 19, 16)}.
Chapter 5

Summary and Recommendations

In this paper, the edge magic total labeling prism graphs may or may not have a edge-magic
labeling. Thus, we are able to conclude with some observations.

1. A prism graph denoted by Yn is said to be edge-magic if for m ∈ N is odd, n = m.

11m + 3
• If m is odd, the minimum magic sum of Ym is m(k) = .
2
15n + 3
• If m is odd, then Ym has a magic sum k =
2
19m + 3
• If m is odd, the maximum magic sum of Ym is M (k) =
2

2. We can get the edge magic total labeling of the maximum magic sum k by applying
the dual labeling of the graph.

3. A prism graph Yn is not edge-magic if n ≡ 2(mod4).

4. Yn cannot have an edge-magic total labeling if and only if p ≡ 4(mod8)

38
5.1. RECOMMENDATIONS 39

5.1 Recommendations

Based on the results derived, the researchers come up with the following recommendations:

1. Get the possible magic sum of prism such that min(k) < k < max(k) and show their
edge magic total labeling.

2. Use other graph and give its edge magic total labeling.

3. We found a labeling of Yn , for n = 4. Prove or disprove that Yn is edge-magic if

n ≡ 0(mod8).
References

[1] Enomoto, H., Llado, A., Nakamigawa, T., and Ringel, E., Super Edge Magic Graphs
(1998).

[2] Hegde, S.M. and Shetty, S., On Magic Graphs (2003).

[3] Jager, G., and Kolonko, M., Totally Magic Graphs (2012).

[4] Kotzig, A. and Rosa, A., Magic Valuations of Complete Graphs (1972).

[5] Sedlacĕk, J., (Math Slov) On Magic Graphs (1976).

[6] Sugeng, K. A. and Miller, M., On Consecutive Edge Magic Total Labeling of Graphs
(2006).

[7] Vasuki, S. A., and Jeyanthi, P., Magic Strength of a Graph (1999).

[8] D. Craft and E. H. Tesar, On a question by Erdos about edge-magic graphs, Discrete
Math., 207 (1999) 271-276.

[9] Wallis, W., Baskoro, E., Miller, M., and Slamin, Edge-Magic Total Labelings (2000).

[10] K. Wijaya and E. Baskoro, Edge-magic labelings of a product of two graphs (in
indonesian), Proc. Seminar MIPA, ITB Bandung Indonesia, October 2000.

[11] A. Gallian, A Dynamic Survey of Graph Labeling (1996)

40
Chapter 6

Revisions

(1) pg. 4 - ”we are going to find and study ...”

(2) pg. 10 - definition 6, def. 8

(3) pg. 16

(4) pg. 18 - theorem 5 proof

41