University of Newcastle

Faculty of Science and Information Technology

School of Mathematics and Physical Science

A thesis submitted in total fulfilment of the requirement for the

degree of Doctor of Philosophy

Graph Labeling Techniques

Supervisors: Student:
Dr. Joe Ryan Dushyant Tanna
Dr. Thomas Kalinowski B.Sc., M.Sc., M.Phil.
C3194176

External Co-supervisor:
A/Prof. Andrea Semaničová-Feňovčı́ková

November, 2017
 Statements

Statement of Originality

The thesis contains no material which has been accepted for the award of any other degree
or diploma in any university or other tertiary institution and, to the best of my knowledge and
belief, contains no material previously published or written by another person, except where
due reference has been made in the text. I give consent to the final version of my thesis being
made available worldwide when deposited in the University’s Digital Repository, subject to the
provisions of the Copyright Act 1968.

Statement of Authorship

I hereby certify that the work embodied in this thesis contains published papers of which I
am a joint author. I have included as a part of the thesis a written statement, endorsed by my
supervisor, attesting to my contribution to the joint publication.

Signed:
DUSHYANT TANNA

The candidate has six papers, of which he is a joint author, embodied in the thesis. As
principal supervisor I can attest that the candidate’s contribution, in all the cases were about
75% of the final article.

Signed:
Dr. JOE RYAN

1
 List of Publications

Substantial sections of the following chapters are based upon the following articles submitted
to the nominated journals;

1. Tanna, D., Ryan, J., Semaničová-Feňovčı́ková, A., Edge Irregular Reflexive Labeling of
Prisms and Wheels, Electronic Notes in Discrete Mathematics (Elsevier), 2016.

2. Tanna, D., Ryan, J., Semaničová-Feňovčı́ková, A., Bača, M., Vertex Irregular Reflexive
Labeling of Prisms and Wheels, Electronic Journal of Graph Theory and Applications,
2017.

3. Bača, M., Irfan, M., Ryan, J., Semaničová-Feňovčı́ková, A., Tanna, D., Note on Re-
flexive Edge Irregular Labelings of Graphs, AKCE International Journal of Graphs and
Combinatorics, 2017.

4. Bača, M., Irfan, M., Ryan, J., Semaničová-Feňovčı́ková, A., Tanna, D., On Reflexive Edge
Irregular Labelings for the Generalized Friendship Graphs, Turkish Journal of Mathemat-
ics, 2017.

5. Dafik, Slamin, Tanna, D., Semaničová-Feňovčı́ková, A., Bača, M., Constructions of H-

antimagic Graphs Using Smaller Edge-antimagic Graphs, Ars Combinatoria, 2017.

6. Ryan, J., Munushinge, B., Tanna, D., Reflexive Irregular Labelings, Preprint, 2014.

The following works are in still progress at the time of writing thesis and so it is not included
here;

1. Tanna, D., Ryan, J., Semaničová-Feňovčı́ková, A., Vertex Irregular Reflexive Labeling of
Circulant Graphs, Preprint, (2017).

2. Tanna, D., Ryan, J., Semaničová-Feňovčı́ková, A., Vertex Irregular Reflexive Labeling of
Ladders and Generalised Petersen Graph, Preprint, (2017)

2
 Dedication

To my loving wife, Zenit, my whole family and to the memory of my supervisor, Mirka Miller.

3
 Acknowledgements
First and foremost I would like to acknowledge my sincere gratitude towards my supervisors,
Emeritus Professor Mirka Miller, Dr. Joe Ryan, Dr. Thomas Kalinowski, Laureate Professor
Jonathan Borwein and A/Professor Andrea Semaničová-Feňovčı́ková, who have supported me
abundantly with their encouragement, guidance and patience.
Mirka, it has always been an honour to be your Ph.D. student. She has been the real reason
and support for completion of my Ph.D. Her valuable advices and teachings has proved to be
a catalyst throughout the journey of Ph.D. Her inclination and passion towards research also
motivated and kept me ongoing during the tough times of my Ph.D. pursuit.
My debt of gratitude to Joe for always being there for me and making my vision turn
to reality. Joe has always helped, motivated and guided me through out my candidature. I
appreciate all his contribution in terms of time and efforts that he has invested to make my
Ph.D. fruitful and productive. His rigorous guidance and ever-friendly nature has helped to
earn better insights towards mathematical problem solving. Your encouragements and unfailing
support were crucial in overcoming the setbacks and hurdles from start to completion of Ph.D.
at University of Newcastle.
I owe my special thanks to A/Professor Andrea Semaničová-Feňovčı́ková, for her vital sugges-
tions and her expert knowledge that has immensely contributed to my candidature. Discussions
with her were always thought provoking. She helped me with her suggestions at all stages in
my writing journey. Her promptness in replying to my queries and persistent problem solving is
appreciable. She stood behind me and provided continuous stimulation throughout the process.
Thanks a lot, without your help it would not be possible. A big thanks to Dr. Thomas Kali-
nowski and Laureate Professor Jonathan Borwein for being my co-supervisor. Advices given by
Thomas were precious and it always lead me to good research directions.
I would like to bestow my gratitude to the University of Newcastle for providing scholarships
(UNIPRS and UNRSC 50-50) to pursue my dream. I am thankful to International Leadership
Experience and Development (ILEAD) Plus Program for the generous support for participating
in various conferences. I acknowledge my gratitude to the Faculty of Sciences for the Conference
scholarship. My sincere thanks to the School of Mathematical and Physical Sciences, all the staff,
both academic and administrative staff who created a healthy and comfortable environment for
me to pursue my research. Many thanks to all of them who helped through the completion of this
endeavor with their support. Special thanks to Vicki Gumbleton for giving me the opportunity
for tutoring.
I convey my gratitude to my wife, Zenit. She has always been by my side when I needed her
the most with her endless support. She is the pillar of my ambition and all hard work. Thanks
to her that I am able to complete my Ph.D. She is the reason for my success. Big thanks to my
in-laws, mom, dad, brothers, sisters and whole family. Their encouragement and support helped
me in completion of my Ph.D. Last but not least, I thank God Almighty for blessing me with
this opportunity to pursue my research and for giving me sound body and mind to complete
this successfully.

4
Contents

List of Figures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
List of Tables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9

1 Introduction 11
1.1 Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
1.2 Research Objectives . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
1.3 Outline of the Thesis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13

2 Basic Concepts and Literature Review 15

2.1 Basic Graph Theory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
2.2 An Introduction to Graph Labelings with Examples . . . . . . . . . . . . . . . . 20

3 Edge Irregular Reflexive Labeling 35

3.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35
3.2 Edge Irregular Reflexive Labeling . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
3.3 Edge Irregular Reflexive Labeling of Prisms . . . . . . . . . . . . . . . . . . . . . 44
3.4 Edge Irregular Reflexive Labeling of Wheels . . . . . . . . . . . . . . . . . . . . . 46
3.5 Edge Irregular Reflexive Labeling of Join of Graphs . . . . . . . . . . . . . . . . 50
3.6 Edge Irregular Reflexive Labeling for Generalised Friendship Graphs . . . . . . . 57
3.7 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66

4 Vertex Irregular Reflexive Labeling 67

4.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67
4.2 Vertex Irregular Reflexive Labeling for Some Graphs . . . . . . . . . . . . . . . . 69
4.3 Vertex Irregular Reflexive Labeling of Prisms . . . . . . . . . . . . . . . . . . . . 78
4.4 Vertex Irregular Reflexive Labeling of Wheels . . . . . . . . . . . . . . . . . . . . 81

5
 4.5 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 89

5 H-antimagic Labeling 91
5.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 91
5.2 Partitions with Determined Differences . . . . . . . . . . . . . . . . . . . . . . . . 92
5.3 Counting the Upper Bound of the Difference d . . . . . . . . . . . . . . . . . . . 94
5.4 Main Result . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 97
5.5 Special Families of Graphs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 99
5.6 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 102

6 Applications of Graph Labelings 103

6.1 Application of Graph Labeling . . . . . . . . . . . . . . . . . . . . . . . . . . . . 103
6.1.1 Linear Congruence . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 104
6.1.2 Irregular Reflexive Labeling . . . . . . . . . . . . . . . . . . . . . . . . . . 108
6.2 Conjectures and Open Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . 109
6.2.1 Future Work . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 110
6.3 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 111

6
List of Figures

2.1 A simple graph of order 7 and size 11 . . . . . . . . . . . . . . . . . . . . . . . . 16

2.2 Pseudograph . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
2.3 A 2-regular graph . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
2.4 Complete graph K5 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
2.5 A bipartite graph . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
2.6 Complete bipartite graph K3,5 . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
2.7 Wheel W4 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
2.8 Helm H4 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
2.9 Web graph W4 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
2.10 Fan graph F7 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
2.11 Ladder L5 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
2.12 Friendship graph f6 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
2.13 Graceful labeling of wheel W4 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
2.14 Harmonious labeling of wheel W4 . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
2.15 Supermagic labeling of wheel W4 . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
2.16 Vertex-magic total labeling of wheel W4 . . . . . . . . . . . . . . . . . . . . . . . 25
2.17 Super edge-magic total labeling of cycle C5 . . . . . . . . . . . . . . . . . . . . . 26
2.18 Antimagic labeling of wheel W4 . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
2.19 A (25,1)-vertex-antimagic total labeling of wheel W4 . . . . . . . . . . . . . . . . 28
2.20 Super (10, 2)-edge-antimagic total labeling of cycle C5 . . . . . . . . . . . . . . . 29
2.21 Super (39, 1)-C3 -antimagic total labeling of wheel W4 . . . . . . . . . . . . . . . 30
2.22 Edge irregular total labeling of cycle C5 . . . . . . . . . . . . . . . . . . . . . . . 32
2.23 Vertex irregular total labeling of cycle C5 . . . . . . . . . . . . . . . . . . . . . . 33

3.1 Edge irregular reflexive labeling of K1,6 . . . . . . . . . . . . . . . . . . . . . . . 38

7
3.2 Edge irregular reflexive labeling of K1,5 . . . . . . . . . . . . . . . . . . . . . . . 38
3.3 Edge irregular reflexive labeling of P7 . . . . . . . . . . . . . . . . . . . . . . . . 39
3.4 Edge irregular reflexive labeling of P8 . . . . . . . . . . . . . . . . . . . . . . . . 39
3.5 Edge irregular reflexive labeling of C3 . . . . . . . . . . . . . . . . . . . . . . . . 40
3.6 Edge irregular reflexive labeling of C10 . . . . . . . . . . . . . . . . . . . . . . . . 42
3.7 Edge irregular reflexive labeling of C12 . . . . . . . . . . . . . . . . . . . . . . . . 42
3.8 Edge irregular reflexive labeling of C4 2C3 . . . . . . . . . . . . . . . . . . . . . . 45
3.9 Edge irregular reflexive labeling of D8 . . . . . . . . . . . . . . . . . . . . . . . . 46
3.10 Edge irregular reflexive labeling of D9 . . . . . . . . . . . . . . . . . . . . . . . . 46
3.11 Edge irregular reflexive labeling of W3 . . . . . . . . . . . . . . . . . . . . . . . . 47
3.12 Edge irregular reflexive labeling of W4 . . . . . . . . . . . . . . . . . . . . . . . . 47
3.13 Edge irregular reflexive labeling of W5 . . . . . . . . . . . . . . . . . . . . . . . . 47
3.14 Edge irregular reflexive labeling of W6 . . . . . . . . . . . . . . . . . . . . . . . . 47
3.15 Edge irregular reflexive labeling of W10 . . . . . . . . . . . . . . . . . . . . . . . . 49
3.16 Edge irregular reflexive labeling of W12 . . . . . . . . . . . . . . . . . . . . . . . . 49
3.17 Edge irregular reflexive labeling of B10 . . . . . . . . . . . . . . . . . . . . . . . . 49
3.18 Edge irregular reflexive labeling of F12 . . . . . . . . . . . . . . . . . . . . . . . . 49
3.19 Edge irregular reflexive labeling of P2 ⊕ (2K1 ) . . . . . . . . . . . . . . . . . . . . 51
3.20 Edge irregular reflexive labeling of C4 ⊕ (2K1 ) . . . . . . . . . . . . . . . . . . . 54
3.21 Edge irregular reflexive labeling of f3,2 . . . . . . . . . . . . . . . . . . . . . . . . 58
3.22 Edge irregular reflexive labeling of f5,2 . . . . . . . . . . . . . . . . . . . . . . . . 63
3.23 Edge irregular reflexive labeling of f5,3 . . . . . . . . . . . . . . . . . . . . . . . . 63

4.1 Vertex irregular reflexive labeling of K1,6 . . . . . . . . . . . . . . . . . . . . . . . 70

4.2 Vertex irregular reflexive labeling of K1,5 . . . . . . . . . . . . . . . . . . . . . . . 70
4.3 Vertex irregular reflexive labeling of P7 . . . . . . . . . . . . . . . . . . . . . . . . 74
4.4 Vertex irregular reflexive labeling of P8 . . . . . . . . . . . . . . . . . . . . . . . . 74
4.5 Vertex irregular reflexive labeling of C10 . . . . . . . . . . . . . . . . . . . . . . . 78
4.6 Vertex irregular reflexive labeling of C12 . . . . . . . . . . . . . . . . . . . . . . . 78
4.7 Vertex irregular reflexive labeling of K6 . . . . . . . . . . . . . . . . . . . . . . . 79
4.8 Vertex irregular reflexive labeling of D9 . . . . . . . . . . . . . . . . . . . . . . . 81
4.9 Vertex irregular reflexive labeling of D12 . . . . . . . . . . . . . . . . . . . . . . . 81

8
4.10 Vertex irregular reflexive labeling of W3 . . . . . . . . . . . . . . . . . . . . . . . 83
4.11 Vertex irregular reflexive labeling of W4 . . . . . . . . . . . . . . . . . . . . . . . 83
4.12 Vertex irregular reflexive labeling of W12 . . . . . . . . . . . . . . . . . . . . . . . 84
4.13 Vertex irregular reflexive labeling of F3 . . . . . . . . . . . . . . . . . . . . . . . . 86
4.14 Vertex irregular reflexive labeling of F4 . . . . . . . . . . . . . . . . . . . . . . . . 86
4.15 Vertex irregular reflexive labeling of B4 . . . . . . . . . . . . . . . . . . . . . . . 87
4.16 Vertex irregular reflexive labeling of B10 . . . . . . . . . . . . . . . . . . . . . . . 88
4.17 Vertex irregular reflexive labeling of F12 . . . . . . . . . . . . . . . . . . . . . . . 88

6.1 Supermagic labeling of caterpillar of order 7 and size 6 . . . . . . . . . . . . . . . 107

6.2 Supermagic labeling of caterpillar of order 11 and size 10 . . . . . . . . . . . . . 107

9
 Abstract

We give some background to the labeling schemes like graceful, harmonious, magic, an-
timagic and irregular total labelings. Followed by this we give some preliminary results and
open problems in these schemes. We will introduce a new branch of irregular total labeling,
irregular reflexive labeling. This new labeling technique has few variations on vertices labels
from irregular total labeling. They are,

• The vertices labels are non negative even integers.

• The vertex label 0 is permissible, representing the vertex without loop.

The vertex (edge) irregular reflexive labeling is a total irregular labeling with above con-
ditions on vertices labeling such that the vertices (edges) weights are distinct. The idea is to
use minimum possible labels for vertices (edges) and thus keeping the reflexive vertex (edge)
strength as low as possible.
We believe that this new technique is closer in concept to the original irregular labeling as pro-
posed by Chartrand et al., since the vertex labels are also being used to represent edges(loops).
Again the objective is to minimize the total strength by using the smallest vertices/edges
labels. We will give edge and vertex irregular reflexive strengths for many graphs such as paths,
cycles, stars, complete graphs, prisms, wheels, baskets, friendship graphs, join of graphs and
generalised friendship graphs and present labeling techniques for these graphs.
We also describe edge covering, H-edge covering, H-magic and H-antimagic graphs and prove
some theorems based on these concepts. Many results have been established for construction of
H-antimagic labelings of graphs we will use the partitions of a set of integers with determined
differences, the upper bound of the difference d if the graph GH is super (a, d) − H-antimagic,
establishment of connection between H-antimagic labelings and edge-antimagic total labelings.
We have also posed some open problems.
Finally we address why study of graph labeling is important by explaining some applications
of graph labeling and give some open problems and conjectures.

10
Chapter 1

Introduction

1.1 Overview

Magic labeling is considered to be the oldest labeling techniques. Many researcher have
introduced labeling schemes to generalize the idea of a magic square. A magic square of order n
is n × n array of integers 1, 2, . . . , n2 such that the sum of numbers along any row, column and
main diagonals is a fixed constant that equals to n(n2 + 1)/2. In 1963, Sedláček [102] pointed
out the correspondence between a magic square of order n and magic labeling of a complete
bipartite graph Kn,n . He discovered that if we label every edge ui vj of Kn,n with the number
from ith row and j th column of the magic square of order n, we obtain the supermagic labeling
of Kn,n . Sedláček [102] in 1963 defined a graph to be magic if it has an edge labeling, with the
range of the real numbers so that the sum of the edge labels around any vertex is always the
same, a constant, independent of the choice of a vertex.
Ringel, [97], in 1963 posed a conjecture that for any tree T with n vertices, the complete
graph K2n−1 can be decomposed into 2n − 1 trees isomorphic to T . Rosa, [99], introduced
β-labeling in 1967. Rosa conjectured that if it was possible that every tree was labeled with
β-labeling then Ringel’s conjecture holds.
Rosa, [99], called a function f a β-valuation of a graph G with q edges if f is an injection
from the vertices of G to the set {0, 1, . . . , q} such that, when each edge xy between vertices x
and y is assigned the label |f (x) − f (y)|, the resulting edge labels are distinct.
Later in 1972 Golomb, [50] called β-labeling as graceful labeling which is a popular term now.
Rosa posed a conjecture, Graceful Tree Conjecture which claims that “every tree is graceful”.
Rosa showed that if every tree is graceful then Ringel’s conjecture holds. Since then, researchers
have been trying to prove Ringel’s conjecture through the Graceful Tree Conjecture. So far no
one is able to solve this conjecture for trees in general but there has been several classes of trees
for which this conjecture holds true and as a result there are various types of trees, which are
defined in Section 2.1.
Harmonious labeling, naturally arose in the study by Graham and Sloane, [51] of modular
versions of additive bases problems stemming from error-correcting codes in 1980. They defined

11
a graph G with q edges to be harmonious if there is an injection f from the vertices of G to the
group of integers modulo q, Zq = {0, 1, . . . , q − 1} such that when each edge xy between vertices
x and y is assigned the label (f (x) + f (y)) (mod q), the resulting edge labels are distinct. They
proved that almost all graphs are not harmonious, [51].
Antimagic labeling is opposite to magic labeling. Here the weights of the elements of graph
have to be pairwise distinct. Antimagic labeling was first introduced by Hartsfield and Ringel
[54] in 1990. They called a graph G with q edges to be antimagic if there is a bijective function
f from edges of G to the set {1, 2, . . . , q} such that all vertex weights are pairwise distinct.
Hartsfield and Ringel [54] conjectured that all graphs except K2 are antimagic. Many re-
searcher have tried to prove this conjecture. Using probabilistic method and techniques of
analytical number theory, Alon et al., [7] showed that this conjecture is true for all graphs hav-
ing minimum degree Ω(log |V (G)|). They also proved that if G is a graph with |V (G)| ≥ 4 and
maximum degree ∆(G) = |V (G)| − 2 then G is antimagic. Hartsfield and Ringel proved that
many classes of graphs are antimagic. In [54], it is proved that paths Pn , n ≥ 3, cycles Cn ,
wheels Wn and complete graphs Kn , n ≥ 3 are antimagic. In this thesis, we will discuss different
types of antimagic labeling and present results in a new variation, H-antimagic.
Irregular labeling was introduced by Chartrand et al., in [33]. They posed the following
problem, “assign positive integer labels to the edges of a simple connected graph of order at
least 3 in such a way that the graph becomes irregular, that is, the label sums (weights) at
each vertex are distinct. What is the minimum value of the largest label over all such irregular
assignments?”Such a labeling is known as irregular labeling and the minimum value of largest
label over all possible irregular labeling is known as irregularity strength.
Bača et al., [20] defined vertex (edge) irregular total k-labeling of a graph G. Here the
vertices and edges both are labelled from integers {1,2,. . . ,k} so that the vertex (edge) weights
are pairwise distinct.
We present here a new type of irregular total labeling, known as vertex (edge) irregular
reflexive labeling. This new labeling technique has few variation on vertices labels from irregular
total labeling. They are,

• The vertex labels have be non negative even integers.

• The vertex label 0 is permissible, representing a vertex without a loop.

The vertex (edge) irregular reflexive labeling is a total irregular labeling with the above
condition on vertex labeling such that the vertices (edges) weights are distinct. The idea is to
use minimum possible labels for vertices (edges) and thus keeping the reflexive vertex (edge)
strength as low as possible.
We believe that this new technique is closer in concept to the original irregular labeling as pro-
posed by Chartrand et al., since the vertex labels are also being used to represent edges(loops).
Again the objective is to minimize the reflexive strength total weight by using the smallest
maximum vertex/edge labels. We will give results for edge and vertex irregular reflexive labelings
for many graphs such as paths, cycles, stars, complete graphs, prisms, wheels, baskets, friendship
graphs, join of graphs and generalised friendship graphs.

12
 We also describe edge covering, H-edge covering, H-magic and H-antimagic graphs and prove
some theorems based on these concepts. Some results have been established for construction of
H-antimagic labelings of graphs. We will use the partitions of a set of integers with determined
differences, the upper bound of the difference d if the graph GH is super (a, d)−H-antimagic. We
have also established the connection between H-antimagic labelings and edge-antimagic total
labelings.
Graph labeling is a rich area. There are other labeling techniques like radio labeling, sum
labeling, prime labeling, cordial labeling, filicitous labeling and many more. But all these
techniques must necessary fall beyond the scope of this thesis. For details study about labeling
techniques, refer [47].
Finally, we address why study of graph labeling is important by explaining some applications
of graph labeling and finish with some open problems and conjectures and suggestions for future
research endeavours.

1.2 Research Objectives

Overall research aims of study are to investigate graph labelings and their applications. In
particular:

• Study various graph labeling schemes.

• Extend existing methods and develop new techniques for investigating graph labelings.

• Find more applications of existing graph labelings.

• Motivated by real life applications, design new graph labelings.

1.3 Outline of the Thesis

This thesis is structured as follows.

Chapter 2 on page 15: Basic Concepts and Literature Review This chapter contains the
necessary and useful concepts from graph theory and an extended literature review of various
dimensions needed for the thesis.
Chapter 3 on page 35: Edge Irregular Reflexive Labeling In this chapter we will intro-
duce the edge irregular reflexive labeling and we will present results for reflexive edge strength
of some graphs.
Chapter 4 on page 67: Vertex Irregular Reflexive Labeling In this chapter we will
introduce vertex irregular reflexive labeling and we will show results for reflexive vertex strength
of some graphs.
Chapter 5 on page 91: H-antimagic Labeling In this chapter we will discuss partitions of
integers with determined differences and we will obtain some new results.

13
Chapter 6 on page 103: Applications of Graph Labelings In this chapter we will explain
why study of graph labeling is important, pose some open problems and conjectures and discuss
possible future directions.
In this thesis, all original results are indicated by the symbol ♦.

14
Chapter 2

Basic Concepts and Literature

Review

2.1 Basic Graph Theory

In this section, we introduce the terminology, definitions and notations that will be used
throughout this thesis.
A graph G is an ordered triplet (V (G), E(G), ψG ) consisting of a non-empty set V (G) of
vertices, a set E(G), disjoint from V (G), of edges, and an incidence function ψG that associates
with each edge of G an unordered pair of (not necessarily distinct) vertices of G. The graph
is generally denoted by G = G(V, E). If the vertex set V (G) and edge set E(G) are finite sets
than we call graph G a finite graph. Otherwise G is called an infinite graph.
In pictorial form the vertices are usually represented by points on a plane and edges are
represented by lines, connecting two vertices. A graph which can be drawn on a plane in such
a way that its edges intersect only at their end vertices is called a planar graph.
If two vertices x and y are connected by one or more edges than these vertices are said to be
adjacent vertices. It is often convenient to write adjacent vertices as x ∼ y. If an edge connects
with two vertices then that edge is said to be incident with those vertices and vice versa. The
order p of a graph is the number of vertices in a graph and the size q of a graph G is the number
of edges in a graph. The graph of order p and size q is also denoted by (p, q)-graph. In Figure
2.1, the order of the graph is 7 and the size of the graph is 11.
An edge connecting a vertex to itself is called a loop, while a multiple edge in a graph means
more than one edge exists between a pair of vertices. In Figure 2.2, e1 is a loop while e10 and
e11 are multiple edges. A graph is simple if it has no loops and multiple edges. A graph that
contains multiple edges is called a multigraph, while a pseudograph may contain loops as well
as multiple edges. Figure 2.2 gives an example of a pseudograph. Figure 2.1 presents a simple
graph.
The degree deg(v) of a vertex v in G is the number of edges incident with v, each loop is

15
counted as two edges when calculating the degree of a vertex.
We denote by δ(G) and ∆(G) the minimum and maximum degrees, respectively, of vertices
of G. A vertex with degree 1 is called a pendant vertex and a vertex with degree 0 is called an
isolated vertex. If a graph G has vertices v1 , v2 , . . . , vn , the monotonic non increasing sequence
(deg(v1 ), deg(v2 ), . . . , deg(vn )) is called the degree sequence of G. A graph is k-regular if deg(v) =
k for all vertices in G; a regular graph is one that is k-regular for some k. Figure 2.3 shows a
2-regular graph.

Figure 2.1: A simple graph of order 7 and size 11

v4 v7

e5 e4 e8

e7
e1 v1 e6 v3 v5 e10 e11

e2 e3 e9

v2 v6

Figure 2.2: Pseudograph

Figure 2.3: A 2-regular graph Figure 2.4: Complete graph K5

16
 Two graphs G and H are said to be isomorphic, written as G ∼ = H, if there are bijections
θ : V (G) → V (H) and φ : E(G) → E(H) such that for the incidence function ψG (e) = uv if and
only if ψH (φ(e)) = θ(u)θ(v); for edge e and vertices u, v then such a pair (θ, φ) of mappings is
called an isomorphism between G and H.
A simple graph in which each pair of distinct vertices is joined by an edge is called a complete
graph. A complete graph with p vertices is denoted by Kp . Figure 2.4 depicts the complete graph
K5 .
A bipartite graph is one whose vertex set can be partitioned into two subsets X and Y , so
that each edge has one end vertex in X and other end vertex in Y . Such a partition (X, Y ) is
called a bipartition of the graph. A complete bipartite graph is a simple bipartite graph with
bipartition (X, Y ) in which each vertex of X is joined to each vertex of Y ; if |X| = m and
|Y | = n, such a graph is denoted by Km,n .
Figure 2.5 presents a bipartite graph and Figure 2.6 gives the complete bipartite graph K3,5 .

Figure 2.5: A bipartite graph Figure 2.6: Complete bipartite graph K3,5

A graph H is a subgraph of G, written as H ⊆ G, if V (H) ⊆ V (G), E(H) ⊆ E(G) and the

incidence function ψH is the restriction of the incidence function ψG to E(H). When H ⊆ G
but H 6= G, we write H ⊂ G and call H a proper subgraph of G. If H is a subgraph of G,
G is a supergraph of H. A spanning subgraph (or spanning supergraph) of G is a subgraph (or
supergraph) H with V (H) = V (G).
We say that two graphs G1 and G2 are disjoint if they have no vertex in common, and
edge-disjoint if they have no edge in common. The union G1 ∪ G2 of two graphs G1 and G2 is
the graph with vertex set V (G1 ) ∪ V (G2 ) and edge set E(G1 ) ∪ E(G2 ). The Cartesian product
of two graphs G1 and G2 , denoted by G1 2G2 , is the graph with vertex set V (G1 ) × V (G2 ), in
which (u1 , v1 ) is adjacent to (u2 , v2 ) if and only if either u1 = u2 and v1 v2 ∈ E(G2 ) or v1 = v2
and u1 u2 ∈ E(G1 ).
A walk in G is a finite non-null sequence W = v0 e1 v1 e2 v2 . . . ek vk , whose terms are alternately
vertices and edges, such that, for 1 ≤ i ≤ k, the end vertices of ei are vi−1 and vi . The vertices
v0 and vk are called the origin and terminus of W , respectively and v1 , v2 , . . . , vk−1 are internal
vertices. The integer k is called the length of W . Given two walks W1 = v0 e1 v1 e2 v2 . . . ek vk
and W2 = vk ek+1 vk+1 . . . en vn , concatenation of W1 and W2 , denoted by W1 W2 , is a walk of the
form v0 e1 v1 e2 v2 . . . ek vk ek+1 vk+1 . . . en vn . If the edges e1 , e2 , . . . , ek of a walk W are distinct, W
is called a trail. If, in addition, the vertices v0 , v1 , . . . , vk are distinct, W is called a path. Path
with n vertices is denoted by Pn .
A walk is closed if it has positive length and its origin and terminus are the same. A cycle

17
is a closed path that is a path combined with the edge vk , v1 . A cycle of length k is called a
k-cycle and is denoted by Ck . A cycle Ck is odd if k is odd and it is even if k is even. A 3-cycle
is often called a triangle. A tour in G is a closed walk that traverses each edge of G at least
once. An Euler tour is a tour which traverses each edge exactly once. A graph is said to be
Eulerian if it contains an Euler tour.
Two vertices u and v of G are said to be connected if there is a (u, v)-path in G. If the vertex
set V can be partitioned into subsets V1 , V2 , . . . , Vn such that two vertices u and v are connected
if and only if both u and v belong to the same set Vi then the subgraphs G[V1 ], G[V2 ], . . . , G[Vn ]
are called the components of G. If G has only one component, G is connected, otherwise G is
disconnected. We denote the number of components of G by ω(G).
An acyclic graph is a graph that contains no cycle. A forest is an acyclic graph. A tree
is a connected acyclic graph. Each component of a forest is a tree. A spanning tree of G is
a spanning subgraph of G that is a tree. A rooted tree is a tree having a distinguished vertex
r called the root such that each edge is implicitly directed away from the root. A leaf in a
tree is a vertex of degree 1. A caterpillar is a tree with the property that the removal of its
pendant vertices give a path. In other words, a caterpillar is described as a graph obtained by
attaching any number of leaves to the vertices of a path. A rooted tree consisting of k branches,
where the ith branch is a path of length i, is called an olive tree. The star graph Sn of order
n, sometimes simply known as an n-star is a tree with one vertex having degree n − 1 and the
other n − 1 vertices having degree 1. The star graph Sn is isomorphic to the complete bipartite
graph K1,n−1 . An (n, k)-banana tree is obtained by connecting one leaf of each of n copies of an
k-star graph with a single root vertex v that is not present in any of the starts. A firecracker is
a graph obtained from the concatenation of stars by linking one leaf from each star to a leaf of
another star by an edge. Many of these trees were originated to find a solution to the famous
graceful tree conjecture, defined in 2.2.1.
A cut-edge of G is an edge e such that deletion of an edge e leads to an increase in number
of components in G, that is, ω(G − e) > ω(G). If H is a subgraph of G, the complement of H
in G, denoted by, H(G), is the subgraph G − E(H). A vertex v of G is a cut-vertex if E(G) can
be partitioned into two nonempty subsets E1 and E2 such that subgraph graphs induced by E1
and E2 , that is, G[E1 ] and G[E2 ], have just the vertex v in common. A connected graph that
has no cut-vertex is called a block.
A wheel graph Wn is obtained by joining every vertex of a cycle Cn to a further vertex called
center. The other vertices of wheel Wn are called rim vertices. The helm graph Hn is the graph
obtained from a wheel Wn joining a pendant edge to each rim vertex of Wn . A web graph Wn is
the graph obtained from helm Hn by joining the pendant vertices of a helm to form a cycle and
then adding a single pendant edge to each vertex of this outer cycle. A flower Fn is a graph
obtained from a helm Hn by joining each pendant vertex to the central vertex of the helm. The
ladder Ln is a graph obtained by the Cartesian product of path P2 and path Pn , n ≥ 2, and
the prism Dn is the graph obtained by the Cartesian product of path P2 and cycle Cn . Thus,
Ln ∼= P2 2Pn and Dn ∼ = P2 2Cn . A friendship graph fn is a graph which consists of n triangles
with a common vertex. The fan graph Fn is a graph obtained by joining all vertices of Pn , n ≥ 2,
to a further vertex called the center. Alternatively, for n ≥ 3, the fan graph Fn can be obtained
from the wheel Wn by deleting one edge joining any two rim vertices.

18
 Figure 2.7 shows a wheel W4 , Figure 2.8 depicts a helm H4 and Figure 2.9 presents a web
W4 . Figure 2.10 shows a fan graph F7 , Figure 2.11 represents a ladder L5 and Figure 2.12
depicts a friendship graph f6 .

Figure 2.7: Wheel W4

Figure 2.8: Helm H4

Figure 2.10: Fan graph F7

Figure 2.9: Web graph W4

Figure 2.11: Ladder L5

Figure 2.12: Friendship graph f6

19
2.2 An Introduction to Graph Labelings with Examples

Any field of investigation becomes interesting when it gives rise to the number of problems
that pose challenges to our mind for their eventual solutions. The problems arising from the
study of a variety of labeling techniques of the elements of a graph or of any discrete structure
form one such potential area of challenge. Graph labeling problems are really not of recent
origin, for example, coloring of the vertices arose in connection with the well known Four Color
Theorem, which remained for a long time known by the name Four Color Conjecture and took
more than 150 years for its solution in 1976. Here, we will describe some labeling techniques.
Recall that a labeling of a graph is a mapping (function) that carries graph elements to
numbers (usually non-negative integers). The most common choice for the domain are the
vertex set (vertex labeling), the edge set (edge labeling) or the set of all vertices and edges (total
labeling). Other domains are also possible.
In many cases, it is interesting to consider the sum of all labels associated with a graph
element. This will be called the weight of the element. For example, the weight of the vertex v
if it exists under total labeling is the sum of the label of vertex v and edge labels of all edges
incident with v and the weight of an edge uv under total labeling is the sum of the edge label
if it exists and labels of its end vertices.
Before we begin the study of labeling schemes, let us first define some useful concepts from
algebra.

Definition 2.2.1. A group (G, ·) is a nonempty set G together with a binary operation · on G
such that the following conditions hold:

• Closure: For all a, b ∈ G the element a · b is a uniquely defined element of G.

• Associativity: For all a, b, c ∈ G, we have a · (b · c) = (a · b) · c.

• Identity: There exists an identity element e ∈ G such that e · a = a and a · e = a for all
a ∈ G.

• Inverses: For each a ∈ G there exists an inverse element a−1 ∈ G such that a · a−1 = e
and a−1 · a = e.

Definition 2.2.2. Let (G1 , ·) and (G2 , ∗) be groups and f a function from G1 into G2 . Then f
is called a homomorphism of G1 into G2 if for all a, b ∈ G1 , f (a · b) = f (a) ∗ f (b).

Definition 2.2.3. A homomorphism f of a group G1 into a group G2 is called an isomorphism

of G1 onto G2 if f is bijective. In this case, we write G1 ∼
= G2 and say that G1 and G2 are
isomorphic. An isomorphism of a group G1 onto G1 is called an automorphism.

Graceful Labeling

This labeling scheme was introduced by Rosa, [99] who called it a β-valuation, later named
by Golomb, [50] as a graceful labeling. The origin of a β-valuation was due to the efforts to

20
solve the famous conjecture by Ringel that “the complete graph K2n+1 can be decomposed into
2n + 1 subgraphs, each isomorphic to a given tree with n edges”. Rosa defined another labeling
which was known as an α-valuation with some additional properties to β-valuation. α-valuation
is a graceful labeling such that for each edge e = (u, v) ∈ E, one of the labels f (u), f (v) ≤ k
and the other is > k, for some fixed k, refer [99]. In the same paper, Rosa showed that if G is
a (p, q)-graph and has α-valuation then for every natural number q, the complete graph with
2q + 1 vertices can be decomposed into copies of G in such a way that the automorphism group
of decomposition contains Zq itself.
Definition 2.2.4. A function g is called a graceful labeling of a graph G if g : V (G) → {0, 1, . . . ,
|E(G)|} is injective and the induced function g ∗ : E(G) → {1, 2, . . . , |E(G)|}, defined as

g ∗ (uv) = |g(u) − g(v)|

for every uv ∈ E(G), is bijective. A graph which admits a graceful labeling is called a graceful
graph.

Figure 2.13 demonstrates a graceful labeling of wheel W4 . Here and in whole document, the
weights will be given in circles.

6
0 6

1 5
8 1 2
7
3

8 4
4

Figure 2.13: Graceful labeling of wheel W4

Ringel and Kotzig posed below conjecture, which is still open.

Conjecture 2.2.1. [97] All trees are graceful.

This was the pioneer to major research development in graceful labeling. Using computer
search, Aldred and McKay [5] proved that trees with at most 27 vertices are graceful. This
result was improved by Fang [40] who proved that trees with at most 35 vertices are graceful. In
1976, Bermond and Sotteau [29] proved that rooted trees in which all the vertices at the same
distance from the root and have the same degree (symmetrical trees) are graceful. In 1982,
Huang et al., [58] proved that trees with at most 4 end vertices are graceful. Presently there
are many types of trees which are known to be graceful including paths, caterpillars, olive trees,
banana trees, firecrackers and many more. Rosa [99] in 1967 proved that any Eulerian graph
with number of edges congruent 1 or 2 modulo 4 is not graceful. In 1972, Golomb [50] proved

21
that complete graphs Kn are not graceful for n ≥ 5. In 1979 Frucht et al., [46, 56] proved that
all wheels are graceful. Recently in 2016, Adamaszek et al., [1] proved that trees with maximum
degree O(n/ log n) are graceful.
Truszczynski [119] studied unicyclic graphs and proved several classes of such graphs are
graceful. He conjectured that all unicyclic graphs except Cn , where n ≡ 1 or 2 (mod 4), are
graceful. In 1984, Ayel and Favaron [11] published a paper proving that helms are graceful.
Kang et al., [71] proved that all web graphs are graceful. Seoud and Youssef, [105] established
that all flowers are graceful. Delorme et al., [35] proved that cycles with a chord are graceful.
Graceful labeling is still an open area of research and while the conjecture is still open, many
types of trees have been shown to be graceful.

Harmonious Labeling

Harmonious labeling was introduced by Graham and Sloane [51] in 1980, during the study of
additive basis in number theory. Harmonious labeling has many applications in communication
networks. The following example demonstrates the utility of harmonious labeling.
Consider a network which transmits signals with a criteria that every station must commu-
nicate with some other of that network via some signal. The total bandwidth, say e, is divided
among all the connection channels. Every station x is assigned some number f (x), the label of
x. When two channels x and y communicate they use the flow f (x) + f (y). If the harmonious
labeling f exists for such a network then it is assured that each channel is assigned a unique
link, since harmonious labeling is a bijective function.
Definition 2.2.5. Let G be a graph of order p. A function h is called a harmonious labeling
of G if h : V (G) → Zq , where Zq is the additive group, is an injective function and the induced
function h∗ : E(G) → Zq defined by

h∗ (uv) = (h(u) + h(v)) (mod q)

is bijective. A graph admitting a harmonious labeling is called a harmonious graph.

A harmonious labeling of the friendship graph fn may be regarded as a modular generaliza-

tion of the Langford-Skolem problem, [93], which states that, is it possible to partition the set
P = {1, 2, . . . , 2n} in n pairs (ai , bi ) such that the set of differences bi − ai = i, for i = 1, 2, . . . , n.
This problem has attracted many researchers and there has been significant progress in solving
this problem, see [86]. In order to obtain the harmonious labeling of friendship graph fn , we
will label the vertices of the triangles with (0, i, n + ai ), for i = 1, 2, . . . , n.

There are more applications of harmonious labeling, like embedding of graphs in the plane
and modular versions of many combinatorial problems, see [51].
Figure 2.14 demonstrates a harmonious labeling of wheel W4 .
It was proved by Graham and Sloane [51] that in case of trees exactly one vertex label
has to be repeated for tree to be harmonious and this repeated vertex can be any element of
Zq . Graham and Sloane [51] conjectured that all trees are harmonious, which is still an open

22
 7
0 7

5 4
6 5 2
3
0

6 3
1

Figure 2.14: Harmonious labeling of wheel W4

problem. Aldred and McKay [5] proved that trees with at most 26 vertices are harmonious. In
2012, this result was improved by Fang [41]. Fang proved that trees with at most 31 vertices are
harmonious. Graham and Sloane [51] proved that cycles Cn , n ≥ 3, are harmonious if and only if
n is odd. There are several families of graphs which have been proved to be harmonious. In [51]
Graham and Sloane proved that ladders except L2 are harmonious. There are several results for
existence of harmonious labeling of prisms. In [48, 51, 70] was proved that the generalized prisms
Cm 2Pn are harmonious if n is odd. Further, generalized prisms Cm 2Pn are harmonious if n = 2,
m 6= 4 and also if m = 4 and n ≥ 3. Petersen graph is another graph which is harmonious, see
[51]. Complete bipartite graphs Km,n are harmonious if and only if either m = 1 or n = 1, [51].
Gnanajothi [49] has shown that web graphs with odd cycles are harmonious. Gnanajothi and
Liu [49, 77] proved that helms are harmonious. For more results on harmonious labeling, refer
[47].

Magic Labelings

Many authors have introduced labeling schemes that generalize the idea of a magic square.
A magic square of order n is n × n array of integers 1, 2, . . . , n2 such that the sum of numbers
along any row, column and main diagonals is a fixed constant that equals to n(n2 + 1)/2. In
1963, Sedláček [102] pointed out the correspondence between a magic square of order n and
magic labeling of a complete bipartite graph Kn,n . He found out that if we label every edge ui vj
of Kn,n with the number from ith row and j th column of the magic square of order n, we obtain
the supermagic labeling of Kn,n . Sedláček [102] in 1963 defined a graph to be magic if it has an
edge labeling, with the range of the real numbers so that the sum of the edge labels around any
vertex is always the same, a constant, independent of the choice of a vertex. More precisely, see
the following definition.

Definition 2.2.6. A function f is called a magic labeling of G if f : E(G) → Z+ , is an injective

function and the induced function has the property that for every vertex v ∈ V (G) the associated

23
vertex weight equals to a constant independent of the choice of a vertex, that is,
X
wf (v) = f (uv) = λ,
uv∈E(G)

where λ is a constant. A magic labeling is called supermagic if the set of all the labels of the
edges consists of consecutive integers. A graph admitting a magic (supermagic) labeling is called
a magic (supermagic) graph.

Figure 2.15 demonstrates a supermagic labeling of wheel W4 with constant λ = 24.

24 9 24

4 5
11 24 10
7
8

24 6 24

Figure 2.15: Supermagic labeling of wheel W4

Some authors refer to a graph a supermagic if its edges, in a magic labeling can be labeled
with numbers {1, 2, . . . , |E(G|}, see [54, 106]. Note that for regular graphs these definitions are
equivalent, see [62].
Some sufficient conditions for the existence of magic graphs are established in [14, 90, 103,
110]. A characterization of regular magic graphs was given by Doob in [37]. There were in-
dependently published two different characterizations of all magic graphs, see Jeurissen, [66]
and Jezný and Trenkler in [68]. To this time only some special classes of supermagic graphs
have been characterized. In [111], Stewart characterized supermagic complete graphs and the
characterization of supermagic regular complete multipartite graphs and supermagic cubes is
given by Ivančo in [62].
Now we will introduce other types of magic labelings where not only the edges but also the
vertices of a graph are labeled.
Definition 2.2.7. A function f is called a vertex-magic total labeling of G if f : V (G)∪E(G) →
{1, 2, . . . , |V (G)|+|E(G)|} is a bijective function with the property that for every vertex v ∈ V (G)
the associated vertex weight equals to a constant independent of the choice of vertex, that is.,
X
wtf (v) = f (v) + f (uv) = k,
uv∈E(G)

where k is a constant, known as a magic constant. If the vertices are labeled with the smallest

24
possible labels the vertex-magic total labeling is called super vertex-magic total labeling. A graph
admitting a (super) vertex-magic total labeling is called a (super) vertex-magic total graph.

Figure 2.16 gives an example of vertex-magic total labeling of wheel W4 with the magic
constant k = 29.

29 29
13
3 1

5 6
8 29 9
4 12
2

10 11
7
29 29

Figure 2.16: Vertex-magic total labeling of wheel W4

The definition of a vertex-magic total labeling was introduced by MacDougall et al., in

[80] and [81]. MacDougall [78] conjectured that all regular graphs other than K2 and 2K3 are
vertex-magic total.
Definition 2.2.8. A function f is called an edge-magic total labeling of G if f : V (G)∪E(G) →
{1, 2, . . . , |V (G)|+|E(G)|} is a bijective function with the property that for every edge uv ∈ E(G)
the associated edge weight equals to a constant independent of the choice of edge, that is,

wtf (uv) = f (u) + f (uv) + f (v) = k,

where k is a constant, known as a magic constant. If the vertices are labeled with the smallest
possible labels the edge-magic total labeling is called a super edge-magic total labeling. A graph
admitting a (super) edge-magic total labeling is called a (super) edge-magic total graph.

Figure 2.17 gives an example of super edge-magic total labeling of cycle C5 with the magic
constant k = 14.
Edge-magic total labeling was introduced by Kotzig and Rosa [73, 74] under the name magic
valuation. In [73] Kotzig and Rosa proved that complete bipartite graphs Km,n are edge-magic
total for all m and n, and cycles Cn are edge-magic total for all n ≥ 3. Kotzig and Rosa [74]
proved that complete graphs Kn are edge-magic total if and only if n = 1, 2, 3, 5 or 6. Wallis et
al., [123] described edge-magic total labelings of Kn , n = 1, 2, 3, 5 and 6, for all possible values
of the magic constant k.
One of the variations of magic labeling is a bimagic labeling.
Babujee [12, 13] introduced the idea of a bimagic labeling, in which there are two constants,
say k1 and k2 , such that all weights under this labeling equal to one or to other of those two

25
 1

10 9
14 14
3 4

6 14 14 8

14
5 2
7

Figure 2.17: Super edge-magic total labeling of cycle C5

constants. For example, we would define an edge-bimagic total labeling f of G to be a total

labeling such that the edge weights wtf (xy) = f (x) + f (xy) + f (y) equals either k1 or k2 for
every edge xy ∈ E(G). It is interesting to consider the following two cases.

• All edges but one have the common weight. Such a labeling is said to be almost magic.

• The number of edges with one weight differs at most by 1 to the number of edges with the
other weight. In this case the labeling is known as equitable bimagic.

In [82] Marr et al., proved that when n ≡ 3 (mod 4), the wheel Wn has both an equitable
bimagic labeling and an almost magic labeling.
For more information about magic or bimagic type labelings see Gallian’s dynamic survey
of graph labeling [47] or [122].

Antimagic Labelings

In the previous section we introduced the definitions of some magic type labelings. These
labelings have one common property, that the weights of the considered elements are all the
same. Now we will deal the case when all the weights of the considered elements are pairwise
distinct. Such labeling is called an antimagic type labeling.
Antimagic labeling was first introduced by Hartsfield and Ringel [54] in 1990.
Definition 2.2.9. A function f is called an antimagic labeling of G if f : E(G) → {1, 2, . . . ,
|E(G)|} is a bijective function with the property that all the vertex weights are pairwise distinct.
A graph admitting an antimagic labeling is called an antimagic graph.

Figure 2.18 gives an example of an antimagic labeling of wheel W4 .

Hartsfield and Ringel [54] gave the following conjecture.
Conjecture 2.2.2. [54] All graphs except K2 are antimagic.

26
 8 1 13

5 8
2 26 4
6
7

11 3 14

Figure 2.18: Antimagic labeling of wheel W4

This conjecture is still open. Using probabilistic method and techniques of analytical number
theory, Alon et al., [7] showed that this conjecture is true for all graphs having minimum degree
Ω(log |V (G)|). They also proved that if G is a graph with |V (G)| ≥ 4 and maximum degree
∆(G) = |V (G)| − 2 then G is antimagic. Hartsfield and Ringel proved that many classes of
graphs are antimagic. In [54], it is proved that paths Pn , n ≥ 3, cycles Cn , wheels Wn and
complete graphs Kn , n ≥ 3, are antimagic. In 1999, there has been a progress in Hartsfield -
Ringel’s conjecture by Alon [6]. Alon proved that all dense graphs, the graphs with number of
edges close to |V (G)|2 , are antimagic.
Miller et al., [88] have contributed by introducing a new antimagic scheme for different types
of graphs. They applied finite combinatorics methods to find antimagic labeling for graphs and
using these techniques they proved that for a given degree sequence for a tree or a forest, they
can provide an antimagic tree or forest with that degree sequence.
Further they proved that all trees and forests have edge-antimagic vertex labeling, that is,
it is possible to label their vertices such that all edge weights are distinct. Moreover, they also
proved that the labeling is super.
A total labeling f is said to be edge-antimagic total (EAT) labeling if all edge-weights are
pairwise distinct. Similarly f is said to be vertex-antimagic total (VAT) labeling if all vertex-
weights are pairwise distinct. In [87], Miller et al., proved that all graphs have VAT labelings.
They also proved in the same paper that for every graph there exist VAT labeling which are
super, repus (that is the vertex labels are {q + 1, q + 2, . . . , q + p}) and neither super nor repus.
The same is for EAT graphs. In [87], Miller et al., proved that all graphs have EAT labelings.
Thus as this problem for VAT graphs is completely solved, it has sense to give some restriction
on vertex weights. We will get the concept of (a, d)-VAT labeling defined by Bača et al., in [15]
that the weights are not only distinct, but also form arithmetic sequence with a difference d.
Another avenue of research is to consider that as all graphs have VAT labelings and also EAT
labelings, do there exist graphs possessing a labeling that is simultaneously VAT and EAT? We
get the concept of totally antimagic total labeling defined by Bača et al., [25] which is a natural
extension of the concept of a totally magic labeling defined by Exoo et al., in [39].

27
 Bača et al., [15] defined the concept of an (a, d)-vertex-antimagic total labeling.

Definition 2.2.10. A function f is called an (a, d)-vertex-antimagic total labeling of G, denoted

by (a, d)-VAT, if f : V (G) ∪ E(G) → {1, 2, . . . , |V (G)| + |E(G)|} is a bijective function with the
property that the set of vertex weights is

{wtf (v) : v ∈ V (G)} = {a, a + d, . . . , a + (|V (G)| − 1)d},

where a > 0 and d ≥ 0 are two fixed integers. If the vertices are labeled with the smallest possible
labels, an (a, d)-vertex-antimagic total labeling is called a super (a, d)-vertex-antimagic total
labeling. A graph admitting a (super) (a, d)-vertex-antimagic total labeling is called a (super)
(a, d)-vertex-antimagic total graph.

Figure 2.19 gives an example of (25, 1)-vertex-antimagic total labeling of wheel W4 .

29 25
3
5 13

11 2
10 28 7
1 8
6

12 9
4
27 26

Figure 2.19: A (25,1)-vertex-antimagic total labeling of wheel W4

Analogously, the natural extensions of the edge-magic total labeling and the super edge-
magic total labeling are the (a, d)-edge-antimagic total and super (a, d)-edge-antimagic total
labelings, respectively. The definition of (a, d)-edge-antimagic total labeling was introduced by
Simanjuntak et al., in [108].

Definition 2.2.11. A function f is called an (a, d)-edge-antimagic total labeling of G, denoted

by (a, d)-EAT, if f : V (G) ∪ E(G) → {1, 2, . . . , |V (G)| + |E(G)|} is a bijective function with the
property that the set of edge weights is

{wtf (e) : e ∈ E(G)} = {a, a + d, . . . , a + (|E(G)| − 1)d},

where a > 0 and d ≥ 0 are two fixed integers. If the vertices are labeled with the smallest possible
labels the (a, d)-edge-antimagic total labeling is called super (a, d)-edge-antimagic total. A graph
admitting a (super) (a, d)-edge-antimagic total labeling is called a (super) (a, d)-edge-antimagic
total graph. If d = 0 then f is called an edge-magic total labeling.

Figure 2.20 gives an example of super (10, 2)-edge-antimagic total labeling of cycle C5 .

28
 1

6 7
10 12
3 4

10 18 14 8

16
5 2
9

Figure 2.20: Super (10, 2)-edge-antimagic total labeling of cycle C5

The generalization of the definition of edge-magic total and edge-antimagic total labelings is
the concept of H-magic, H-antimagic labelings, respectively.
Let G be a finite simple graph. An edge-covering of G is a family of subgraphs H1 , H2 , . . . , Ht
such that each edge of E(G) belongs to at least one of the subgraphs Hi , i = 1, 2, . . . , t. Then it
is said that G admits an (H1 , H2 , . . . , Ht )-(edge) covering. If every Hi is isomorphic to a given
graph H, then G admits an H-covering. Note that in this case every subgraph isomorphic to H
must be in the H-covering.
Definition 2.2.12. Suppose that a graph G admits an H-covering. A function f is called an
(a, d)-H-antimagic total labeling of G if f : V (G) ∪ E(G) → {1, 2, . . . , |V (G)| + |E(G)|} is a
bijective function with the property that for all subgraphs H 0 of G isomorphic to H, the H-weights
constitute an arithmetic progression

a, a + d, . . . , a + (t − 1)d,

where a > 0 and d ≥ 0 are two integers, and t is the number of all subgraphs of G isomorphic
to H. For the subgraph H under the total labeling f the associated H-weight is defined as
X X
wtf (H) = f (v) + f (e).
v∈V (H) e∈E(H)

If the vertices are labeled with the smallest possible labels the (a, d)-H-antimagic total labeling is
called a super (a, d)-H-antimagic total labeling. A graph admitting a (super) (a, d)-H-antimagic
labeling is called a (super) (a, d)-H-antimagic total graph.

Note that for d = 0 the considered labeling is called an H-(super)magic total labeling.
In [104], Semaničová-Feňovčı́ková et al., proved that wheels are cycle antimagic. Figure 2.21
illustrates super (39, 1)-C3 -antimagic total labeling of wheel W4 , the C3 -weights are depicted in
circles.
The H-(super)magic total labelings were first studied by Gutiérrez and Lladó in [52]. They
considered star-(super)magic and path-(super)magic labelings of some connected graphs and

29
 9
1 2
42
13 12
39
6 41 8
10 5
11
40
4 3
7

Figure 2.21: Super (39, 1)-C3 -antimagic total labeling of wheel W4

proved that the path Pn and the cycle Cn are Ph -supermagic for some h. Maryati et al., [83]
gave Ph -(super)magic labelings of some trees such as shrubs, subdivision of shrubs and banana
tree graphs. Lladó and Moragas [75] investigated Cn -(super)magic graphs and proved that
wheels, windmills, books and prisms are Ch -magic for some h. Some results on Cn -supermagic
labelings of several classes of graphs can be found in [91]. Other examples of H-supermagic
graphs with different choices of H have been given by Jeyanthi and Selvagopal in [67]. Inayah
et al., [59] gave a connection between graceful trees and antimagic H-decomposition of complete
graphs. Maryati et al., [84] investigated the G-supermagicness of a disjoint union of c copies of
a graph G and showed that disjoint union of any paths is cPh -supermagic for some c and h.
Motivated by H-(super)magic labelings, Inayah et al., [60] introduced the (a, d)-H-antimagic
labeling. In [61] they investigated the super (a, d)-H-antimagic labelings for some shackles of
a connected graph H. In [104] was proved that wheels are super (a, 1)-Ck -antimagic for every
k = 3, 4, . . . , n − 1, n + 1.
For more information about antimagic type labelings see [24, 47].

Irregular Labelings

The following problem was proposed by Chartrand et al., in [33]. Assign positive integer
labels to the edges of a simple connected graph of order at least 3 in such a way that the graph
becomes irregular, that is, the label sums (weights) at each vertex are distinct. What is the
minimum value of the largest label over all such irregular assignments? This leads to the concept
of irregular labeling.

Definition 2.2.13. A function ψ is called an irregular assignment of G if ψ : E(G) →

{1, 2, . . . , k} has the property that the associated vertex weights are pairwise distinct, that is,

wψ (u) 6= wψ (v)

30
for all vertices u, v ∈ V (G), u 6= v. The weight of a vertex v ∈ V (G) is
X
wψ (v) = ψ(uv),
uv∈E(G)

where the sum is over all vertices u adjacent to v. The irregularity strength s(G) of a graph G
is defined as the minimum k for which G has an irregular assignment using labels at most k.

The irregularity strength s(G) can be interpreted as the smallest integer k for which G can
be turned into a multigraph G0 by replacing each edge by a set of at most k parallel edges, such
that the degrees of the vertices in G0 are all different.
Finding the irregularity strength of a graph is a tough task, see [8, 36, 42, 53, 64, 65].

Definition 2.2.14. A function ρ is called an edge irregular k-labeling of G if ρ : V (G) →

{1, 2, . . . , k} has the property that the associated edge weights are pairwise distinct, that is,

wρ (uv) 6= wρ (u0 v 0 )

for every two different edges uv and u0 v 0 . The weight of an edge uv ∈ E(G) is

wρ (uv) = ρ(u) + ρ(v).

The minimum k for which the graph G has an edge irregular k-labeling is called the edge irreg-
ularity strength of G, denoted by es(G).

The notion of the edge irregularity strength was defined by Ahmad et al., [3]. They de-
termined the exact value of the edge irregularity strength for paths, stars, double stars and
Cartesian product of two paths and there is also given a lower bound for es(G).
Bača et al., [20] defined the edge irregular total k-labeling.

Definition 2.2.15. A function ϕ is called an edge irregular total k-labeling of G if ϕ : V (G) ∪

E(G) → {1, 2, . . . , k} has the property that the associated edge weights are pairwise distinct, that
is,
wtϕ (uv) 6= wtϕ (u0 v 0 )
for every two different edges uv and u0 v 0 . The weight of an edge uv ∈ E(G) is

wtϕ (uv) = ϕ(u) + ϕ(v) + ϕ(uv).

The minimum k for which the graph G has an edge irregular total k-labeling is called the total
edge irregularity strength of G, denoted by tes(G).

In [20], the bound for tes(G) is given. For a graph G, d(|E(G)| + 2)/3e ≤ tes(G) ≤ |E(G)|.
They also gave tes for paths and cycles as, for n > 1, tes(Pn ) = tes(Cn ) = d(n + 2)/3e.
Moreover they proved that for stars K1,n , n ≥ 1 have the same edge and vertex total irregularity
strength, that is, tes(K1,n ) = tvs(K1,n ) = d(n + 1)/2e. They also proved that for trees T with
n pendant vertices and no vertex of degree 2 it holds d(n + 1)/2e ≤ tvs(T ) ≤ n and the

31
 2

3 2
7 6
2 2

2 5 4 1

3
1 1
1

Figure 2.22: Edge irregular total labeling of cycle C5

total irregular strength for complete graphs is tvs(Kn ) = 2. Jendrol’ et al., [64] proved that
tes(Kp ) = d(p2 − p + 4)/6e for any p ≥ 6.
In Figure 2.22, edge irregular total labeling of C5 is shown.
Bača et al., [20] defined a vertex irregular total k-labeling of a graph G.

Definition 2.2.16. A function ϕ is called a vertex irregular total k-labeling of G if ϕ : V (G) ∪

E(G) → {1, 2, . . . , k} has the property that the associated vertex weights are different for all
vertices, that is,
wtϕ (u) 6= wtϕ (u)

for all vertices u, v ∈ V (G), u 6= v. The weight of a vertex v ∈ V (G) is

X
wtϕ (v) = ϕ(v) + ϕ(uv),
uv∈E(G)

where the sum is over all vertices u adjacent to v. The minimum k for which the graph G has
an vertex irregular total k-labeling is called the total vertex irregularity strength of G, denoted
by tvs(G).

Note that irregularity strength s(G) of a graph G is defined only for graphs containing at
most one isolated vertex and no connected component of order 2. However, the total vertex
irregularity strength tvs(G) is defined for every graph G. Thus for graphs with no component
of order at most 2, tvs(G) ≤ s(G).
Ahmed et al., [4] described the vertex irregularity strength of helms Hn , they proved tvs(Hn ) =
d(n+1)/2e, for n ≥ 4. Let fm,n be a generalized friendship graph, where m is the number of cycles
and n is the order of cycle. For n = 3 it has been shown in [124] that tvs(fm,n ) = d(2m + 2)/3e.
In [20], Bača et al., determined an exact value of the total vertex irregularity strength for
the prism Dn , n ≥ 3, as tvs(Dn ) = d(2n + 3)/4e. They also proved that tvs(Cn ) = d(n + 2)/3e.
Wijaya and Slamin [124] found the exact value of the total vertex irregularity strength for
the wheels, fans, sun graphs and friendship graphs as follows. For wheel graphs Wn , n ≥ 3,

32
tvs(Wn ) = d(n + 3)/4e. For fan graphs Fn , n ≥ 3, tvs(Fn ) = d(n + 2)/4e. For friendship graphs
fn , n ≥ 3, tvs(fn ) = d(2n + 2)/3e.
In Figure 2.23, vertex irregular total labeling of C5 is shown.

7
3

2 2
5 2 2 6

1 2

3 1 1 4
1

Figure 2.23: Vertex irregular total labeling of cycle C5

In [100], Ryan et al., introduced the concept of irregular reflexive labeling in 2014, in which
the vertex labels are interpreted as (double) the number of loops at a vertex. This new labeling
provides two differences from irregular total labeling as follows:

• The labels of vertices have to be even number considering the fact that each loop con-
tributes 2 to the degree of a vertex.

• Vertex label 0 is permissible which represents loopless vertex.

The irregular reflexive labeling is defined as below.

Definition 2.2.17. For a graph G, we define labelings ρe : E(G) → {1, 2, . . . ke } and ρv :

V (G) → {0, 2, . . . , 2kv }. Let ρ = ρe ∪ ρv and k = max{ke , 2kv }.
The labeling ρ is said to be an edge irregular reflexive k-labeling if distinct edges e and f
have distinct weights, that is,
wtρ (e) 6= wtρ (f ).
Similarly, the labeling ρ is said to be a vertex irregular reflexive k-labeling if distinct vertices u
and v have distinct weights, that is,

wtρ (u) 6= wtρ (v).

The smallest value of k for which such labelings exist is called the reflexive edge strength of the
graph G (resp. reflexive vertex strength of G), res(G) (resp. rvs(G)).

The subsequent chapters explain irregular reflexive labeling in details and several theorems
have been proved.

33
34
Chapter 3

Edge Irregular Reflexive Labeling

3.1 Introduction

Regular graphs have been an area of interest almost as long as graphs have been studied.
However, as a consequence of the Handshaking Lemma, no simple graphs can be completely
irregular. That is, no simple graph can have each vertex bearing a distinct degree. Multigraphs
however can display this property. In [33] authors asked, “In a loopless multigraph, determine the
fewest parallel edges required to ensure that all vertices have distinct degree”. For convenience,
the problem was recast in terms of a labeled simple graph with the edge labels representing the
number of parallel edges. The degree of a vertex was then determined by adding the labels of
the edges incident to that vertex. Then the problem became,
“Assign positive values to the edges of a simple connected graph of order at least 3 in such a
way that the graph becomes irregular. What is the minimum value of the largest label over all
such irregular assignments?” The minimum value of the largest label is known as the irregularity
strength of a graph.
Recognising the problem of finding irregularity strength as a labeling problem, Bača et al.,
in [20], considered total labelings on graphs that is, labeling both edges and vertices. This
expanded the concept of weight which could now be measured, not only at vertices, but also at
edges. In [20] they introduced the total vertex irregularity strength of a graph G, denoted by
tvs(G), and the total edge irregularity strength of G, denoted by tes(G), as being the minimum
maximum label so that the vertex weights (resp. edge weights) were pairwise distinct, for
definitions see Subsection 2.2. Work of authors in [20] has inspired further work in this field
such as [9, 63, 94, 107].
The concept of irregular reflexive multigraphs originated in [100] as a natural consequence
of irregular multigraphs by allowing loops. Following the initiative of [33] and rewording the
problem as a graph labeling exercise, irregular reflexive labeling include vertex labels representing
degrees contributed by the loops. The weight of a vertex v, denoted by wt(v), is now determined
by adding the incident edge labels and the label of v.
Previously in [20] authors proposed an irregular total labeling in which the vertices were

35
labeled by positive integers. The difference between this idea and irregular reflexive labeling is
threefold:

1. The concept is consistent with the genesis of the problem by considering multigraphs with
loops.
2. The vertex label must be even non negative integers, representing the fact that each loop
contributes 2 to the vertex degree.
3. Vertex label 0 is permissible as representing a loopless vertex.

As in the case of irregular total labelings, this new scheme allows to consider not just vertex
weights but also edge weights. An edge weight is the sum of the edge label and the labels of
the vertices incident to the edge. Thus we are able to propose vertex irregular reflexive labelings
and edge irregular reflexive labelings.
We present here some basic results concerning irregular reflexive labeling and provide ir-
regular reflexive edge and vertex strengths for some families of graphs including stars, paths,
cycles, complete graphs, prisms, wheel, join of graphs and generalised friendship graphs. Hence-
forth, reflexive edge strength and reflexive vertex strength will be abbreviated as res and rvs
respectively.
We recall the definition of irregular reflexive labeling.
Definition 3.1.1. For a graph G, we define labelings ρe : E(G) → {1, 2, . . . ke } and ρv : V (G) →
{0, 2, . . . , 2kv }. Let ρ = ρe ∪ ρv and k = max{ke , 2kv }.
The labeling ρ is said to be an edge irregular reflexive k-labeling if distinct edges e and f
have distinct weights, that is,
wtρ (e) 6= wtρ (f ).
Similarly, the labeling ρ is said to be a vertex irregular reflexive k-labeling if distinct vertices u
and v have distinct weights, that is,

wtρ (u) 6= wtρ (v).

The smallest value of k for which such labelings exist is called the reflexive edge strength of the
graph G (resp. reflexive vertex strength of G), res(G) (resp. rvs(G)).

In Section 3.2, we will consider edge irregular reflexive labelings and give examples of res(G)
for some well known graphs. Due to the similarity of the labeling schemes, many of the theorems
and proofs presented here will be similar to those given in [20] however respective strengths (on
the same graph) may be different. For example, res(K5 ) = 4 while in [20] it was shown that
tes(K5 ) = 5.

3.2 Edge Irregular Reflexive Labeling

We begin the study of edge irregular reflexive labelings by showing that all graphs can bear
an edge irregular reflexive labeling and so every graph has a reflexive edge strength.

36
♦ Lemma 3.2.1. [100] For every graph G,
l |E(G)| m
 3 if |E(G)| 6≡ 2, 3 (mod 6),
res(G) ≥ l m
 |E(G)| + 1 if |E(G)| ≡ 2, 3 (mod 6).
3

Proof. The lower bound for res(G) follows from the fact that the minimal edge weight under
a edge irregular reflexive labeling is 1 and the minimum of the maximal edge weights, that is
|E(G)|, can be achieved only as the sum of 3 numbers from which at least two are even.

♦ Theorem 3.2.2. [100] Let G be a simple graph, then

l m
|E(G)|
3 ≤ res(G) ≤ |E(G)|.

Proof. To get the upper bound we label each vertex of G with label 0 and the edges of G are
labeled consecutively with labels 1, 2, . . . , |E(G)|. This labeling ensures that wt(e) 6= wt(f ) for
any two distinct edges e and f of G.
Let ρ be an optimal labeling with respect to the res(G). Then the heaviest edge e of G
has weight wt(e) ≥ |E(G)|. This weight is the sum of three labels. So one label is at least
|E(G)|/3.

We now consider the edge irregular reflexive labeling for some standard classes of graphs.
The first case demonstrates that the lower bound of Theorem 3.2.2 is tight.

♦ Theorem 3.2.3. [100] The reflexive edge strength for the star K1,n is
( 
n
if n 6≡ 2 (mod 4),
res(K1,n ) =  n2 
2 + 1 if n ≡ 2 (mod 4).

Proof. The graph K1,n has n edges with the smallest possible edge weight being 1 and the largest
being at least n. If we label the central vertex 0 then the least maximum label must be dn/2e
when n 6≡ 2 (mod 4). For n ≡ 2 (mod 4), dn/2e is an odd number and so cannot be a vertex
label. In this case we can increase the vertex label to n/2 + 1 and reduce the corresponding edge
label, or reduce the vertex label by 1 which would require increasing the corresponding edge
label to n/2 + 1. So we have,
( 
n
if n 6≡ 2 (mod 4),
res(K1,n ) ≥  n2 
2 + 1 if n ≡ 2 (mod 4).

The following labeling scheme provides the equality in the above bounds, thus proving the
theorem.
Label the central vertex 0. Order the pendant vertices v1 , v2 , . . . , vn and corresponding

37
incident edges e1 , e2 , . . . , en . Label ρ(v1 ) = 0, ρ(e1 ) = 1, ρ(v2 ) = 0, ρ(e2 ) = 2. For i ≥ 3 label,

ρ(vi ) = 2i , ρ(ei ) = i
if i ≡ 0 (mod 4),
2 i 
ρ(vi ) = 2i ,
 
ρ(ei ) = 2 if i ≡ 1 (mod 4),
i
ρ(vi ) = − 1, ρ(ei ) = 2i + 1 if i ≡ 2 (mod 4),
2 i 
ρ(ei ) = 2i
 
ρ(vi ) = 2 , if i ≡ 3 (mod 4).

It is very easy to observe that the weights of edges will fall between 1, 2, . . . , n.

Figure 3.1 and Figure 3.2 provides edge irregular reflexive labeling of K1,6 and K1,5 respec-
tively.

2 0
0
2 2 2
3
3
1 2 2
2 1 1 0 1
0 1 0 0
4 2 1
3 4 2 3
4
2
5 6 5

2 2
2
Figure 3.1: Edge irregular reflexive labeling of Figure 3.2: Edge irregular reflexive labeling of
K1,6 K1,5

The next theorem provides the reflexive edge strength of the path Pn on n vertices.

♦ Theorem 3.2.4. [100] The reflexive edge strength for the path Pn is
(
n+1
3 if n ≡ 2 (mod 6),
res(Pn ) =  n+3 
2 6 otherwise.

Proof. Clearly the smallest maximum edge weight for the path Pn is n − 1, the number of edges.
The following labeling scheme produces such a weight and provides the smallest maximum label
by allocating labels as close as possible to one third the weight subject to the constraint on the
vertices being labeled with even numbers.
Number the vertices and edges of the path from one of the leaves, beginning with 1 so that
the edge numbered e1 is incident with vertices v1 and v2 . Define a reflexive labeling φ as follows;

38
 φ(v1 ) = 0, φ(e1 ) = 1, φ(v2 ) = 0, φ(e2 ) = 2, φ(v3 ) = 0, φ(e3 ) = 1, φ(v4 ) = 2, φ(e4 ) = 2, φ(v5 ) =
0, φ(e5 ) = 3, φ(v6 ) = 2. For the remaining vertices and edges, label as follows;

φ(vi ) = i+3 − 1, φ(ei ) = i+3

   
3 3 −1 if i ≡ 0 (mod 6),
 i+3   i+3 
φ(vi ) = 3 − 1, φ(ei ) = 3 if i ≡ 1 (mod 6),
 i+3   i+3 
φ(vi ) = 3 − 1, φ(ei ) = 3 − 1 if i ≡ 2 (mod 6),
 i+3   i+3 
φ(vi ) = 3 , φ(ei ) = 3 − 3 if i ≡ 3 (mod 6),
 i+3   i+3 
φ(vi ) = 3 , φ(ei ) = 3 − 2 if i ≡ 4 (mod 6),
 i+3   i+3 
φ(vi ) = 3 , φ(ei ) = 3 − 1 if i ≡ 5 (mod 6).

It is a simple matter to check that φ(vi ) + φ(ei ) + φ(vi+1 ) = i for all i ≥ 6 so that for each edge,
wt(ei ) = i.

Figure 3.3 and Figure 3.4 provide edge irregular reflexive labelings of P7 and P8 respectively.

0 1 0 2 0 1 2 2 0 3 2 2 2
1 2 3 4 5 6

Figure 3.3: Edge irregular reflexive labeling of P7

0 1 0 2 0 1 2 2 0 3 2 2 2 3 2
1 2 3 4 5 6 7

Figure 3.4: Edge irregular reflexive labeling of P8

♦ Theorem 3.2.5. [18] For every positive integer n, n ≥ 3

(
d n3 e if n 6≡ 2, 3 (mod 6),
res(Cn ) = n
d 3 e + 1 if n ≡ 2, 3 (mod 6).

Proof. Let us denote the vertex set and the edge set of the cycle Cn , n ≥ 3 such that V (Cn ) =
{xi : i = 1, 2, . . . , n}, E(Cn ) = {xi xi+1 : i = 1, 2, . . . , n}, where indices are taken modulo n.
As the cycle Cn has n edges, immediately from Lemma 3.2.1 we get that
(
d n3 e if n 6≡ 2, 3 (mod 6),
res(Cn ) ≥ n
d 3 e + 1 if n ≡ 2, 3 (mod 6).

Now we distinguish two subcases.

Case 1. Let n ≡ 3 (mod 6). If n = 3, from the lower bound we get res(C3 ) ≥ 2 and the edge
irregular reflexive labeling of C3 , as illustrated shown equality in Figure 3.5.

39
 0

3 1 1 1

4
2
2 0

Figure 3.5: Edge irregular reflexive labeling of C3

For n ≥ 9 we define the total (n/3 + 1)-labeling f of Cn in the following way

f (xi ) = 2 d i+1 i = 1, 2, . . . , n+3

3 e−1 2 ,
f (xn−i+1 ) = 2d i−1
3 e i = 1, 2, . . . , n−3
2 ,
f (x n+3 x n+5 ) = 2b n6 c,
2 2
f (xi xi+1 ) = 2d 3i e − 1 i = 1, 2, . . . , n+1
2 ,
f (xn−i xn−i+1 ) = 2d i+1
3 e i = 1, 2, . . . , n−5
2 ,
f (xn x1 ) = 2.

The vertices of Cn are labeled with even numbers.

The edge weights of the edges in Cn under the labeling f are the following. For i =
1, 2, . . . , (n + 1)/2 is

wtf (xi xi+1 ) =f (xi ) + f (xi xi+1 ) + f (xi+1 ) = 2 d i+1 i i+2

3 e − 1 + 2d 3 e − 1 + 2 d 3 e − 1
=2 d 3i e + d i+1 i+2


3 e + d 3 e − 5 = 2(i + 2) − 5 = 2i − 1.

Thus, the corresponding edge weights are 1, 3, . . . , n and

wtf (x n+3 x n+5 ) =f (x n+3 ) + f (x n+3 x n+5 ) + f (x n+5 )

2 2
 2n+3  2
 2 2n−3 
+1 n 2 −1
 n+5  n
 n−5 

=2 2 − 1 + 2b c + 2 = 2 + b c + −2
3 6 3 6 6 6

=n − 1.

For i = 1, 2, . . . , (n − 5)/2 is

wtf (xn−i xn−i+1 ) =f (xn−i ) + f (xn−i xn−i+1 ) + f (xn−i+1 )

=2d 3i e + 2d i+1 i−1 i−1 i i+1


3 e + 2d 3 e = 2 d 3 e + d 3 e + d 3 e = 2i + 2.

Thus, these edge weights are 4, 6, . . . , n − 3.

40
 Moreover,

wtf (xn x1 ) =f (xn ) + f (xn x1 ) + f (x1 ) = 0 + 2 + 0 = 2.

That means that the edge weights are distinct numbers from the set {1, 2, . . . , n}.

Case 2. Let n 6≡ 3 (mod 6). We define the total labeling f of Cn such that

f (xi ) = 2 d i+1 i = 1, 2, . . . , d n2 e,


3 e − 1
f (xn−i+1 ) = 2d i−1
3 e i = 1, 2, . . . , b n2 c,
f (xi xi+1 ) = 2d 3i e − 1 i = 1, 2, . . . , d n2 e,
f (xn−i xn−i+1 ) = 2d i+1
3 e i = 1, 2, . . . , b n2 c − 1,
f (xn x1 ) = 2.

Evidently the vertices of Cn are labeled with even numbers and the labels used are at most
dn/3e if n 6≡ 2 (mod 6) or they are at most (dn/3e + 1) if n ≡ 2 (mod 6).
The edge weights of the edges in Cn under the labeling f are the following.
For i = 1, 2, . . . , dn/2e − 1 is

wtf (xi xi+1 ) =f (xi ) + f (xi xi+1 ) + f (xi+1 ) = 2 d i+1 i i+2

3 e − 1 + 2d 3 e − 1 + 2 d 3 e − 1
=2 d 3i e + d i+1 i+2


3 e + d 3 e − 5 = 2(i + 2) − 5 = 2i − 1.

Thus, for n even the edge weights are 1, 3, . . . , n−3 and for n odd the edge weights are 1, 3, . . . , n−
2.

wtf (xd n e xd n e+1 ) =f (xd n e ) + f (xd n e xd n e+1 ) + f (xd n e+1 )

2 2 2 2 2 2
=f (xd n e ) + f (xd n e xd n e+1 ) + f (xn−b n c+1 )
 2 n  2
  2
 n   2  n 
d 2 e+1 d2e b 2 c−1
=2 3 − 1 + 2 3 − 1 + 2 3

d n2 e−1 d n2 e b n2 c+1
     
− 3 = 2 d n2 e + 1 − 3 = 2d n2 e − 1,


=2 3 + 3 + 3

which is equal to (n − 1) for n even and is equal to n for n odd.

For i = 1, 2, . . . , bn/2c − 1 is

wtf (xn−i xn−i+1 ) =f (xn−i ) + f (xn−i xn−i+1 ) + f (xn−i+1 ) = 2d 3i e + 2d i+1 i−1

3 e + 2d 3 e
=2 d i−1 i i+1


3 e + d 3 e + d 3 e = 2i + 2.

Thus, for n even these edge weights are 4, 6, . . . , n and for n odd these edge weights are
4, 6, . . . , n − 1.
Moreover,

wtf (xn x1 ) =f (xn ) + f (xn x1 ) + f (x1 ) = 0 + 2 + 0 = 2.

41
Combining the previous facts we get that weights of the edges are distinct numbers from the set
{1, 2, . . . , n}. This completes the proof.

Figures 3.6 and 3.7 demonstrate edge irregular reflexive labelings of C10 and C12 respectively.
5 7 5
1 3 2 1
7 2 2 3 2 2
3 9 3
1 3 1
2
0 4 0
9 1 11 3
3 1 1 1

4 0 4 0

2 2 12 4 2 2
10 2 4 0
4 4 2
0
2 10 4
2 2 2
8 2 2 4 4 2 2
2 8 6
6

Figure 3.6: Edge irregular reflexive labeling Figure 3.7: Edge irregular reflexive labeling
of C10 of C12
In the next theorem we give the exact values of reflexive edge strength of Cartesian product
of a cycle Cn and C3 .
♦ Theorem 3.2.6. [18] For every positive integer n, n ≥ 3

res(Cn 2C3 ) = 2n.

Proof. We denote the vertex set and the edge set of Cn 2C3 , n ≥ 3 such that

V (Cn 2C3 ) = {xi , yi , zi : i = 1, 2, . . . , n},

E(Cn 2C3 ) = {xi yi , xi zi , yi zi , xi xi+1 , yi yi+1 , zi zi+1 : i = 1, 2, . . . , n},

where indices are taken modulo n.

The graph Cn 2C3 has 6n edges thus using Lemma 3.2.1 we have

res(Cn 2C3 ) ≥ 2n.

Now we distinguish two subcases according to the parity of n.

Case 1. When n is even let us consider the total 2n-labeling f defined in the following way

f (xi ) = 0 i = 1, 2, . . . , n,
f (yi ) = n i = 1, 2, . . . , n,

42
 f (zi ) = 2n i = 1, 2, . . . , n,
f (xi xi+1 ) = i i = 1, 2, . . . , n − 1,
f (xn x1 ) = n,
f (yi yi+1 ) = n + i i = 1, 2, . . . , n − 1,
f (yn y1 ) = 2n,
f (zi zi+1 ) = n + i i = 1, 2, . . . , n − 1,
f (zn z1 ) = 2n,
f (xi yi ) = i i = 1, 2, . . . , n,
f (yi zi ) = n + i i = 1, 2, . . . , n,
f (xi zi ) = i i = 1, 2, . . . , n.

Evidently f is a 2n-labeling. Now we calculate the edge weights.

wtf (xi xi+1 ) =f (xi ) + f (xi xi+1 ) + f (xi+1 ) = 0 + i + 0 = i

for i = 1, 2, . . . , n − 1,
wtf (x1 xn ) =f (x1 ) + f (x1 xn ) + f (xn ) = 0 + n + 0 = n,
wtf (xi yi ) =f (xi ) + f (xi yi ) + f (yi ) = 0 + i + n = n + i
for i = 1, 2, . . . , n,
wtf (xi zi ) =f (xi ) + f (xi zi ) + f (zi ) = 0 + i + 2n = 2n + i
for i = 1, 2, . . . , n,
wtf (yi yi+1 ) =f (yi ) + f (yi yi+1 ) + f (yi+1 ) = n + (n + i) + n = 3n + i
for i = 1, 2, . . . , n,
wtf (yi zi ) =f (yi ) + f (yi zi ) + f (zi ) = n + (n + i) + 2n = 4n + i
for i = 1, 2, . . . , n,
wtf (zi zi+1 ) =f (zi ) + f (zi zi+1 ) + f (zi+1 ) = 2n + (n + i) + 2n = 5n + i
for i = 1, 2, . . . , n.

Thus the set of edge weights is {1, 2, . . . , 6n}.

Case 2. When n is odd we define the total 2n-labeling f Cn 2C3 such that

f (xi ) = 0 i = 1, 2, . . . , n,
f (yi ) = n + 1 i = 1, 2, . . . , n − 1,
f (yn ) = n − 1,
f (zi ) = 2n i = 1, 2, . . . , n,
f (xi xi+1 ) = i i = 1, 2, . . . , n − 1,
f (xn x1 ) = n,
f (yi yi+1 ) = n + i i = 1, 2, . . . , n − 2,
f (yn y1 ) = n + 1,
f (yn−1 yn ) = n + 2,

43
 f (zi zi+1 ) = n + i i = 1, 2, . . . , n − 1,
f (z1 zn ) = 2n,
f (xi yi ) = i i = 1, 2, . . . , n − 1,
f (xn yn ) = 2,
f (yi zi ) = n + i i = 1, 2, . . . , n − 1,
f (yn zn ) = n + 2,
f (xi zi ) = i i = 1, 2, . . . , n.

Also in this case the vertices are labeled with even numbers and the labels are at most 2n. For
the edge weights we have

wtf (xi xi+1 ) =f (xi ) + f (xi xi+1 ) + f (xi+1 ) = 0 + i + 0 = i

for i = 1, 2, . . . , n − 1,
wtf (x1 xn ) =f (x1 ) + f (x1 xn ) + f (xn ) = 0 + n + 0 = n,
wtf (xi yi ) =f (xi ) + f (xi yi ) + f (yi ) = 0 + i + (n + 1) = n + i + 1
for i = 1, 2, . . . , n − 1,
wtf (xn yn ) =f (xn ) + f (xn yn ) + f (yn ) = 0 + 2 + (n − 1) = n + 1,
wtf (xi zi ) =f (xi ) + f (xi zi ) + f (zi ) = 0 + i + 2n = 2n + i
for i = 1, 2, . . . , n,
wtf (yi yi+1 ) =f (yi ) + f (yi yi+1 ) + f (yi+1 ) = (n + 1) + (n + i) + (n + 1) = 3n + i + 2
for i = 1, 2, . . . , n − 2,
wtf (yn−1 yn ) =f (yn−1 ) + f (yn−1 yn ) + f (yn ) = (n + 1) + (n + 2) + (n − 1) = 3n + 2,
wtf (yn y1 ) =f (yn ) + f (yn y1 ) + f (y1 ) = (n − 1) + (n + 1) + (n + 1) = 3n + 1,
wtf (yi zi ) =f (yi ) + f (yi zi ) + f (zi ) = (n + 1) + (n + i) + 2n = 4n + i + 1
for i = 1, 2, . . . , n − 1,
wtf (yn zn ) =f (yn ) + f (yn zn ) + f (zn ) = (n − 1) + (n + 2) + 2n = 4n + 1,
wtf (zi zi+1 ) =f (zi ) + f (zi zi+1 ) + f (zi+1 ) = 2n + (n + i) + 2n = 5n + i
for i = 1, 2, . . . , n.

Which means that also for n odd the edge weights are distinct numbers from the set
{1, 2, . . . , 6n}.

Figure 3.8 gives edge irregular reflexive labeling of C4 2C3 .

3.3 Edge Irregular Reflexive Labeling of Prisms

The prism Dn , n ≥ 3, is a trivalent graph which can be defined as the Cartesian product
P2 2Cn of a path on two vertices with a cycle on n vertices. We denote the vertex set and the
edge set of Dn such that V (Dn ) = {xi , yi : i = 1, 2, . . . , n} and E(Dn ) = {xi xi+1 , yi yi+1 , xi yi :

44
 3 3
0 0
8 7
4
12 4 3 3
4 7 15 11
20 8 4
8 19
8 7
4
16 23 7 22 14 2
8
4 8 6 6 2
24
5 21
5 8 8
17
4 6 18
9
1 10
5 1 5 13 4 2
2 6
0 0
1 1

Figure 3.8: Edge irregular reflexive labeling of C4 2C3

i = 1, 2, . . . , n}, where indices are taken modulo n. Here we have considered the outer cycle
vertices as yi and inner cycle vertices as xi .

♦ Theorem 3.3.1. [116] For n ≥ 3,

(
n+1 if n is odd,
res(Dn ) =
n if n is even.

Proof. As the prism Dn has 3n edges, immediately from Lemma 3.2.1 we get that res(Dn ) ≥ n+1
for n is odd and res(Dn ) ≥ n for n is even.
Let k = n for n even and let k = n + 1 for n odd. We define the labeling f of Dn such that

f (xi ) = 0 i = 1, 2, . . . , n,
f (yi ) = k i = 1, 2, . . . , n,
f (xi xi+1 ) = f (yi yi+1 ) = i i = 1, 2, . . . , n − 1,
f (x1 xn ) = f (y1 yn ) = n,
f (xi yi ) = i i = 1, 2, . . . , n.

Evidently f is k-labeling. The edge weights of the edges in Dn under the labeling f are the
following.

wtf (xi xi+1 ) =0 + i + 0 = i for i = 1, 2, . . . , n − 1,

wtf (x1 xn ) =0 + n + 0 = n,
wtf (xi yi ) =0 + i + k = k + i for i = 1, 2, . . . , n,

45
 wtf (yi yi+1 ) =k + i + k = 2k + i for i = 1, 2, . . . , n − 1,
wtf (y1 yn ) =k + n + k = 2k + n.

That means that for n odd {wtf (e) : e ∈ E(Dn )} = {1, 2, . . . , n, n + 2, n + 3, . . . , 2n + 1, 2n +

3, 2n + 4, . . . , 3n + 2} and for n even {wtf (e) : e ∈ E(Dn )} = {1, 2, . . . , 3n}. Thus the edge
weights are distinct, that is f is a edge irregular reflexive labeling of a prism Dn .

Figures 3.9 and 3.10 demonstrate edge irregular reflexive labeling of D8 and D9 respectively.
19 8 18 23
3 3 10 22
2 10
8 2 10 8 2
24 14 3 13
11 3 2 1 3 10
3 0 17 4 4
20 4 3 2 0 9 0 2 2
0 1 4 0 3 2 0 12 1 21
4 4 10 15 4
1 1
12 5 0 1 1
8 4 0 0 8 8
16 25 5 5 5 0 1 10
5 5 8 8 16 0 9 11
5 9
13 0 6 7 0 7 8 24 10 6 6 9
21 8 0 29
5 6 0 7 15 6 0 7 9
0 8 19
8 14 6 8 6 17 7 10
7 18 8
6 7 26 8
8 10
23 7 10 28
22
27

Figure 3.9: Edge irregular reflexive labeling of Figure 3.10: Edge irregular reflexive labeling
D8 of D9

3.4 Edge Irregular Reflexive Labeling of Wheels

The wheel Wn , n ≥ 3, is a graph obtained by joining all vertices of Cn to a further vertex

called the centre. We denote the vertex set and the edge set of Wn such that V (Wn ) = {x, xi :
i = 1, 2, . . . , n} and E(Wn ) = {xi xi+1 , xxi : i = 1, 2, . . . , n}, where indices are taken modulo n.
We prove the following result for wheels.

♦ Theorem 3.4.1. [116] For n ≥ 3,


4
 if n = 3,
res(Wn ) = d 2n3 e if n ≡ 0, 2 (mod 3) and n ≥ 5,

 2n
d 3 e+1 if n ≡ 1 (mod 3).

Proof. According to the fact that the wheel Wn has 2n edges, using Lemma 3.2.1 we obtain the
following lower bound for wheel res(Wn ) ≥ k = d 2n 3 e if n ≡ 0, 2 (mod 3) and res(Wn ) ≥ k =
2n
d 3 e + 1 if n ≡ 1 (mod 3). It is easy to see that k is even.

46
 2 2 2
0 0
6
3 7
1 4 2 3
7 1
3 8 4
1 3
2
1 8
2 6 5
4 1 2
2
3 0 2
0 2 4 2
5

Figure 3.11: Edge irregular reflexive labeling Figure 3.12: Edge irregular reflexive
of W3 labeling of W4

4 11 3 4
2 5
4
2 1 2 9 12
0 2 10
8 9 10 6 4 3 4
3 8
4 4 3
4 0 4
2 1 5 2
2
3 2 2 2 7
2 7 3 1
6 2 4 3 3
0 0
1 1 0 0
1 1

Figure 3.13: Edge irregular reflexive labeling Figure 3.14: Edge irregular reflexive
of W5 labeling of W6

Thus res(W3 ) ≥ 2. If res(W3 ) ≤ 3 then the vertices of W3 can be labeled only with 0’s or 2’s
and the set of all possible edge weights is a subset of the set {1, 2, . . . , 7}. As the edge weights 1
and 2 could be realizable only as 1=0+1+0; 2=0+2+0 then five edges have end vertex labeled
with 0. However, this is a contradiction as the edge weights 6 and 7 can be realizable only as
6=2+2+2 and 7=2+3+2. Thus res(W3 ) ≥ 4. The corresponding labelings for Wn , n = 3, 4, 5, 6
are illustrated on Figures 3.11, 3.12, 3.13 and 3.14 respectively.
For n ≥ 7 we define a total labeling of Wn such that

f (x) = k,



 0 i = 1, 2, . . . , k − 1,

2 i = k,
f (xi ) =


 k i = k + 1, k + 2, . . . , n − 1,

k − 2 i = n,

47
 


i i = 1, 2, . . . , k − 1,

k − 2 i = k,
f (xxi ) =


n − 2k + 1 + i i = k + 1, k + 2, . . . , n − 1,

5 i = n,



i i = 1, 2, . . . , k − 2,




k−3 i = k − 1,

k − 1 i = k,
f (xi xi+1 ) =


i − k + 3 i = k + 1, k + 2, . . . , n − 2,




4 i = n − 1,

2 i = n.

Evidently for n ≥ 7 f is a k-labeling. Now we calculate the edge weights.

wtf (xi xi+1 ) =0 + i + 0 = i for i = 1, 2, . . . , k − 2,

wtf (xk−1 xk ) =0 + (k − 3) + 0 = k − 1,
wtf (xk xk+1 ) =2 + (k − 1) + k = 2k + 1,
wtf (xi xi+1 ) =k + (i − k + 3) + k = k + 3 + i for i = k + 1, k + 2, . . . , n − 2,
wtf (xn−1 xn ) =k + 4 + (k − 2) = 2k + 1,
wtf (xn x1 ) =(k − 2) + 2 + 0 = k,
wtf (xxi ) =k + i + 0 = k + i for i = 1, 2, . . . , k − 1,
wtf (xxk ) =k + (k − 2) + 2 = 2k,
wtf (xxi ) =k + (n − 2k + 1 + i) + k = n + 1 + i for i = k + 1, k + 2, . . . , n − 1,
wtf (xxn ) =k + 5 + (k − 2) = 2k + 3.

It is easy to check that the edge weights are from the set {1, 2, . . . , 2n}. This concludes the
proof.

Figure 3.15 and 3.16 demonstrate edge irregular reflexive labelings of W10 and W12 respec-
tively.
Let us receall the following definitions from Section 2.1.
A fan graph Fn is obtained from wheel Wn if one rim edge, say x1 xn is deleted. A basket Bn
is obtained by removing a spoke, say xxn , from wheel Wn . Before we will give the exact value
of reflexive edge strength of fan graphs and baskets we give the following observation.

♦ Observation 3.4.2. [116] Let e be an arbitrary edge in G. Then

res(G − {e}) ≤ res(G).

Proof. The proof is trivial.

48
 3 4 3
3 4 0 3
4 0 0 2 0 0
4 5 2
2 5 2
0 12 12 11
11 0 13
13 4 3 0 5 4 3 0
5 5 1 10
5 1 6 6 6 1 1
10 14
14 6 2 7 2 9
8 0 8 0
0 0 15 6 1
1 9
7 19
6 15 2 7 5 5
2 8
6 5 8 16 6
6 4 2 7 8 24 6
0 16 19 6 22 23
20 7 4
5 17 18
4 8 8
7 2 8 18 4 8 5
7
20 21
17

Figure 3.15: Edge irregular reflexive labeling Figure 3.16: Edge irregular reflexive labeling
of W10 of W12

1 2 1
1 2 0 1
2 0 0 6 0 0
2 3 8
2 3 2
0 8 10 9
7 4 0 11 2 6
9 2 1 3 1 19
3 14 4 4
3 3
15 4 12 4 4 18
5 5 5 20
10 4 6 8
0 6 0 8
4 16 13 6 4
5 5 5 17
4 11 7 1
4 19 14 5 7
4 5 0 5 23 8
0 12 15 21
17 6 4 8
3 6 24
6 2 8
5 2 6 18 6 8 6
5 16 22
13

Figure 3.17: Edge irregular reflexive labeling Figure 3.18: Edge irregular reflexive labeling
of B10 of F12

♦ Theorem 3.4.3. [116] For n ≥ 3,


3
 if n = 3,
res(Fn ) = res(Bn ) = 4 if n = 4,

 2n
d3e if n ≥ 4.

Proof. Using similar arguments as in the proof of Theorem 3.4.1 we get that res(F3 ) ≥ 3 and

49
res(F4 ) ≥ 4. The corresponding labeling for F3 can be obtained from the labeling illustrated
for W3 , see Figure 3.11, when the edge labeled with 4 is deleted. For the fan graph F4 we can
delete arbitrary rim edge from W4 .
For n ≥ 5, combining Lemma 3.2.1 and Observation 3.4.2 we get that res(Fn ) = res(Wn ) =
d2n/3e for n ≡ 0, 2 (mod 3) and d2n/3e ≤ res(Fn ) ≤ d2n/3e + 1 for n ≡ 1 (mod 3).
Let n ≡ 1 (mod 3), n ≥ 7. Then the number k = d2n/3e is odd. We define k-labeling of Fn
such that

f (x) = k − 1,


 k − 1 i = 1,

k − 3 i = 2,



f (xi ) = 0 i = 3, 4, . . . , k,

2 i = k + 1,





k − 1 i = k + 2, k + 3, . . . , n,



 4 i = 1,

5 i = 2,



f (xxi ) = i − 2 i = 3, 4, . . . , k,

k − 3

 i = k + 1,


2i − 2k + 1 i = k + 2, k + 3, . . . , n,




 4 i = 1,




 2 i = 2,

i − 2 i = 3, 4, . . . , k − 1,
f (xi xi+1 ) =


 k−4 i = k,

k − 2

 i = k + 1,


2i − 2k + 2 i = k + 2, k + 3, . . . , n − 1.

It is not difficult to check that the edge weights are distinct numbers from the set 1, 2, . . . , 2n−1.
The proof for basket Bn can be done analogously as for the fan graph. Evidently V (Bn ) =
V (Fn ) and E(Bn ) = E(Fn ) ∪ {x1 xn } − {xxn }. For n ≡ 1 (mod 3), n ≥ 7, the following d2n/3e-
labeling g of Bn defined such that g(y) = f (y) for y ∈ V (Fn ) or y ∈ E(Fn ) − {xxn } and
g(x1 xn ) = f (xxn ) has desired properties.

Figures 3.17 and 3.18 demonstrate edge irregular reflexive labelings of B10 and F12 respec-
tively.

3.5 Edge Irregular Reflexive Labeling of Join of Graphs

The join G ⊕ H of the disjoint graphs G and H is the graph G ∪ H together with all the
edges joining vertices of V (G) and vertices of V (H).

50
 In the next two theorems we will deal with the join of a path or a cycle with 2K1 .

♦ Theorem 3.5.1. [18] For every positive integer n, n ≥ 2


3
 if n = 2,
res(Pn ⊕ (2K1 )) = n + 1 if n is odd, n ≥ 3,

n if n is even, n ≥ 4.


Proof. We denote the vertex set and the edge set of Pn ⊕ (2K1 ), n ≥ 2 such that

V (Pn ⊕ (2K1 )) = {xi : i = 1, 2, . . . , n} ∪ {y, z},

E(Pn ⊕ (2K1 )) = {xi xi+1 : i = 1, 2, . . . , n − 1} ∪ {yxi , zxi : i = 1, 2, . . . , n}.

As |E(Pn ⊕ (2K1 ))| = 3n − 1 then by Lemma 3.2.1 we have

(
n if n is even,
res(Pn ⊕ (2K1 )) ≥
n + 1 if n is odd.

However, it is easy to see that res(P2 ⊕ (2K1 )) ≥ 3. The corresponding 3-labeling for P2 ⊕ (2K1 )
is illustrated in Figure 3.19.

1 1
0 0

3 5 4

1 3 2

2 2
2 6

Figure 3.19: Edge irregular reflexive labeling of P2 ⊕ (2K1 )

For n ≥ 3 we distinguish two subcases according to the parity of n.

Case 1. When n is even we define n-labeling f of Pn ⊕ (2K1 ) such that

f (xi ) = 0 i = 1, 2, . . . , n2 ,
f (x n +1 ) = n − 2,
2
n
f (xi ) = n i= 2 + 2, n2 + 3, . . . , n,
f (y) = 0,
f (z) = n,
n
f (xi xi+1 ) = 2 +i i = 1, 2, . . . , n2 − 1,
f (x n x n +1 ) = 2,
2 2

51
 f (x n +1 x n +2 ) = 3,
2 2
n n
f (xi xi+1 ) = i − 2 i= 2 + 2, n2 + 3, . . . , n − 1,
f (yxi ) = i i = 1, 2, . . . , n2 ,
f (yx n +1 ) = 3,
2
n n
f (yxi ) = i − 2 i= 2 + 2, n2 + 3, . . . , n,
n
f (zxi ) = 2 +i i = 1, 2, . . . , n2 ,
n
f (zx n +1 ) = 2 + 2,
2
n
f (zxi ) = i − 1 i= 2 + 2, n2 + 3, . . . , n.

For the edge weights we get

wtf (xi xi+1 ) =f (xi ) + f (xi xi+1 ) + f (xi+1 ) = 0 + ( n2 + i) + 0 = n

2 +i
n
for i = 1, 2, . . . , 2 − 1,
wtf (x n x n +1 ) =f (x n ) + f (x n x n +1 ) + f (x n +1 ) = 0 + 2 + (n − 2) = n,
2 2 2 2 2 2
wtf (x n +1 x n +2 ) =f (x n +1 ) + f (x n +1 x n +2 ) + f (x n +2 ) = (n − 2) + 3 + n = 2n + 1,
2 2 2 2 2 2
wtf (xi xi+1 ) =f (xi ) + f (xi xi+1 ) + f (xi+1 ) = n + (i − n2 ) + n = 3n
2 +i
n n
for i = 2 + 2, 2 + 3, . . . , n − 1,
wtf (yxi ) =f (y) + f (yxi ) + f (xi ) = 0 + i + 0 = i
for i = 1, 2, . . . , n2 ,
wtf (yx n +1 ) =f (y) + f (yx n +1 ) + f (x n +1 ) = 0 + 3 + (n − 2) = n + 1,
2 2 2
wtf (yxi ) =f (y) + f (yxi ) + f (xi ) = 0 + (i − n2 ) + n = n
2 +i
n n
for i = 2 + 2, 2 + 3, . . . , n,
wtf (zxi ) =f (z) + f (zxi ) + f (xi ) = n + ( n2 + i) + 0 = 3n
2 +i
n
for i = 1, 2, . . . , 2,
wtf (zx n +1 ) =f (z) + f (zx n +1 ) + f (x n +1 ) = n + ( n2 + 2) + (n − 2) = 5n
2 ,
2 2 2
wtf (zxi ) =f (z) + f (zxi ) + f (xi ) = n + (i − 1) + n = 2n + i − 1
n
for i = 2 + 2, n2 + 3, . . . , n.

Thus the set of edge weights is {1, 2, . . . , 3n − 1}.

Case 2. When n is odd we define (n + 1)-labeling f in the following way

f (xi ) = 0 i = 1, 2, . . . , n+1
2 ,
f (x n+3 ) = n − 1,
2
n+5 n+7
f (xi ) = n + 1 i= 2 , 2 , . . . , n,
f (y) = 0,
f (z) = n + 1,

52
 n+1
f (xi xi+1 ) = 2 +i i = 1, 2, . . . , n−1
2 ,
f (x n+1 x n+3 ) = 2,
2 2
f (x n+3 x n+5 ) = 2,
2 2
n+3 n+5 n+7
f (xi xi+1 ) = i − 2 i= 2 , 2 ,...,n − 1,
f (yxi ) = i i= 1, 2, . . . , n+1
2 ,
f (yx n+3 ) = 3,
2
n+1 n+5 n+7
f (yxi ) = i − 2 i= 2 , 2 , . . . , n,
n−1
f (zxi ) = 2 + i i= 1, 2, . . . , n+1
2 ,
n+1
f (zx n+3 ) = 2 ,
2
n+5 n+7
f (zxi ) = i − 3 i= 2 , 2 , . . . , n.

Evidently, the vertices are labeled with even numbers and the label of every element is at most
n + 1.
Now we will calculate the weights of the edges.

wtf (xi xi+1 ) =f (xi ) + f (xi xi+1 ) + f (xi+1 ) = 0 + ( n+1

2 + i) + 0 =
n+1
2 +i
n−1
for i = 1, 2, . . . , 2 ,
wtf (x n+1 x n+3 ) =f (x n+1 ) + f (x n+1 x n+3 ) + f (x n+3 ) = 0 + 2 + (n − 1) = n + 1,
2 2 2 2 2 2
wtf (x n+3 x n+5 ) =f (x n+3 ) + f (x n+3 x n+5 ) + f (x n+5 ) = (n − 1) + 2 + (n + 1) = 2n + 2,
2 2 2 2 2 2
n+3 3n+1
wtf (xi xi+1 ) =f (xi ) + f (xi xi+1 ) + f (xi+1 ) = (n + 1) + (i − 2 ) + (n + 1) = 2 +i
n+5 n+7
for i = 2 , 2 ,...,n − 1,
wtf (yxi ) =f (y) + f (yxi ) + f (xi ) = 0 + i + 0 = i
for i = 1, 2, . . . , n+1
2 ,
wtf (yx n+3 ) =f (y) + f (yx n+3 ) + f (x n+3 ) = 0 + 3 + (n − 1) = n + 2,
2 2 2
n+1 n+1
wtf (yxi ) =f (y) + f (yxi ) + f (xi ) = 0 + (i − 2 ) + (n + 1) = 2 +i
n+5 n+7
for i = 2 , 2 , . . . , n,
wtf (zxi ) =f (z) + f (zxi ) + f (xi ) = (n + 1) + ( n−1
2 + i) + 0 =
3n+1
2 +i
n+1
for i = 1, 2, . . . , 2 ,
n+1 5n+1
wtf (zx n+3 ) =f (z) + f (zx n+3 ) + f (x n+3 ) = (n + 1) + 2 + (n − 1) = 2 ,
2 2 2
wtf (zxi ) =f (z) + f (zxi ) + f (xi ) = (n + 1) + (i − 3) + (n + 1) = 2n + i − 1
n+5 n+7
for i = 2 , 2 , . . . , n.

It is easy to check that the edge weights are distinct consecutive integers {1, 2, . . . , 3n − 1}.
This concludes the proof.

53
 In the next theorem we give bounds for the graph Cn ⊕ (2K1 ).
♦ Theorem 3.5.2. [18] For every positive integer n, n≥3

5
 if n = 4,
res(Cn ⊕ (2K1 )) = n + 1 if n is odd,

n if n is even.


Proof. We denote the vertex set and the edge set of Cn ⊕ (2K1 ), n ≥ 2 such that

V (Cn ⊕ (2K1 )) = {xi : i = 1, 2, . . . , n} ∪ {y, z},

E(Cn ⊕ (2K1 )) = {xi xi+1 , yxi , zxi : i = 1, 2, . . . , n},

where the indices are taken modulo n. The graph Cn ⊕(2K1 ) has 3n edges thus applying Lemma
3.2.1 we obtain
(
n if n is even, n ≥ 6,
res(Cn ⊕ (2K1 )) ≥
n + 1 if n is odd.

Let us consider two subcases according to the parity of n.

Case 1. Let n be even number.
It is easy to see that res(C4 ⊕ (2K1 )) ≥ 5. The corresponding 5-labeling for C4 ⊕ (2K1 ) is
illustrated in Figure 3.20.

4 10
2 1 3 0
2 1 5
9 1 6
4 1
3 0 2 4
8 7 5
3
4 13
4
3 11 4
12
4

Figure 3.20: Edge irregular reflexive labeling of C4 ⊕ (2K1 )

For n ≥ 6 we define n-labeling f of Cn ⊕ (2K1 ) such that

f (xi ) = 0 i = 1, 2, . . . , n2 − 1,
f (x n ) = n − 4,
2
f (x n +1 ) = n − 2,
2

54
 f (x n +2 ) = n − 2,
2
n
f (xi ) = n i= 2 + 3, n2 + 4, . . . , n,
f (y) = 0,
f (z) = n,
n
f (xi xi+1 ) = 2 +i−1 i = 1, 2, . . . , n2 − 2,
f (x n −1 x n ) = 2,
2 2
f (x n x n +1 ) = 6,
2 2
f (x n +1 x n +2 ) = 5,
2 2
f (x n +2 x n +3 ) = 4,
2 2
n n
f (xi xi+1 ) = i − 2 i= 2 + 3, n2 + 3, . . . , n − 1,
f (xn x1 ) = n − 1,
f (yxi ) = i i = 1, 2, . . . , n2 − 1,
f (yx n ) = 3,
2
f (yx n +1 ) = 2,
2
f (yx n +2 ) = 3,
2
n n
f (yxi ) = i − 2 −1 i= 2 + 3, n2 + 4, . . . , n,
n
f (zxi ) = 2 +i−1 i = 1, 2, . . . , n2 − 1,
n
f (zx n ) = 2 + 4,
2
n
f (zx n +1 ) = 2 + 3,
2
n
f (zx n +2 ) = 2 + 4,
2
n
f (zxi ) = i i= 2 + 3, n2 + 4, . . . , n.

For the edge weights we get

wtf (xi xi+1 ) =f (xi ) + f (xi xi+1 ) + f (xi+1 ) = 0 + ( n2 + i − 1) + 0 = n

2 +i−1
n
for i = 1, 2, . . . , 2 − 2,
wtf (x n −1 x n ) =f (x n −1 ) + f (x n −1 x n ) + f (x n ) = 0 + 2 + (n − 4) = n − 2,
2 2 2 2 2 2
wtf (x n x n +1 ) =f (x n ) + f (x n x n +1 ) + f (x n +1 ) = (n − 4) + 6 + (n − 2) = 2n,
2 2 2 2 2 2
wtf (x n +1 x n +2 ) =f (x n +1 ) + f (x n +1 x n +2 ) + f (x n +2 ) = (n − 2) + 5 + (n − 2) = 2n + 1,
2 2 2 2 2 2
wtf (x n +2 x n +3 ) =f (x n +2 ) + f (x n +2 x n +3 ) + f (x n +3 ) = (n − 2) + 4 + n = 2n + 2,
2 2 2 2 2 2
wtf (xi xi+1 ) =f (xi ) + f (xi xi+1 ) + f (xi+1 ) = n + (i − n2 ) + n = 3n
2 +i
n n
for i = 2 + 3, 2 + 4, . . . , n − 1,
wtf (xn x1 ) =f (xn ) + f (xn x1 ) + f (x1 ) = n + (n − 1) + 0 = 2n − 1,

55
 wtf (yxi ) =f (y) + f (yxi ) + f (xi ) = 0 + i + 0 = i
for i = 1, 2, . . . , n2 − 1,
wtf (yx n ) =f (y) + f (yx n ) + f (x n ) = 0 + 3 + (n − 4) = n − 1,
2 2 2
wtf (yx n +1 ) =f (y) + f (yx n +1 ) + f (x n +1 ) = 0 + 2 + (n − 2) = n,
2 2 2
wtf (yx n +2 ) =f (y) + f (yx n +2 ) + f (x n +2 ) = 0 + 3 + (n − 2) = n + 1,
2 2 2
n n
wtf (yxi ) =f (y) + f (yxi ) + f (xi ) = 0 + (i − 2 − 1) + n = 2 +i−1
n n
for i = 2 + 3, 2 + 4, . . . , n,
wtf (zxi ) =f (z) + f (zxi ) + f (xi ) = n + ( n2 + i − 1) + 0 = 3n
2 +i−1
for i = 1, 2, . . . , n2 − 1,
wtf (zx n ) =f (z) + f (zx n ) + f (x n ) = n + ( n2 + 4) + (n − 4) = 5n
2 ,
2 2 2
wtf (zx n +1 ) =f (z) + f (zx n +1 ) + f (x n +1 ) = n + ( n2 + 3) + (n − 2) = 5n
2 + 1,
2 2 2
wtf (zx n +2 ) =f (z) + f (zx n +2 ) + f (x n +2 ) = n + ( n2 + 4) + (n − 2) = 5n
2 + 2,
2 2 2
wtf (zxi ) =f (z) + f (zxi ) + f (xi ) = n + i + n = 2n + i
n
for i = 2 + 3, n2 + 4, . . . , n.

It is easy to get that the edge weights are {1, 2, . . . , 3n}.

Case 2. Let n be odd. Then we define (n + 1)-labeling f in the following way

f (xi ) = 0 i = 1, 2, . . . , n+1
2 ,
f (x n+3 ) = n − 1,
2
n+5 n+7
f (xi ) = n + 1 i= 2 , 2 , . . . , n,
f (y) = 0,
f (z) = n + 1,
n+1
f (xi xi+1 ) = 2 +i i = 1, 2, . . . , n−1
2 ,
f (x n+1 x n+3 ) = 2,
2 2
f (x n+3 x n+5 ) = 3,
2 2
n+1 n+5 n+7
f (xi xi+1 ) = i − 2 i= 2 , 2 ,...,n − 1,
f (xn x1 ) = n + 1,
f (yxi ) = i i = 1, 2, . . . , n+1
2 ,
f (yx n+3 ) = 3,
2
n+1 n+5 n+7
f (yxi ) = i − 2 i= 2 , 2 , . . . , n,
n−1
f (zxi ) = 2 + i i= 1, 2, . . . , n+1
2 ,
n+3
f (zx n+3 ) = 2 ,
2
n+5 n+7
f (zxi ) = i − 2 i= 2 , 2 , . . . , n.

56
Thus the vertices are labeled with even numbers 0, n − 1 or n + 1.
For the edge weights we get the following.

wtf (xi xi+1 ) =f (xi ) + f (xi xi+1 ) + f (xi+1 ) = 0 + ( n+1

2 + i) + 0 =
n+1
2 +i
n−1
for i = 1, 2, . . . , 2 ,
wtf (x n+1 x n+3 ) =f (x n+1 ) + f (x n+1 x n+3 ) + f (x n+3 ) = 0 + 2 + (n − 1) = n + 1,
2 2 2 2 2 2
wtf (x n+3 x n+5 ) =f (x n+3 ) + f (x n+3 x n+5 ) + f (x n+5 ) = (n − 1) + 3 + (n + 1) = 2n + 3,
2 2 2 2 2 2
n+1 3n+3
wtf (xi xi+1 ) =f (xi ) + f (xi xi+1 ) + f (xi+1 ) = (n + 1) + (i − 2 ) + (n + 1) = 2 +i
n+5 n+7
for i = 2 , 2 ,...,n − 1,
wtf (xn x1 ) =f (xn ) + f (xn x1 ) + f (x1 ) = (n + 1) + (n + 1) + 0 = 2n + 2,
wtf (yxi ) =f (y) + f (yxi ) + f (xi ) = 0 + i + 0 = i
for i = 1, 2, . . . , n+1
2 ,
wtf (yx n+3 ) =f (y) + f (yx n+3 ) + f (x n+3 ) = 0 + 3 + (n − 1) = n + 2,
2 2 2
n+1 n+1
wtf (yxi ) =f (y) + f (yxi ) + f (xi ) = 0 + (i − 2 ) + (n + 1) = 2 +i
n+5 n+7
for i = 2 , 2 , . . . , n,
wtf (zxi ) =f (z) + f (zxi ) + f (xi ) = (n + 1) + ( n−1
2 + i) + 0 =
3n+1
2 +i
n+1
for i = 1, 2, . . . , 2 ,
n+3 5n+3
wtf (zx n+3 ) =f (z) + f (zx n+3 ) + f (x n+3 ) = (n + 1) + 2 + (n − 1) = 2 ,
2 2 2
wtf (zxi ) =f (z) + f (zxi ) + f (xi ) = (n + 1) + (i − 2) + (n + 1) = 2n + i
n+5 n+7
for i = 2 , 2 , . . . , n.

Evidently, the edge weights are distinct numbers from the set {1, 2, . . . , 3n}.

3.6 Edge Irregular Reflexive Labeling for Generalised Friend-

ship Graphs

In this section we will investigate the edge irregular reflexive labeling for generalized friend-
ship graphs. The friendship graph fm is a collection of m triangles with a common vertex. It
may be also pictured as a wheel W2m with every alternate rim edge removed. Let us mention
that the reflexive edge strength for wheels can be found in Theorem 3.4.1. The generalized
friendship graph fn,m is a collection of m cycles all of order n, meeting at a common vertex.
We will refer to the friendship graph fm as an instance of the generalized friendship graph and
write it as f3,m . For our purposes, we refer to vertices in the following way: the central vertex
is named x and all other vertices addressed in the form xji , where j, 1 ≤ j ≤ m indicates which
cycle contains the vertex and i, 1 ≤ i ≤ n points to the position of the vertex within the cycle.
So x = xj0 for all j. Then we denote the edge set of the generalized friendship graph such that
E(fn,m ) = {xji xji+1 : 1 ≤ i ≤ n − 2, 1 ≤ j ≤ m} ∪{xxj1 , xxjn−1 : 1 ≤ j ≤ m}.

57
 In this section we determine the exact value of the reflexive edge strength for the generalized
friendship graphs fn,m for n = 3, 4, 5 and m ≥ 1.

♦ Theorem 3.6.1. [19] For every positive integer m ≥ 1


3
 if m = 2,
res(f3,m ) = m if m is even, m ≥ 4,

m + 1 if m is odd.


Proof. The graph f3,m has 3m edges thus by Lemma 3.2.1 we have
(
m if m is even,
res(f3,m ) ≥
m + 1 if m is odd.

As f3,1 is isomorphic to C3 thus according to Theorem 3.2.5 we get res(C3 ) = 2.

It is easy to see that res(f3,2 ) ≥ 3. The corresponding edge irregular reflexive labeling for
f3,2 is illustrated in Figure 3.21. This graph is also known as a bowtie graph.

2
0
4 1
2 1
0
6 2 3 3
3 2
5 2

2 0

Figure 3.21: Edge irregular reflexive labeling of f3,2

For m ≥ 3 we distinguish two cases.

Case 1. When m is even we define a m-labeling f of f3,m such that

f (x) = m − 2,
f (xj1 ) = 0 j = 1, 2, . . . , m − 2,
f (xj1 ) = m j = m − 1, m,
f (xj2 ) = 2d 2i e − 2 j = 1, 2, . . . , m − 2,
f (xj2 ) =m j = m − 1, m,
f (xxj1 ) =i j = 1, 2, . . . , m − 2,
f (xxm−1
1 ) = m − 3,
m
f (xx1 ) = m − 1,

58
 f (xxj2 ) = m − 1 j = 1, 3, . . . , m − 3,
f (xxj2 ) =m j = 2, 4, . . . , m,
f (xxm−12 ) = m − 2,
j j
f (x1 x2 ) =1 j = 1, 3, . . . , m − 3,
f (xj1 xj2 ) =2 j = 2, 4, . . . , m − 2,
f (xm−1
1 xm−1
2 ) = m − 1,
m m
f (x1 x2 ) = m.

Then we get

wtf (xxj1 ) =f (x) + f (xxj1 ) + f (xj1 ) = (m − 2) + j + 0 = m + j − 2

for j = 1, 2, . . . , m − 2,
wtf (xxm−1
1 ) =f (x) + f (xxm−1
1 ) + f (xm−1
1 ) = (m − 2) + (m − 3) + m = 3m − 5,
wtf (xxm m m
1 ) =f (x) + f (xx1 ) + f (x1 ) = (m − 2) + (m − 1) + m = 3m − 3,
wtf (xxj2 ) =f (x) + f (xxj2 ) + f (xj2 ) = (m − 2) + (m − 1) + (2d 2j e − 2) = 2m − 4 + j
for j = 1, 3, . . . , m − 3,
wtf (xxj2 ) =f (x) + f (xxj2 ) + f (xj2 ) = (m − 2) + m + (2d 2j e − 2) = 2m − 4 + j
for j = 2, 4, . . . , m − 2,
wtf (xxm−1
2 ) =f (x) + f (xxm−1
2 ) + f (xm−1
2 ) = (m − 2) + (m − 2) + m = 3m − 4,
m
wtf (xx2 ) =f (x) + f (xxm m
2 ) + f (x2 ) = (m − 2) + m + m = 3m − 2,
wtf (xj1 xj2 ) =f (xj1 ) + f (xj1 xj2 ) + f (xj2 ) = 0 + 1 + (2d 2j e − 2) = j
for j = 1, 3, . . . , m − 3,
wtf (xj1 xj2 ) =f (xj1 ) + f (xj1 xj2 ) + f (xj2 ) = 0 + 2 + (2d 2j e − 2) = j
for j = 2, 4, . . . , m − 2,
wtf (xm−1
1 xm−1
2 ) =f (xm−1
1 ) + f (xm−1
1 xm−1
2 ) + f (xm−1
2 ) = m + (m − 1) + m = 3m − 1,
m m m m m m
wtf (x1 x2 ) =f (x1 ) + f (x1 x2 ) + f (x2 ) = m + m + m = 3m.

Its is not difficult to see that the edge weights are {1, 2, . . . , 3m}. This shows that f is a
edge irregular reflexive labeling of f3,m for m ≥ 4 even.
Case 2. When m is odd we define a (m + 1)-labeling f of f3,m such that

f (x) = m − 1,
f (xj1 ) = 0 j = 1, 2, . . . , m − 1,
f (xm
1 ) = m − 1,
f (xj2 ) = 2d 2j e − 2 j = 1, 2, . . . , m − 1,
f (xm
2 ) = m + 1,
f (xxj1 ) =j j = 1, 2, . . . , m,

59
 f (xxj2 ) = m j = 1, 3, . . . , m − 2,
f (xxj2 ) =m+1 j = 2, 4, . . . , m − 1,
f (xxm 2 ) = m − 1,
j j
f (x1 x2 ) =1 j = 1, 3, . . . , m − 2,
f (xj1 xj2 ) =2 j = 2, 4, . . . , m − 1,
f (xm m
1 x2 ) = m.

Thus the vertices are labeled with even numbers and the edge weights are

wtf (xxj1 ) =f (x) + f (xxj1 ) + f (xj1 ) = (m − 1) + j + 0 = m + j − 1

for j = 1, 2, . . . , m − 1,
wtf (xxm
1 ) =f (x) + f (xxm m
1 ) + f (x1 ) = (m − 1) + m + (m − 1) = 3m − 2,
wtf (xxj2 ) =f (x) + f (xxj2 ) + f (xj2 ) = (m − 1) + m + (2d 2j e − 2) = 2m − 2 + j
for j = 1, 3, . . . , m − 2,
wtf (xxj2 ) =f (x) + f (xxj2 ) + f (xj2 ) = (m − 1) + (m + 1) + (2d 2j e − 2) = 2m − 2 + j
for j = 2, 4, . . . , m − 1,
wtf (xxm2 ) =f (x) + f (xxm m
2 ) + f (x2 ) = (m − 1) + (m − 1) + (m + 1) = 3m − 1,
j j
wtf (x1 x2 ) =f (xj1 ) + f (xj1 xj2 ) + f (xj2 ) = 0 + 1 + (2d 2j e − 2) = j
for j = 1, 3, . . . , m − 2,
wtf (xj1 xj2 ) =f (xj1 ) + f (xj1 xj2 ) + f (xj2 ) = 0 + 2 + (2d 2j e − 2) = j
for j = 2, 4, . . . , m − 1,
wtf (xm m
1 x2 ) =f (xm
1 ) + f (xm m m
1 x2 ) + f (x2 ) = (m − 1) + m + (m + 1) = 3.

Thus also for n odd, m ≥ 3, the edge weights are distinct numbers from the set {1, 2, . . . , 3m}.
This concludes the proof.

♦ Theorem 3.6.2. [19] For every positive integer m ≥ 1

(
d 4m
3 e if m ≡ 0, 1 (mod 3),
res(f4,m ) = 4m
d 3 e + 1 if m ≡ 2 (mod 3).

Proof. Let us denote m = 3r + t, where r ≥ 0 and t ∈ {0, 1, 2}.

As |E(f4,3r+t )| = 4(3r + t) = 12r + 4t then according to Lemma 3.2.1 we get

(
d 12r+4t
3 e if t 6= 2,
res(f4,3r+t ) ≥ 12r+4t
d 3 e+1 if t = 2

that is
res(f4,3r+t ) ≥ 4r + 2t

60
for every r ≥ 0 and t ∈ {0, 1, 2}.
We define a (4r + 2t)-labeling f of f4,3r+t such that

f (x) = 0,
f (xji ) = 0 j = 1, 2, . . . , r + t, i = 1, 3,
f (xji ) = 2j − 2 j = r + t + 1, r + t + 2, . . . , 2r + t, i = 1, 3,
f (xji ) = 4r + 2t j = 2r + t + 1, 2r + t + 2, . . . , 3r + t, i = 1, 3,
f (xj2 ) = 4r + 2t j = 1, 2, . . . , 3r + t,
f (xxj1 ) = 2j − 1 j = 1, 2, . . . , r + t,
f (xxj1 ) =1 j = r + t + 1, r + t + 2, . . . , 2r + t,
f (xxj1 ) = 6r + 2t + 2 − 2j j = 2r + t + 1, 2r + t + 2, . . . , 3r + t,
f (xxj3 ) = 2j j = 1, 2, . . . , r + t,
f (xxj3 ) =2 j = r + t + 1, r + t + 2, . . . , 2r + t,
f (xxj3 ) = 6r + 2t + 1 − 2j j = 2r + t + 1, 2r + t + 2, . . . , 3r + t,
f (xj1 xj2 ) = 2r − 1 + 2j j = 1, 2, . . . , r + t,
f (xj1 xj2 ) = 2r + 1 j = r + t + 1, r + t + 2, . . . , 2r + t,
f (xj1 xj2 ) = 8r + 2t + 2 − 2j j = 2r + t + 1, 2r + t + 2, . . . , 3r + t,
f (xj3 xj2 ) = 2r + 2j j = 1, 2, . . . , r + t,
f (xj3 xj2 ) = 2r + 2 j = r + t + 1, r + t + 2, . . . , 2r + t,
f (xj3 xj2 ) = 8r + 2t + 1 − 2j j = 2r + t + 1, 2r + t + 2, . . . , 3r + t.

Evidently, the vertices are labeled by even numbers and all the labels are not greater then 4r+2t.
Moreover, for the edge weights we get the following.

wtf (xxj1 ) =f (x) + f (xxj1 ) + f (xj1 ) = 0 + (2j − 1) + 0 = 2j − 1

for j = 1, 2, . . . , r + t,
wtf (xxj1 ) =f (x) + f (xxj1 ) + f (xj1 ) = 0 + 1 + (2j − 2) = 2j − 1
for j = r + t + 1, r + t + 2, . . . , 2r + t,
wtf (xxj1 ) =f (x) + f (xxj1 ) + f (xj1 ) = 0 + (6r + 2t + 2 − 2j) + (4r + 2t) = 10r + 4t + 2 − 2j
for j = 2r + t + 1, 2r + t + 2, . . . , 3r + t
wtf (xxj3 ) =f (x) + f (xxj3 ) + f (xj3 ) = 0 + 2j + 0 = 2j
for j = 1, 2, . . . , r + t
wtf (xxj3 ) =f (x) + f (xxj3 ) + f (xj3 ) = 0 + 2 + (2j − 2) = 2j
for j = r + t + 1, r + t + 2, . . . , 2r + t,
wtf (xxj3 ) =f (x) + f (xxj3 ) + f (xj3 ) = 0 + (6r + 2t + 1 − 2j) + (4r + 2t) = 10r + 4t + 1 − 2j
for j = 2r + t + 1, 2r + t + 2, . . . , 3r + t,

61
 wtf (xj1 xj2 ) =f (xj1 ) + f (xj1 xj2 ) + f (xj2 ) = 0 + (2r − 1 + 2j) + (4r + 2t) = 6r + 2t − 1 + 2j
for j = 1, 2, . . . , r + t,
wtf (xj1 xj2 ) =f (xj1 ) + f (xj1 xj2 ) + f (xj2 ) = (2j − 2) + (2r + 1) + (4r + 2t) = 6r + 2t − 1 + 2j
for j = r + t + 1, r + t + 2, . . . , 2r + t,
wtf (xj1 xj2 ) =f (xj1 ) + f (xj1 xj2 ) + f (xj2 ) = (4r + 2t) + (8r + 2t + 2 − 2j) + (4r + 2t)
=16r + 6t + 2 − 2j
for j = 2r + t + 1, 2r + t + 2, . . . , 3r + t,
wtf (xj3 xj2 ) =f (xj3 ) + f (xj3 xj2 ) + f (xj2 ) = 0 + (2r + 2j) + (4r + 2t) = 6r + 2t + 2j
for j = 1, 2, . . . , r + t,
wtf (xj3 xj2 ) =f (xj3 ) + f (xj3 xj2 ) + f (xj2 ) = (2j − 2) + (2r + 2) + (4r + 2t) = 6r + 2t + 2j
for j = r + t + 1, r + t + 2, . . . , 2r + t,
wtf (xj3 xj2 ) =f (xj3 ) + f (xj3 xj2 ) + f (xj2 ) = (4r + 2t) + (8r + 2t + 1 − 2j) + (4r + 2t)
=16r + 6t + 1 − 2j
for j = 2r + t + 1, 2r + t + 2, . . . , 3r + t.

Its is not difficult to see that the edge weights {1, 2, . . . , 12r + 4t}. This shows that f is a
edge irregular reflexive labeling of f4,m for m ≥ 1.

♦ Theorem 3.6.3. [19] For every positive integer m ≥ 1

(
d 5m
3 e if m 6≡ 3, 4 (mod 6),
res(f5,m ) = 5m
d 3 e + 1 if m ≡ 3, 4 (mod 6).

Proof. As the number of edges of f5,m is 5m then using Lemma 3.2.1 we obtain
(
d 5m
3 e if m ≡
6 3, 4 (mod 6),
res(f5,m ) ≥ k = 5m
d 3 e + 1 if m ≡ 3, 4 (mod 6).

It is easy to see that for m ≡ 5 (mod 6) the number k is odd and otherwise k is even.
As the graph f5,1 is isomorphic to C5 thus according to Theorem 3.2.5 we get res(C5 ) = 2.
From the lower bound for res(f5,m ) we get that res(f5,2 ) ≥ 4 and res(f5,3 ) ≥ 6. A corre-
sponding edge irregular reflexive labeling for f5,2 and an edge irregular reflexive labeling for f5,3
are illustrated in Figures 3.22 and 3.23 respectively.
We distinguish two cases according to the parity of k.
Let m ≥ 4.
Case 1. When m 6≡ 5 (mod 6), that is when k is an even number, we define a k-labeling f
of f5,m in the following way.

62
 7 0 2 3
3 2 0 3
0 0
4 0 4
4
2 2 1 3 1
10 2 1 6 1 0
3 5
15 3 3
9
5 6
4 0 0 6 0 0
6
1 8 2 4 14 2 4 10
9 4 4 6 2
6 1 6
8 7
4 2 1 1
13 11
2 1 6 4
5 6 2
12

Figure 3.22: Edge irregular reflexive labeling Figure 3.23: Edge irregular reflexive labeling
of f5,2 of f5,3

f (x) = 0,
f (xji ) = 0 j = 1, 2, . . . , m − 2d m−4
6 e, i = 1, 4,
f (xji ) = k − 2d m−4
6 e j = m − 2d m−4 m−4
6 e + 1, m − 2d 6 e + 2, . . . , m, i = 1, 4,
f (xji ) = k j = 1, 2, . . . , m, i = 2, 3,
f (xxj1 ) = 2j − 1 j = 1, 2, . . . , m − 2d m−4
6 e,
f (xxj1 ) = 2m + 2 − 2j j = m − 2d m−4 m−4
6 e + 1, m − 2d 6 e + 2, . . . , m,
f (xxj4 ) = 2j j = 1, 2, . . . , m − 2d m−4
6 e,
f (xxj4 ) = 2m + 1 − 2j j = m − 2d m−4 m−4
6 e + 1, m − 2d 6 e + 2, . . . , m,
f (xj1 xj2 ) = k + 2 − 2j j = 1, 2, . . . , m − 2d m−4
6 e,
f (xj1 xj2 ) = k − 2m − 4d m−4 m−4 m−4
6 e − 1 + 2j j = m − 2d 6 e + 1, m − 2d 6 e + 2, . . . , m,
m≡0 (mod 6),
f (xj1 xj2 ) = k − 2m − 4d m−4
6 e − 3 + 2j j = m − 2d m−4 m−4
6 e + 1, m − 2d 6 e + 2, . . . , m,
m ≡ 1, 2 (mod 6),
f (xj1 xj2 ) = k − 2m − 4d m−4 m−4 m−4
6 e − 7 + 2j j = m − 2d 6 e + 1, m − 2d 6 e + 2, . . . , m,
m≡3 (mod 6),
f (xj1 xj2 ) = k − 2m − 4d m−4 m−4 m−4
6 e − 9 + 2j j = m − 2d 6 e + 1, m − 2d 6 e + 2, . . . , m,
m≡4 (mod 6),
f (xj3 xj4 ) = k + 1 − 2j j = 1, 2, . . . , m − 2d m−4
6 e,
f (xj3 xj4 ) = k − 2m − 4d m−4
6 e + 2j j = m − 2d m−4 m−4
6 e + 1, m − 2d 6 e + 2, . . . , m,

63
 m≡0 (mod 6),
f (xj3 xj4 ) = k − 2m − 4d m−4 m−4 m−4
6 e − 2 + 2j j = m − 2d 6 e + 1, m − 2d 6 e + 2, . . . , m,
m ≡ 1, 2 (mod 6),
f (xj3 xj4 ) = k − 2m − 4d m−4
6 e − 6 + 2j j = m − 2d m−4 m−4
6 e + 1, m − 2d 6 e + 2, . . . , m,
m≡3 (mod 6),
f (xj3 xj4 ) = k − 2m − 4d m−4
6 e − 8 + 2j j = m − 2d m−4 m−4
6 e + 1, m − 2d 6 e + 2, . . . , m,
m≡4 (mod 6),
f (xj2 xj3 ) = k + 1 − j j = 1, 2, . . . , m, m ≡ 0 (mod 6),
f (xj2 xj3 ) =k−j j = 1, 2, . . . , m, m ≡ 1 (mod 6),
f (xj2 xj3 ) =k−1−j j = 1, 2, . . . , m, m ≡ 2 (mod 6),
f (xj2 xj3 ) =k−2−j j = 1, 2, . . . , m, m ≡ 3 (mod 6),
f (xj2 xj3 ) =k−3−j j = 1, 2, . . . , m, m ≡ 4 (mod 6).

For the edge weights we get

wtf (xxj1 ) =f (x) + f (xxj1 ) + f (xj1 ) = 0 + (2j − 1) + 0 = 2j − 1

for j = 1, 2, . . . , m − 2d m−4
6 e,
wtf (xxj1 ) =f (x) + f (xxj1 ) + f (xj1 ) = 0 + (2m + 2 − 2j) + (k − 2d m−4
6 e)
=2m + k − 2d m−4
6 e + 2 − 2j
for j = m − 2d m−4 m−4
6 e + 1, m − 2d 6 e + 2, . . . , m,
wtf (xxj4 ) =f (x) + f (xxj4 ) + f (xj4 ) = 0 + 2j + 0 = 2j
for j = 1, 2, . . . , m − 2d m−4
6 e,
wtf (xxj4 ) =f (x) + f (xxj4 ) + f (xj4 ) = 0 + (2m + 1 − 2j) + (k − 2d m−4
6 e)
=2m + k − 2d m−4
6 e + 1 − 2j
for j = m − 2d m−4 m−4
6 e + 1, m − 2d 6 e + 2, . . . , m,
wtf (xj1 xj2 ) =f (xj1 ) + f (xj1 xj2 ) + f (xj2 ) = 0 + (k + 2 − 2j) + k = 2k + 2 − 2j
for j = 1, 2, . . . , m − 2d m−4
6 e.

For j = m − 2d m−4 m−4

6 e + 1, m − 2d 6 e + 2, . . . , m we have

wtf (xj1 xj2 ) =f (xj1 ) + f (xj1 xj2 ) + f (xj2 ) = (k − 2d m−4 j j

6 e) + f (x1 x2 ) + k



 3k − 2m − 6d m−4 6 e − 1 + 2j, m ≡ 0 (mod 6),
3k − 2m − 6d m−4 e − 3 + 2j, m ≡ 1, 2 (mod 6),

= 6
m−4


 3k − 2m − 6d 6 e − 7 + 2j, m ≡ 3 (mod 6),
3k − 2m − 6d m−4 e − 9 + 2j, m ≡ 4 (mod 6).

6

64
Furthermore, the weights of edges xj3 xj4 for j = 1, 2, . . . , m − 2d m−4
6 e are

wtf (xj3 xj4 ) =f (xj3 ) + f (xj3 xj4 ) + f (xj4 ) = k + (k + 1 − 2j) + 0 = 2k + 1 − 2j,

and for j = m − 2d m−4 m−4

6 e + 1, m − 2d 6 e + 2, . . . , m the weights are

wtf (xj3 xj4 ) =f (xj3 ) + f (xj3 xj4 ) + f (xj4 ) = k + f (xj3 xj4 ) + (k − 2d m−4
6 e)

m−4


 3k − 2m − 6d 6 e + 2j, m ≡ 0 (mod 6),
3k − 2m − 6d m−4 e − 2 + 2j, m ≡ 1, 2 (mod 6),

= 6
m−4


 3k − 2m − 6d 6 e − 6 + 2j, m ≡ 3 (mod 6),
3k − 2m − 6d m−4 e − 8 + 2j, m ≡ 4 (mod 6).

6

And for j = 1, 2, . . . , m we get

wtf (xj2 xj3 ) =f (xj2 ) + f (xj2 xj3 ) + f (xj3 ) = k + f (xj2 xj3 ) + k



 3k + 1 − j, m ≡ 0 (mod 6),

3k − j, m ≡ 1 (mod 6),



= 3k − 1 − j, m ≡ 2 (mod 6),

3k − 2 − j, m ≡ 3 (mod 6),





3k − 3 − j, m ≡ 4 (mod 6).


Its is not difficult to see that the edge weights are distinct numbers from the set {1, 2, . . . , 5m}.
Case 2. When m ≡ 5 (mod 6), that is when k = d 5m
3 e is odd. Then we define a k-labeling
f of f5,m such that.

f (x) = 0,
f (xji ) = 0 j = 1, 2, . . . , m − 2d m−4
6 e, i = 1, 4,
f (xji ) = k − 1 − 2d m−4
6 e j = m − 2d m−4 m−4
6 e + 1, m − 2d 6 e + 2, . . . , m, i = 1, 4,
f (xji ) = k − 1 j = 1, 2, . . . , m, i = 2, 3,
f (xxj1 ) = 2j − 1 j = 1, 2, . . . , m − 2d m−4
6 e,
f (xxj1 ) = 2m + 2 − 2j j = m − 2d m−4 m−4
6 e + 1, m − 2d 6 e + 2, . . . , m,
f (xxj4 ) = 2j j = 1, 2, . . . , m − 2d m−4
6 e,
f (xxj4 ) = 2m + 1 − 2j j = m − 2d m−4 m−4
6 e + 1, m − 2d 6 e + 2, . . . , m,
f (xj1 xj2 ) = k + 1 − 2j j = 1, 2, . . . , m − 2d m−4
6 e,
f (xj1 xj2 ) = k − 2m − 4d m−4
6 e + 2j j = m − 2d m−4 m−4
6 e + 1, m − 2d 6 e + 2, . . . , m,
f (xj3 xj4 ) = k − 2j j = 1, 2, . . . , m − 2d m−4
6 e,
f (xj3 xj4 ) = k − 2m − 4d m−4 m−4 m−4
6 e + 1 + 2j j = m − 2d 6 e + 1, m − 2d 6 e + 2, . . . , m,
f (xj2 xj3 ) =k+1−j j = 1, 2, . . . , m.

65
The edge weights are

wtf (xxj1 ) =f (x) + f (xxj1 ) + f (xj1 ) = 0 + (2j − 1) + 0 = 2j − 1

for j = 1, 2, . . . , m − 2d m−4
6 e,
wtf (xxj1 ) =f (x) + f (xxj1 ) + f (xj1 ) = 0 + (2m + 2 − 2j) + (k − 1 − 2d m−4
6 e)
=2m + k − 2d m−4
6 e + 1 − 2j
for j = m − 2d m−4 m−4
6 e + 1, m − 2d 6 e + 2, . . . , m,
wtf (xxj4 ) =f (x) + f (xxj4 ) + f (xj4 ) = 0 + 2j + 0 = 2j
for j = 1, 2, . . . , m − 2d m−4
6 e,
wtf (xxj4 ) =f (x) + f (xxj4 ) + f (xj4 ) = 0 + (2m + 1 − 2j) + (k − 1 − 2d m−4
6 e)
=2m + k − 2d m−4
6 e − 2j
for j = m − 2d m−4 m−4
6 e + 1, m − 2d 6 e + 2, . . . , m,
wtf (xj1 xj2 ) =f (xj1 ) + f (xj1 xj2 ) + f (xj2 ) = 0 + (k + 1 − 2j) + (k − 1) = 2k − 2j
for j = 1, 2, . . . , m − 2d m−4
6 e,
wtf (xj1 xj2 ) =f (xj1 ) + f (xj1 xj2 ) + f (xj2 ) = (k − 1 − 2d m−4 m−4
6 e) + (k − 2m − 4d 6 e + 2j) + (k − 1)
=3k − 2m − 6d m−4
6 e − 2 + 2j
for j = m − 2d m−4 m−4
6 e + 1, m − 2d 6 e + 2, . . . , m,
wtf (xj3 xj4 ) =f (xj3 ) + f (xj3 xj4 ) + f (xj4 ) = (k − 1) + (k − 2j) + 0 = 2k − 1 − 2j
for j = 1, 2, . . . , m − 2d m−4
6 e,
wtf (xj3 xj4 ) =f (xj3 ) + f (xj3 xj4 ) + f (xj4 ) = k + (k − 2m − 4d m−4 m−4
6 e + 1 + 2j) + (k − 2d 6 e)
=3k − 2m − 6d m−4
6 e − 1 + 2j
for j = m − 2d m−4 m−4
6 e + 1, m − 2d 6 e + 2, . . . , m,
wtf (xj2 xj3 ) =f (xj2 ) + f (xj2 xj3 ) + f (xj3 ) = (k − 1) + (k + 1 − j) + (k − 1) = 3k − 1 − j
for j = 1, 2, . . . , m.

Also in this case the edge weights are {1, 2, . . . , 5m}.

Thus we constructed a edge irregular reflexive labeling of f5,m for m ≥ 1.

3.7 Conclusion

In this chapter we defined reflexive vertex strength and reflexive edge strength. We showed
that all graphs have reflexive edge strength and produced the reflexive edge strength for several
classes of graphs. For further investigation we state to solve the corresponding problem for the
reflexive vertex strength of graphs. In next chapter, we will recall the definition of reflexive
vertex strength and calculate this parameter for several graphs.

66
Chapter 4

Vertex Irregular Reflexive Labeling

4.1 Introduction

Since the irregularity strength of a graph G, s(G) is based on an edge labeling only, we
can visualise the same labeling scheme as a vertex irregular reflexive labeling with every vertex
having label 0. So we have the following theorem.

♦ Theorem 4.1.1. [100] For all graphs having s(G) is

rvs(G) ≤ s(G).

Nierhoff proved

Theorem 4.1.2. [92] Let G be a graph with no component of order ≤ 2, and G 6= K3 , then

s(G) ≤ |V (G)| − 1.

Using Nierhoff’s result and Theorem 4.1.1 we obtain

♦ Theorem 4.1.3. [100] Let G be a simple graph with no isolated vertices then,

rvs(G) ≤ |V (G)| − 1.

Proof. By Theorem 4.1.2, the theorem holds for all graphs with no component isomorphic to
K2 and not itself isomorphic to K3 .
For K3 label the edges 1, 1, 2 and the vertices with 0 except one vertex incident to edges
with different labels which receives the label 2. This graph is clearly vertex irregular reflexive
with rvs(K3 ) = 2 = |V (G)| − 1.
We will now consider graphs with connected components of order 2. Let V represent the
vertices in the entire graph and let V 0 represent the set of vertices in those components of the
graph not isomorphic to K2 .

67
 If there is a K2 component, label the edge with 1 and the vertices with 0 and 2. Label the
vertices of the remainder of the graph with 2, then by Nierhoff, the graph is still irregular and
all vertex weights are increased by 2. For r components isomorphic to K2 , label the r edges
with integers 1, 2, . . . , r and for the ith K2 , label the vertices i − 1, i + 1, i odd and i − 2, i, i even.
The maximum weight of the K2 components is 2r + 1 and occurs when r is odd. Adding 2r + 2
to the vertices of the remainder of the graph ensures that the graph remains vertex irregular.
The edge labels have not been changed so if 2r + 2 ≤ |V | − 1 then the theorem holds. If
2r + 2 is the new reflexive vertex strength then, since we have added 2r vertices, 2r + 2 ≤ |V | − 1
whenever |V 0 | ≥ 3. If the graph is isomorphic to rK2 , then the same labeling gives a reflexive
vertex strength of r, r even or r + 1, r odd. In either case rvs(rK2 ) ≤ 2r − 1.

♦ Corollary 4.1.4. [100] Let G be a simple graph with r isolated vertices, then

rvs(G) ≤ max{2r, |V (G)| − r − 1}.

Proof. The component(s) of the graph with minimum degree greater than 0 can, by Theorem
4.1.3, be labeled with rvs(G − Kr ) ≤ |V (G)| − r − 1. Label the r isolated vertices with even
integers from 0 to 2r, and add 2r to each remaining vertex. In this way the graph is vertex
irregular and the largest label must be no more than max{2r, |V (G)| − r − 1}.

♦ Observation 4.1.5. For any graph on n vertices with minimum degree δ the least possible
vertex weight is δ and the greatest vertex weight is at least δ + n − 1.

We now give the following lemma.

♦ Lemma 4.1.6. [117] The largest vertex weight of a graph G of order p and the minimum
degree δ under any vertex irregular reflexive labeling is at least

1. p + δ − 1 if p ≡ 0 (mod 4) or p ≡ 1 (mod 4) and δ ≡ 0 (mod 2) or p ≡ 3 (mod 4) and

δ ≡ 1 (mod 2),

2. p + δ otherwise.

Proof. Let f be a vertex irregular reflexive labeling of a graph G of order p and the minimum
degree δ. Let us denote the vertices of G by the symbols v1 , v2 , . . . , vp such that wtf (vi ) <
wtf (vi+1 ) for i = 1, 2, . . . , p − 1.
Then the vertex weight of a vertex v1 is
X X
wtf (v1 ) = f (v1 ) + f (uv1 ) ≥ 0 + 1 ≥ δ.
uv1 ∈E(G) uv1 ∈E(G)

As the vertex weights are distinct we get

wtf (vp ) ≥ wtf (v1 ) + p − 1 ≥ p + δ − 1.

68
Let us consider that wtf (vp ) = p + δ − 1 which means that

{wtf (vi ) : i = 1, 2, . . . , p} = {δ, δ + 1, . . . , p + δ − 1}.

Thus the sum of all vertex weights is

p p
X X p(p + 2δ − 1)
wtf (vi ) = (δ + i − 1) = .
2
i=1 i=1

Evidently, this sum must be an even integer as

p
X p
X X
wtf (vi ) = f (vi ) + 2 f (e)
i=1 i=1 e∈E(G)

and every vertex label is even. Thus

p(p + 2δ − 1) ≡ 0 (mod 4)

but it is not possible if p ≡ 1 (mod 4) and δ ≡ 1 (mod 2) or p ≡ 2 (mod 4) or p ≡ 3 (mod 4)

and δ ≡ 0 (mod 2).

For regular graphs we immediately get

♦ Corollary 4.1.7. [117] Let G an r-regular graph of order p. Then

 
p+r−1

 if p ≡ 0, 1 (mod 4),
 r+1
rvs(G) =  
 p+r

 if p ≡ 2, 3 (mod 4).
r+1

4.2 Vertex Irregular Reflexive Labeling for Some Graphs

Having shown that all graphs can bear a vertex irregular reflexive labeling, we now look at
the reflexive vertex strength of some classes of graphs. The following theorem gives rvs(K1,n ).

♦ Theorem 4.2.1. [100] The reflexive vertex strength for the star K1,n is
( 
n
if n 6≡ 2 (mod 4),
rvs(K1,n ) =  n2 
2 + 1 otherwise.

Proof. The vertex weight for v ∈ V (K1,n ) is calculated by adding the label of v and its incident
edge label. This is the same as the edge weight for the corresponding edge as given in Theorem
3.2.3. The bound and construction of the labeling scheme to prove this theorem is exactly the
same as that offered in the proof of Theorem 3.2.3.

Figures 4.1 and 4.2 provide rvs(K1,6 ) and rvs(K1,5 ) respectively.

69
 2 2
3 0
2
0 2
2
1 2 3 1
1 0 1
2 0 1 1 0
0
4 2
2 2 3
3 4 4
2 2 2 5
6
5

Figure 4.1: Vertex irregular reflexive labeling Figure 4.2: Vertex irregular reflexive labeling
of K1,6 of K1,5

Lemma 4.2.2. [100] (Missing Value Lemma) The sequence of vertex weights in a vertex
irregular reflexive graph cannot include an odd number of odd weights.

Proof. The proof follows immediate from Handshaking Lemma, which states that every finite
undirected graph has even number of vertices of odd degree.
X
deg(v) =2|E(G)|,
v∈V (G)
X X
deg(v) + deg(v) =2|E(G)|,
odd even
X X
deg(v) =2|E(G)| − deg(v).
odd even
P P
Since 2|E(G)| − deg(v) is an even number so deg(v) has to be even. Thus the number of
even odd
odd weights has to be even. This concludes the proof.

In next two theorems we establish rvs of path and cycle.

♦ Theorem 4.2.3. [100] The reflexive vertex strength for the path Pn is

n+1

 3  n ≡ 3, 6, 9 (mod 12),

n+2
rvs(Pn ) = n ≡ 2 (mod 12),
 n 3
otherwise.

3

Proof. We first show that the values mentioned in the theorem are tight lower bounds, and then
provide labeling schemes that achieve these bounds.
The least maximum weight for a path on n vertices is n. So the least maximum label must
be dn/3e.

70
 For n ≡ 6, 9 (mod 12), note that the sequence 1, 2, . . . , n, contains an odd number of odd
numbers and so, by the Missing Value Lemma, the least maximum weight must be at least n + 1.
For n ≡ 3 (mod 12) note that dn/3e is odd however we require an even integer label for the
vertex resulting in a least maximum weight of at least n + 1.
For n ≡ 2 (mod 12), both conditions are involved. We need a least maximum weight of n + 1
to have an even number of odd weights but n + 1 ≡ 3 (mod 12) so we need to consider a least
maximum weight of at least n + 2.
We now provide a labeling scheme that achieves these least minimum maximum labels, called
strengths, which we will refer to as s. We identify all vertices in the path Pn as v1 , v2 , . . . , vn and
edges e1 , e2 , . . . , en−1 . Note here that for n ≡ 3 (mod 12) has different strength than n ≡ 5, 7
(mod 12) but the same labeling scheme is applied.
Define the labeling scheme ψ as follows.

ψ(vi ) = 0 1 ≤ i ≤ 3.

Case 1. When n ≡ 0, 4 (mod 12)


i
4d 6 e − 2
 i ≡ 4 (mod 6),
ψ(vi ) = 4d 6i e i ≡ 5, 0 (mod 6),

 i
4b 6 c i ≡ 1, 2, 3 (mod 6), 4 ≤ i ≤ d n2 e + 1,
ψ(vn−i ) = 2b i+1 c 0 ≤ i ≤ d n2 e − 2,
 3


 4d 6i e − 3 i ≡ 1 (mod 6),
4d i e − 2

i ≡ 2, 4 (mod 6),
ψ(ei ) = 6
4d 6i e − 1
 i ≡ 3, 5 (mod 6),

4b i c

i ≡ 0 (mod 6), 1 ≤ i ≤ d n2 e,
6
ψ(en−i ) = 2d 3i e 1 ≤ i ≤ d n2 e − 1.

Case 2. When n ≡ 3, 5, 7 (mod 12)


i
4d 6 e − 2
 i ≡ 4 (mod 6),
ψ(vi ) = 4d 6i e i ≡ 5, 0 (mod 6),

 i
4b 6 c i ≡ 1, 2, 3 (mod 6), 4 ≤ i ≤ b n2 c,
ψ(vn−i ) = 2b i+1 c 0 ≤ i ≤ b n2 c,
 3


 4d 6i e − 3 i ≡ 1 (mod 6),
4d i e − 2

i ≡ 2, 4 (mod 6),
ψ(ei ) = 6
4d 6i e − 1
 i ≡ 3, 5 (mod 6),

4b i c

i ≡ 0 (mod 6), 1 ≤ i ≤ b n2 c,
6

71
 ψ(en−i ) = 2d 3i e 1 ≤ i ≤ b n2 c.

Case 3. When n ≡ 1 (mod 12)


i
4d 6 e − 2
 i ≡ 4 (mod 6),
ψ(vi ) = 4d 6i e i ≡ 5, 0 (mod 6),

 i
4b 6 c i ≡ 1, 2, 3 (mod 6), 4 ≤ i ≤ d n2 e,
ψ(vi ) = d n3 e − 3 i = d n2 e + 1,
ψ(vn−i ) = 2b i+1 c 0 ≤ i ≤ b n2 c − 2,
 3


 4d 6i e − 3 i ≡ 1 (mod 6),
4d i e − 2

i ≡ 2, 4 (mod 6),
ψ(ei ) = 6
4d 6i e − 1
 i ≡ 3, 5 (mod 6),

4b i c

i ≡ 0 (mod 6), 1 ≤ i ≤ b n2 c − 1,
6
ψ(ei ) = d n3 e i = d n2 e, b n2 c,
ψ(en−i ) = 2d 3i e 1 ≤ i ≤ b n2 c − 1.

Case 4. When n ≡ 2 (mod 12)


i
4d 6 e − 2
 i ≡ 4 (mod 6),
ψ(vi ) = 4d 6i e i ≡ 5, 0 (mod 6),

 i
4b 6 c i ≡ 1, 2, 3 (mod 6), 4 ≤ i ≤ d n2 e,
ψ(vi ) = d n+2
3 e i = d n2 e + 1,
ψ(vn−i ) = 2b i+1 c 0 ≤ i ≤ d n2 e − 2,
 3


 4d 6i e − 3 i ≡ 1 (mod 6),
4d i e − 2

i ≡ 2, 4 (mod 6),
ψ(ei ) = 6
i
4d 6e − 1 i ≡ 3, 5 (mod 6),



4b i c

i ≡ 0 (mod 6), 1 ≤ i ≤ d n2 e,
6
ψ(en−i ) = 2d 3i e 1 ≤ i ≤ d n2 e − 1.

Case 5. When n ≡ 6 (mod 12)


i
4d 6 e − 2
 i ≡ 4 (mod 6),
ψ(vi ) = 4d 6i e i ≡ 5, 0 (mod 6),

 i
4b 6 c i ≡ 1, 2, 3 (mod 6), 4 ≤ i ≤ d n2 e,
ψ(vn−i ) = 2b i+1
3 c 0 ≤ i ≤ d n2 e − 1,

72
 


4d 6i e − 3 i ≡ 1 (mod 6),
4d i e − 2

i ≡ 2, 4 (mod 6),
ψ(ei ) = 6
i
4d 6e − 1 i ≡ 3, 5 (mod 6),



4b i c

i ≡ 0 (mod 6), 1 ≤ i ≤ d n2 e,
6
ψ(en−i ) = 2d 3i e 1 ≤ i ≤ d n2 e − 1.

Case 6. When n ≡ 8 (mod 12)


i
4d 6 e − 2
 i ≡ 4 (mod 6),
ψ(vi ) = 4d 6i e i ≡ 5, 0 (mod 6),

 i
4b 6 c i ≡ 1, 2, 3 (mod 6), 4 ≤ i ≤ d n2 e,
ψ(vn−i ) = 2b i+1 c 0 ≤ i ≤ d n2 e − 1,
 3
i
4d 6 e − 3

 i ≡ 1 (mod 6),
4d i e − 2

i ≡ 2, 4 (mod 6),
ψ(ei ) = 6
i
4d 6e − 1 i ≡ 3, 5 (mod 6),



4b i c

i ≡ 0 (mod 6), 1 ≤ i ≤ d n2 e − 1,
6
ψ(ei ) = d n3 e i = d n2 e,
ψ(en−i ) = 2d 3i e 1 ≤ i ≤ d n2 e − 1.

Case 7. When n ≡ 9 (mod 12)


i
4d 6 e − 2
 i ≡ 4 (mod 6),
ψ(vi ) = 4d 6i e i ≡ 5, 0 (mod 6),

 i
4b 6 c i ≡ 1, 2, 3 (mod 6), 4 ≤ i ≤ d n2 e,
ψ(vn−i ) = 2b i+1 c 0 ≤ i ≤ d n2 e − 1,
 3


 4d 6i e − 3 i ≡ 1 (mod 6),
4d i e − 2

i ≡ 2, 4 (mod 6),
ψ(ei ) = 6
i
4d 6e − 1 i ≡ 3, 5 (mod 6),



4b i c

i ≡ 0 (mod 6), 1 ≤ i ≤ b n2 c,
6
ψ(en−i ) = 2d 3i e 1 ≤ i ≤ b n2 c.

Case 8. When n ≡ 10 (mod 12)


i
4d 6 e − 2
 i ≡ 4 (mod 6),
ψ(vi ) = 4d 6i e i ≡ 5, 0 (mod 6),

 i
4b 6 c i ≡ 1, 2, 3 (mod 6), 4 ≤ i ≤ d n2 e + 1,
ψ(vn−i ) = 2b i+1
3 c 0 ≤ i ≤ d n2 e − 2,

73
 


4d 6i e − 3 i ≡ 1 (mod 6),
4d i e − 2

i ≡ 2, 4 (mod 6),
ψ(ei ) = 6
i
4d 6e − 1 i ≡ 3, 5 (mod 6),



4b i c

i ≡ 0 (mod 6), 1 ≤ i ≤ d n2 e + 1,
6
ψ(en−i ) = 2d 3i e 1 ≤ i ≤ d n2 e − 1.

Case 9. When n ≡ 11 (mod 12)


i
4d 6 e − 2
 i ≡ 4 (mod 6),
ψ(vi ) = 4d 6i e i ≡ 5, 0 (mod 6),

 i
4b 6 c i ≡ 1, 2, 3 (mod 6), 4 ≤ i ≤ b n2 c,
ψ(vn−i ) = 2b i+1 c 0 ≤ i ≤ b n2 c,
 3
i
4d 6 e − 3

 i ≡ 1 (mod 6),
4d i e − 2

i ≡ 2, 4 (mod 6),
ψ(ei ) = 6
i
4d 6e − 1 i ≡ 3, 5 (mod 6),



4b i c

i ≡ 0 (mod 6), 1 ≤ i ≤ d n2 e,
6
ψ(en−i ) = 2d 3i e 1 ≤ i ≤ b n2 c − 1.

It is simple to check that the lower numbered vertices bear odd weights while the higher
numbers have even weights. The weights of the central vertices vary depending on their residue
modulo 12 but the strengths are the values presented as lower bounds.

Figure 4.3 and Figure 4.4 provide vertex irregular reflexive labelings of P7 and P8 , respec-
tively.

0 1 0 2 0 3 2 2 2 2 0 2 0
1 3 5 7 6 4 2

Figure 4.3: Vertex irregular reflexive labeling of P7

0 1 0 2 0 3 2 3 2 2 2 2 0 2 0
1 3 5 8 7 6 4 2

Figure 4.4: Vertex irregular reflexive labeling of P8

74
♦ Theorem 4.2.4. [100] The reflexive vertex strength for the cycle Cn is

n+1

 3  n ≡ 0, 1, 4, 5, 9 (mod 12),

n+2
rvs(Cn ) = 3 n ≡ 2, 3, 6, 10, 11 (mod 12),

 n+3 

3 n ≡ 7, 8 (mod 12).

Proof. We follow the convention of the previous theorem by showing that the values mentioned
in the theorem are lower bounds of for maximum weight and then offer a labeling scheme that
achieves these bounds.
The smallest possible vertex weight for Cn is 2, so the least maximum weight must be greater
than or equal to n + 1 yielding rvs(Cn ) ≥ d(n + 1)/3e. This value is the bound for n ≡ 0, 1, 4, 5, 9
(mod 12). For n ≡ 2, 3, 6, 10, 11 (mod 12) simple counting shows that the sequence contains an
odd number of odd numbers and so the Missing Value Lemma provides for this bound to be
d(n + 2)/3e. For n = 7, 8 the least maximum weight gives a residue of 9 (mod 12). When this
number is divided by 3, the result is three equal odd numbers which does not allow for a labeling.
In these cases the lower bound for the strength is d(n + 3)/3e.
We now provide a labeling scheme that achieves these strengths, which again we will refer to
as s. We have numbered all vertices as v1 , v2 , . . . vn and edges e1 , e2 , . . . , en . One can begin at
any arbitrary vertex v1 and number sequentially. We have numbered clockwise for convenience.
Here is how we will label vertices of Cn for different cases n ≥ 6.

ψ(vi ) = 0 1 ≤ i ≤ 3.

Case 1. When n ≡ 1, 2, 3, 4, 5, 8, 9 (mod 12)


i
4d 6 e − 2
 i ≡ 4 (mod 6),
ψ(vi ) = 4d 6 e i
i ≡ 5, 0 (mod 6),

 i
4b 6 c i ≡ 1, 2, 3 (mod 6), 4 ≤ i ≤ d n2 e,
ψ(vn−i ) = 2d i+1 e 0 ≤ i ≤ b n2 c − 1,
 3
i
4d 6 e − 3

 i ≡ 1 (mod 6),
4d i e − 2

i ≡ 2, 4 (mod 6),
ψ(ei ) = 6
i
4d 6e − 1 i ≡ 3, 5 (mod 6),



4b i c

i ≡ 0 (mod 6), 1 ≤ i ≤ d n2 e,
6
ψ(en−i ) = 2d i+2
3 e−1 0 ≤ i ≤ b n2 c − 1.

Case 2. When n ≡ 5 (mod 12)


i
4d 6 e − 2
 i ≡ 4 (mod 6),
ψ(vi ) = 4d 6i e i ≡ 5, 0 (mod 6),

 i
4b 6 c i ≡ 1, 2, 3 (mod 6), 4 ≤ i ≤ d n2 e,

75
 ψ(vn−i ) = 2d i+1 e 0 ≤ i ≤ b n2 c − 1,
 3


 4d 6i e − 3 i ≡ 1 (mod 6),
4d i e − 2

i ≡ 2, 4 (mod 6),
ψ(ei ) = 6
i
4d 6e − 1 i ≡ 3, 5 (mod 6),



4b i c

i ≡ 0 (mod 6), 1 ≤ i ≤ b n2 c,
6
ψ(ei ) = d n+1
3 e i = d n2 e, d n2 e + 1,
ψ(en−i ) = 2d i+2
3 e−1 0 ≤ i ≤ b n2 c − 2.

Case 3. when n ≡ 7 (mod 12)


i
4d 6 e − 2
 i ≡ 4 (mod 6),
ψ(vi ) = 4d 6i e i ≡ 5, 0 (mod 6),

 i
4b 6 c i ≡ 1, 2, 3 (mod 6), 4 ≤ i ≤ d n2 e + 1,
ψ(vn−i ) = 2d i+1 e 0 ≤ i ≤ b n2 c − 2,
 3
i


 4d 6 e − 3 i ≡ 1 (mod 6),
4d i e − 2

i ≡ 2, 4 (mod 6),
ψ(ei ) = 6
i
4d 6e − 1 i ≡ 3, 5 (mod 6),



4b i c

i ≡ 0 (mod 6), 1 ≤ i ≤ d n2 e,
6
ψ(en−i ) = 2d i+2
3 e−1 0 ≤ i ≤ b n2 c − 1.

Case 4. When n ≡ 10 (mod 12)


i
4d 6 e − 2
 i ≡ 4 (mod 6),
ψ(vi ) = 4d 6i e i ≡ 5, 0 (mod 6),

 i
4b 6 c i ≡ 1, 2, 3 (mod 6), 4 ≤ i ≤ d n2 e,
ψ(vn−i ) = 2d i+1 e 0 ≤ i ≤ b n2 c − 1,
 3


 4d 6i e − 3 i ≡ 1 (mod 6),
4d i e − 2

i ≡ 2, 4 (mod 6),
ψ(ei ) = 6
i
4d 6e − 1 i ≡ 3, 5 (mod 6),



4b i c

i ≡ 0 (mod 6), 1 ≤ i ≤ d n2 e − 1,
6
ψ(ei ) = d n+2
3 e i = d n2 e, d n2 e + 1,
ψ(en−i ) = 2d i+2
3 e−1 0 ≤ i ≤ b n2 c − 2.

76
Case 5. When n ≡ 11 (mod 12)

i
4d 6 e − 2
 i ≡ 4 (mod 6),
ψ(vi ) = 4d 6i e i ≡ 5, 0 (mod 6),

 i
4b 6 c i ≡ 1, 2, 3 (mod 6), 4 ≤ i ≤ d n2 e,
ψ(vn−i ) = 2d i+1 e 0 ≤ i ≤ b n2 c − 1,
 3


 4d 6i e − 3 i ≡ 1 (mod 6),
4d i e − 2

i ≡ 2, 4 (mod 6),
ψ(ei ) = 6
i
4d 6e − 1 i ≡ 3, 5 (mod 6),



4b i c

i ≡ 0 (mod 6), 1 ≤ i ≤ d n2 e,
6
ψ(ei ) = d n+2
3 e i = d n2 e + 1,
ψ(en−i ) = 2d i+2
3 e−1 0 ≤ i ≤ b n2 c − 2.

Case 6. When n ≡ 0 (mod 12)


i
4d 6 e − 2
 i ≡ 4 (mod 6),
ψ(vi ) = 4d 6i e i ≡ 5, 0 (mod 6),

 i
4b 6 c i ≡ 1, 2, 3 (mod 6), 4 ≤ i ≤ d n2 e,
ψ(vn−i ) = 2d i+1 e 0 ≤ i ≤ b n2 c − 1,
 3
i


 4d 6 e − 3 i ≡ 1 (mod 6),
4d i e − 2

i ≡ 2, 4 (mod 6),
ψ(ei ) = 6


 4d 6i e − 1 i ≡ 3, 5 (mod 6),
4b i c

i ≡ 0 (mod 6), 1 ≤ i ≤ d n2 e,
6
ψ(en−i ) = 2d i+2
3 e−1 0 ≤ i ≤ b n2 c − 1.

It will be easy to verify that the weights are distinct.

Figures 4.5 and 4.6 demonstrate vertex irregular reflexive labelings of C10 and C12 respec-
tively.

77
 7 5 7
3 9 2 5
2 0 2 3
2 4 0
10 4 2 3
3 2
0 11 3
4 0
4 1
4 1
12 4 0 2 13 4 0 2
4 1 5 1
4 4 2
11 2 4 12 4
3 3 1
1
2 2 4 2
3 10 3 2 3 6
8 6
8

Figure 4.5: Vertex irregular reflexive labeling Figure 4.6: Vertex irregular reflexive labeling
of C10 of C12

♦ Theorem 4.2.5. [100] For n ≥ 2 we have

rvs(Kn ) = 2.

Proof. Trivially, rvs(Kn ) ≥ 2 for any n ≥ 2.

To show equality, we define a suitable vertex irregular reflexive labeling ψ as follows. Let
V (Kn ) = {v1 , v2 , . . . , vn }.
(
0 for 1 ≤ i ≤ d n2 e,
ψ(vi ) =
2 otherwise.

and
(
1 for i ≤ j ≤ n − i + 1, i 6= j,
ψ(vi vj ) =
2 otherwise.

This labeling yields weights from the sequence {n − 1, n, . . . , 2n − 1} with one weight from
the sequence missing. The missing weight is d(3n − 2)/2e.

Figure 4.7 provides vertex irregular reflexive labeling of K6 .

4.3 Vertex Irregular Reflexive Labeling of Prisms

In this section we will deal with the reflexive vertex strength of prism. First we recall
the definition of prism. The prism Dn , n ≥ 3, is a trivalent graph which can be defined

78
 6 1 0 5
0
1 1 2 1 1 1
1 1

7
2
2
0 2 11

1 2
2
9 2 2
2 10

Figure 4.7: Vertex irregular reflexive labeling of K6

as the Cartesian product P2 2Cn of a path on two vertices with a cycle on n vertices. We
denote the vertex set and the edge set of Dn such that V (Dn ) = {xi , yi : i = 1, 2, . . . , n} and
E(Dn ) = {xi xi+1 , yi yi+1 , xi yi : i = 1, 2, . . . , n}, where indices are taken modulo n. The next
theorem gives rvs for prisms.

♦ Theorem 4.3.1. [117] For n ≥ 3,

lnm
rvs(Dn ) = + 1.
2

Proof. As the prism Dn is 3-regular graph of order 2n, using from Corollary 4.1.7 we get that
lnm
rvs(Dn ) ≥ + 1.
2

We define the total labeling f of Dn in the following way

i = 1, 2, . . . , n2 + 1,
 
f (xi ) = f (yi ) = 0
f (xi ) = f (yi ) = n2 i = n2 + 2, n2 + 3, . . . , n,
     

and n ≡ 0, 3 (mod 4),

n
i = 2 + 2, n2 + 3, . . . , n,
n  
f (xi ) = f (yi ) = 2 +1
and n ≡ 1, 2 (mod 4),
f (xi xi+1 ) = 1 i = 1, 2, . . . , n − 1,
f (x1 xn ) = 1,
f (yi yi+1 ) = n2 + 1
 
i = 1, 2, . . . , n − 1,
f (y1 yn ) = n2 + 1,
 

i = 1, 2, . . . , n2 + 1,
 
f (xi yi ) = i
n
i = n2 + 2, n2 + 3, . . . , n,
   
f (xi yi ) = i − 2

79
 and n ≡ 0, 3 (mod 4),
n
i = 2 + 2, n2 + 3, . . . , n,
n  
f (xi yi ) = i − 1 − 2
and n ≡ 1, 2 (mod 4).

Evidently f is (dn/2e + 1)-labeling and the vertices are labeled with even numbers.
For the vertex weights of the vertices xi , i = 1, 2, . . . , dn/2e + 1 in Dn under the labeling f
we get

wtf (xi ) =0 + 1 + 1 + i = i + 2.

If i = dn/2e + 2, dn/2e + 3, . . . , n and n ≡ 0, 3 (mod 4) then

wtf (xi ) = n2 + 1 + 1 + i − n2 = i + 2
   

and for i = dn/2e + 2, dn/2e + 3, . . . , n and n ≡ 1, 2 (mod 4)

wtf (xi ) = n2 + 1 + 1 + 1 + i − 1 − n2 = i + 2.
 
 

Thus {wtf (xi ) : i = 1, 2, . . . , n} = {3, 4, . . . , n + 2}.

For the vertex weights of the vertices yi , i = 1, 2, . . . , n in Dn under the labeling f we have
the following

wtf (yi ) =0 + n2 + 1 + n2 + 1 + i = i + 2 + 2 n2
 
 
 

for i = 1, 2, . . . , n2 + 1,
 

wtf (yi ) = n2 + n2 + 1 + n2 + 1 + i − n2 = i + 2 + 2 n2
   
 
 
 

for i = n2 + 2, n2 + 3, . . . , n, and n ≡ 0, 3 (mod 4),

   

wtf (yi ) = n2 + 1 + n2 + 1 + n2 + 1 + i − 1 − n2 = i + 2 + 2 n2
 
 
 
 
 

for i = n2 + 2, n2 + 3, . . . , n, and n ≡ 1, 2 (mod 4).

   

Which means

{wtf (yi ) : i = 1, 2, . . . , n} ={2 n2 + 3, 2 n2 + 4, . . . , n + 2 n2 + 2}

     
(
{n + 3, n + 4, . . . , 2n + 2} for n even,
=
{n + 4, n + 5, . . . , 2n + 3} for n odd.

Thus the vertex weights are all distinct, that is f is a vertex irregular reflexive labeling of a prism
Dn .

Figures 4.8 and 4.9 provide vertex irregular reflexive labelings of D9 and D12 respectively.

80
 15 18
16 0 19 0 17
6 7 7
0 0 0
3 6 14 4
6 4 1 0
20
7 5 1 1 3 7 16
0 1 2 0 0
17 0 1 0 1 0 0 1 2 0
6 5 0 6
7 6 7 6 5
4 1 0 0 7
5 0 7 1 8 4 1 1
6 1 3 0 1 0 13 21 0 0 9 3 0 0 15
7
0 8 11 1 1 10 14 1 7
7 6 11
18 0
6 9 10 6 6
12 13 6 6
1 6 3 6 2 1 6 6 1 6
6 1 6
6 1 6 22 1 1 5 26
1 2 7 3 7
6 21 4
6 6 6
6 6 7 7
19 23 6 25
20 24

Figure 4.8: Vertex irregular reflexive labeling Figure 4.9: Vertex irregular reflexive labeling
of D9 of D12

4.4 Vertex Irregular Reflexive Labeling of Wheels

As defined in Section 3.4, the wheel Wn , n ≥ 3, is a graph obtained by joining all vertices of
Cn to a further vertex called the centre. We denote the vertex set and the edge set of Wn such
that V (Wn ) = {x, xi : i = 1, 2, . . . , n} and E(Wn ) = {xi xi+1 , xxi : i = 1, 2, . . . , n}, where indices
are taken modulo n. The wheel is of order n + 1 and size 2n. We prove the following result for
wheels.

♦ Theorem 4.4.1. [117] For n ≥ 3,

(
d n+2
4 e if n 6≡ 2 (mod 8),
rvs(Wn ) = n+2
d 4 e+1 if n ≡ 2 (mod 8).

Proof. Let n ≥ 3. As δ(Wn ) = 3 then the smallest vertex weight is at least 3. The wheel Wn
contains n vertices of degree 3 thus the largest weight over all vertices of degree 3 is at least
n + 2. Every vertex weight of a vertex of degree 3 is the sum of four labels from which at least
one is even thus we have  
n+2
rvs(Wn ) ≥ .
4
However, if n = 8t + 2, t ≥ 1, we get that the fraction
 n+2   8t+4 
4 = 4 = 2t + 1

is odd. The number n + 2 = 8t + 4 can be realizable as the sum of four labels not greater that

81
2t + 1 only in the following way

n + 2 = 8t + 4 = (2t + 1) + (2t + 1) + (2t + 1) + (2t + 1),

but it is a contradiction as the vertex label must be even. Thus, for n ≡ 2 (mod 8) we get

rvs(Wn ) ≥ n+2
 
4 + 1.

Let
(
d n+2
4 e if n 6≡ 2 (mod 8),
R= n+2
d 4 e+1 if n ≡ 2 (mod 8).

Let us denote by K the largest even number not greater than R. Thus
(
R if n ≡ 2, 3, 4, 5, 6 (mod 8),
K=
R − 1 if n ≡ 0, 1, 7 (mod 8).

For n = 3, 4 we get that rvs(Wn ) ≥ 2. The corresponding vertex irregular reflexive labelings
for W3 and W4 are illustrated on Figures 4.10 and 4.11 respectively.
For n ≥ 5 we define the total R-labeling f of Wn such that

f (x) = K,
f (xi ) = 0 i = 1, 2, . . . , 2R + K − 2, i ≤ n − 1,
f (xi ) = K i = 2R + K − 1, 2R + K, . . . , n,
f (xi x) = 3i
 
i = 1, 2, . . . , 2R + K − 2, i ≤ n − 1,
f (xi x) = i + 3 − 2R − K i = 2R + K − 1, 2R + K, . . . , n − 1,
f (xn x) = R,
f (xi xi+1 ) = i−1
 
3 +1 i = 1, 2, . . . , 2R + K − 2, i ≤ n − 1,
f (xi xi+1 ) = R i = 2R + K − 1, 2R + K, . . . , n − 1,
f (x1 xn ) = 1.

82
 0 7 6
3 2
2 0

1 2 2
1
6 2
2 1 2
1 8
2
1 1 1

0 2
2 0 0
4 7 1
3 4

Figure 4.10: Vertex irregular reflexive labeling Figure 4.11: Vertex irregular reflexive labeling
of W3 of W4
For the vertex weights of the vertices of degree 3 under the labeling f we get

wtf (x1 ) =0 + 1 + 1 + 1 = 3,
wtf (xi ) =0 + 3i + i−2 + 1 + i−2
   
 

3 3 +1 =i+2
for i = 2, 3, . . . , 2R + K − 2, i ≤ n − 1,
 2R+K−3 

wtf (x2R+K−1 ) =K + 2 + 3 + 1 + R = 2R + K + 2,
wtf (xi ) =K + (i + 3 − 2R − K) + R + R = i + 3
for i = 2R + K, 2R + K + 1, . . . , n − 1,
wtf (xn ) =K + 1 + R + R = 2R + K + 1.

It is easy to get that the weights of vertices xi , i = 1, 2, . . . , n, n ≥ 5 and n 6= 10 are distinct

numbers from the set {3, 4, . . . , n + 2}. For n = 10 we get {wtf (xi ) : i = 1, 2, . . . , 10} =
{3, 4, . . . , 11, 13}.
The weight of the vertex x is
n
X n
X n−1
X
wtf (x) = f (x) + f (xi x) = K + f (xi x) = K + R + f (xi x) > K + R + n − 1.
i=1 i=1 i=1

Evidently, for n ≥ 5, the vertex weights are distinct.

Figure 4.12 provides vertex irregular reflexive labeling of wheel W12 .

Let us recall from Section 2.1 that a fan graph graph Fn is obtained from wheel Wn if one
rim edge, say x1 xn is deleted. A basket Bn is obtained by removing a spoke, say xxn , from
wheel Wn . Before we will give the exact value of reflexive vertex strength of fan graph graphs
and baskets we give the following observation.

♦ Observation 4.4.2. [117] Let f be a vertex irregular reflexive k-labeling of a graph G. If

83
 3
13 1 0 1 4
4 0

4 1 2
14 4 1 5
4 0
2
4 1 2
4 32 2
12 0 0 6
4
4 2 2
3
0 0
11 3 2 7
4 3 3

0 0
10 3 0 3 8
9

Figure 4.12: Vertex irregular reflexive labeling of W12

there exists an edge uv in G such that

wtf (u) − f (uv) 6∈ {wtf (x) : x ∈ V (G) − {v}}

wtf (v) − f (uv) 6∈ {wtf (x) : x ∈ V (G) − {u}}

then f is a vertex irregular reflexive k-labeling of a graph G − {uv}. It should be noted here that
removing an edge yields no repetition.

Proof. The proof is trivial.

Immediately from this observation we get the corollary.

♦ Corollary 4.4.3. [117] Let rvs(G) = k and let f be the corresponding vertex irregular reflexive
k-labeling of a graph G. If the vertex weights of vertices u, v are the smallest over all vertex
weights under the labeling f and uv ∈ E(G) then

rvs(G − {uv}) ≤ rvs(G).

For the fan graph Fn we prove the following theorem.

♦ Theorem 4.4.4. [117] For n ≥ 3,

(
d n+1
4 e if n 6≡ 3 (mod 8),
rvs(Fn ) = n+1
d 4 e+1 if n ≡ 3 (mod 8).

84
Proof. The fan graph Fn contains two vertices of degree 2, thus the smallest vertex weight is at
least 2. The fan graph Fn contains n − 2 vertices of degree 3, thus the largest weight of a vertex
of degree 3 is at least n.
If all vertex weight of vertices of degree 3 are at most n, then one of the vertices of degree
2 has to have weight at least n + 1 and thus rvs(Fn ) ≥ d(n + 1)/3e. If a vertex of degree 3 has
weight greater than n then rvs(Fn ) ≥ d(n + 1)/4e. As we are trying to minimize the parameter
k for which there exists vertex irregular reflexive k-labeling of Fn we get

rvs(Fn ) ≥ d n+1
4 e

which can be obtain when both vertices of degree 2 in Fn will have weights less than n.
According to the proof of Theorem 4.4.1 and Corollary 4.4.3, for n ≥ 5, we get
(
d n+2
4 e if n 6≡ 2 (mod 8),
rvs(Fn ) ≤ rvs(Wn ) = n+2
d 4 e + 1 if n ≡ 2 (mod 8).

Moreover, we can derive vertex irregular reflexive rvs(Wn )-labeling of Fn from vertex irregular
reflexive rvs(Wn )-labeling of Wn .
Combining the previous facts we get that for n 6≡ 2, 3, 7 (mod 8)

rvs(Fn ) = d n+1
4 e

and n ≡ 2, 3, 7 (mod 8)
d n+1 n+1
4 e ≤ rvs(Fn ) ≤ d 4 e + 1.

For n = 3, 4 we get that rvs(Fn ) ≥ 2. The corresponding reflexive vertex irregular 2-labelings
for F3 and F4 are illustrated on Figure 4.13 and Figure 4.14 respectively.
Let n = 8t + 3, t ≥ 1, then d(n + 1)/4e = 2t + 1. As this is odd we can not get the number
n + 1 = 8t + 4 as the sum of four labels less or equal to 2t + 1 from which at least one is even.
Thus rvs(Fn ) = d(n + 1)/4e + 1 but in this case

rvs(Wn ) = d n+2 n+1

4 e+1=d 4 e+1

and we are done.

We denote the vertex set and the edge set of Fn such that V (Fn ) = {x, xi : i = 1, 2, . . . , n}
and E(Fn ) = {xi xi+1 , xxi : i = 1, 2, . . . , n − 1}.
If n = 8t + 2, t ≥ 1 then d(n + 1)/4e = 2t + 1. We define (2t + 1)-labeling of Fn such that

f (x) = 2t,
f (xi ) = 0 i = 1, 2, . . . , 6t,
f (xi ) = 2t i = 6t + 1, 6t + 2, . . . , 8t + 2,
f (x1 x) = 1,
f (xi x) = i−1
 
3 i = 2, 3, . . . , 6t,

85
 f (xi x) = i − 6t i = 6t + 1, 6t + 2, . . . , 8t + 1,
f (x8t+2 x) = 2t + 1,
f (xi xi+1 ) = i−2
 
3 +1 i = 1, 2, . . . , 6t,
f (xi xi+1 ) = 2t + 1 i = 6t + 1, 6t + 2, . . . , 8t + 1.

0 5 2
5
2 0
2 1 1
1
4 2
0 2 1
2 6
1
1 1 1

0 0
2 3 0 0
4 1 3

Figure 4.13: Vertex irregular reflexive labeling Figure 4.14: Vertex irregular reflexive labeling
of F3 of F4
It is easy to get that the set of all vertex weights is {2, 3, . . . , 8t + 3, 8t2 + 8t + 3}.
If n = 8t + 7, t ≥ 0 then d(n + 1)/4e = 2t + 2. We define (2t + 2)-labeling of Fn such in the
following way

f (x) = 2t + 2,
f (xi ) = 0 i = 1, 2, . . . , 6t + 4,
f (xi ) = 2t + 2 i = 6t + 5, 6t + 6, . . . , 8t + 7,
f (x1 x) = 1,
f (xi x) = i−1
 
3 i = 2, 3, . . . , 6t + 4,
f (xi x) = i − 6t − 4 i = 6t + 5, 6t + 6, . . . , 8t + 6,
f (x8t+7 x) = 2t + 2,
f (xi xi+1 ) = i−2
 
3 +1 i = 1, 2, . . . , 6t + 4,
f (xi xi+1 ) = 2t + 2 i = 6t + 5, 6t + 6, . . . , 8t + 6.

Evidently the vertex weights are distinct.

♦ Theorem 4.4.5. [117] For n ≥ 3,

(
d n+1
4 e if n 6≡ 3 (mod 8),
rvs(Bn ) = n+1
d 4 e+1 if n ≡ 3 (mod 8).

Proof. The basket Bn contains one vertex of degree 2, thus the smallest vertex weight is at least
2 and it contains n − 1 vertices of degree 3, thus the largest weight of a vertex of degree 3 is at

86
least n + 1. Thus

rvs(Bn ) ≥ d n+1
4 e.

Analogously as in the proof of the previous theorem, using Theorem 4.4.1 and Lemma 4.4.2, for
n ≥ 5, we have,
(
d n+2
4 e if n 6≡ 2 (mod 8),
rvs(Bn ) ≤ rvs(Wn ) = n+2
d 4 e + 1 if n ≡ 2 (mod 8).

Moreover, we can derive a vertex irregular reflexive rvs(Wn )-labeling of Bn from the vertex
irregular reflexive rvs(Wn )-labeling of Wn defined in the proof of Theorem 4.4.1 by deleting the
spoke x1 x in Wn .
Combining the previous facts we get that for n 6≡ 2, 3, 7 (mod 8)

rvs(Bn ) = d n+1
4 e

and n ≡ 2, 3, 7 (mod 8)
d n+1 n+1
4 e ≤ rvs(Bn ) ≤ d 4 e + 1.

For n = 3, 4 we get that rvs(Bn ) ≥ 2. The basket B3 is isomorphic to the fan graph F3 . The
vertex irregular reflexive 2-labelings for B4 is illustrated on Figure 4.15.

4 1 6
2 2
2

2
1 1
8
2 2

0 2
5 2 7

Figure 4.15: Vertex irregular reflexive labeling of B4

Let n = 8t + 3, t ≥ 1, then d(n + 1)/4e = 2t + 1. As this is odd we can not get the number
n + 1 = 8t + 4 as the sum of four labels less or equal to 2t + 1 from which at least one is even.
Thus rvs(Bn ) = d(n + 1)/4e + 1 but in this case

rvs(Wn ) = d n+2 n+1

4 e+1=d 4 e+1

and we are done.

Let us denote the vertex set and the edge set of the basket Bn such that V (Bn ) = {x, xi :

87
i = 1, 2, . . . , n} and E(Bn ) = {xxi : i = 2, 3, . . . , n} ∪ {xi xi+1 : i = 1, 2, . . . , n − 1} ∪ {xn x1 }.
2
6 5 13 1 0 3
2 4 0
0 0
2 4 2
7 2 4 1
0 14 4 1 5
0 4 0
2 1 2
3 2 1 4 1 2
1
2 4 32 2
8 0 4 25 0 2 12 0 0 6
4
3
3 1 4 2 2
4
3 3 3
0 0 0
4 11 3 2 7
9 3 13 3
4 4 3
0 4 0
0 0
10 11 10 3 0 3 8
9

Figure 4.16: Vertex irregular reflexive labeling Figure 4.17: Vertex irregular reflexive labeling
of B10 of F12

If n = 8t + 2, t ≥ 1 then d(n + 1)/4e = 2t + 1. We define (2t + 1)-labeling of Bn such that

f (x) = 2t,
f (xi ) = 0 i = 1, 2, . . . , 6t + 1,
f (xi ) = 2t i = 6t + 2, 6t + 3, . . . , 8t + 2,
f (xi x) = i−1
 
3 i = 2, 3, . . . , 6t + 1,
f (xi x) = i − 6t i = 6t + 2, 6t + 3, . . . , 8t + 1,
f (x8t+2 x) = 2t + 1,
f (xi xi+1 ) = i−2
 
3 +1 i = 1, 2, . . . , 6t,
f (xi xi+1 ) = 2t + 1 i = 6t + 1, 6t + 2, . . . , 8t + 1,
f (x1 x8t+2 ) = 1.

If n = 8t + 7, t ≥ 0 then d(n + 1)/4e = 2t + 2. We define (2t + 2)-labeling of Bn such in the

following way

f (x) = 2t + 2,
f (xi ) = 0 i = 1, 2, . . . , 6t + 5,
f (xi ) = 2t + 2 i = 6t + 6, 6t + 7, . . . , 8t + 7,
f (xi x) = i−1
 
3 i = 2, 3, . . . , 6t + 5,

88
 f (xi x) = i − 6t − 4 i = 6t + 6, 6t + 7, . . . , 8t + 6,
f (x8t+7 x) = 2t + 2,
f (xi xi+1 ) = i−2
 
3 +1 i = 1, 2, . . . , 6t + 5,
f (xi xi+1 ) = 2t + 2 i = 6t + 6, 6t + 7, . . . , 8t + 6,
f (x1 x8t+7 ) = 1.

It is not difficult to show that in both cases the described labelings have desired properties.

Figures 4.4 and 4.4 demonstrate vertex irregular reflexive labelings of B10 and F12 , respec-
tively.

4.5 Conclusion

In this chapter we have described vertex irregular reflexive labeling and gave results for the
reflexive vertex strength of several graphs and established their labeling patterns.
In next chapter, we will describe another type of labeling that is H-antimagic labeling of
graphs.

89
90
Chapter 5

H-antimagic Labeling

5.1 Introduction

An edge-covering of a graph G is a family of subgraphs H1 , H2 , . . . , Ht such that each edge

of E belongs to at least one of the subgraphs Hi , i = 1, 2, . . . , t. Then it is said that G admits
an (H1 , H2 , . . . , Ht )-(edge) covering. If every subgraph Hi is isomorphic to a given graph H,
then the graph G admits an H-covering. Note that in this case, every subgraph isomorphic to
H must be in the H-covering.
Gutiérrez and Lladó [52] defined an H-magic labeling as follows.
Definition 5.1.1. The graph G admitting an H-covering is called H-magic if there exists a
total labeling f : V (G) ∪ E(G) → {1, 2, . . . , |V (G)| + |E(G)|} such that, for each subgraph H 0
isomorphic to H, X X
f (v) + f (e) = k
v∈V (H 0 ) e∈E(H 0 )

, where k is constant.
When f (V (G)) = {1, 2, . . . , |V (G)|}, we say that G is H-supermagic.

. The H-(super)magic labelings are an extension of the edge-magic and super edge-magic
labelings introduced by Kotzig and Rosa [73] and Enomoto et al., [38], respectively. In [52],
star-(super)magic and path-(super)magic labelings of some connected graphs were considered
and proved that the path Pn and the cycle Cn are Ph -supermagic for some value of h. Lladó
and Moragas [75] studied the cycle-(super)magic behavior of several classes of connected graphs.
They proved that wheels, windmills, books and prisms are Ch -magic for some value of h. Maryati
et al., [85] and also Salman et al., [101] proved that certain families of trees are path-supermagic.
Ngurah et al., [91] proved that chains, wheels, triangles, ladders and grids are cycle-supermagic.
Maryati et al., [84] investigated the G-supermagicness of a disjoint union of c copies of a graph
G and showed that the disjoint union of any paths is cPh -supermagic for some c and h.
Combining the (a, d)-edge-antimagic total labeling, defined in Definition 2.2.11 and H-
antimagic labeling, Inayah et al., [60] introduced an (a, d)-H-antimagic labeling of a graph G

91
admitting an H-covering as a bijective function f : V (G) ∪ E(G) → {1, 2, . . . , |V (G)| + |E(G)|}
such that for all subgraphs H 0 isomorphic to H, the H 0 -weights
X X
wtf (H 0 ) = f (v) + f (e)
v∈V (H 0 ) e∈E(H 0 )

form an arithmetic progression a, a + d, . . . , a + (t − 1)d, where a > 0 and d ≥ 0 are two integers,
and t is the number of all subgraphs of G isomorphic to H. Such a labeling is called super if the
smallest possible labels appear on the vertices. A graph that admits a (super) (a, d)-H-antimagic
labeling is called (super) (a, d)-H-antimagic.
In [61], super (a, d)-H-antimagic labelings for some shackles of a connected graph H are
investigated. In [104] was proved that wheels are cycle-antimagic. The existence of super (a, 1)-
tree-antimagic labelings for disconnected graphs are studied in [21].
The (super) (a, d)-H-antimagic labeling is related to a super d-antimagic labeling of type
(1, 1, 0) of a plane graph that is the generalization of a face-magic labeling introduced by Lih
[76]. Further information on super d-antimagic labelings can be found in [17, 26].
Let G be an arbitrary graph and H be a connected graph of order at least 2. We define a
graph operation GH in the following way.

1. Denote the edges in G arbitrarily by e1 , e2 , . . . , e|E(G)| ;

2. Take |E(G)| copies of H, say H1 , H2 , . . . , H|E(G)| ;

3. In every Hi , i = 1, 2, . . . , |E(G)|, choose two distinct adjacent vertices, say xi , yi ;

4. Replace every edge ei in E(G) by subgraph Hi in such a way that its end vertices and
xi , yi ∈ V (Hi ) are identified.

The resulting graph GH is of order (|V (H)| − 2)|E(G)| + |V (G)| and size |E(H)||E(G)|. Note
that the graph GH is not defined uniquely. It means for graphs G and H there may be many
non-isomorphic graphs obtained by using this construction.
In this chapter, we investigate the existence of super (a, d)-H-antimagic labelings for GH . We
show connection between H-antimagic labelings and edge-antimagic total labelings and describe
a construction how to obtain the H-antimagic graph from a smaller edge-antimagic total graph
G.

5.2 Partitions with Determined Differences

For construction H-antimagic labelings of graphs we will use the partitions of a set of integers
with determined differences. This concept was introduced in [16].
n of the set {1, 2, . . . , kn}
Let n, k, d and i be positive integers. We will consider the partition Pk,d
into n, n ≥ 2, k-tuples such that the difference between the sum of the numbers in the (i + 1)th
k-tuple and the sum of the numbers in the ith k-tuple is always equal to the constant d, where

92
i = 1, 2, . . . , n − 1. Thus these sums form an arithmetic sequence with the difference d. By
the symbol Pk,d n (i) we denote the ith k-tuple in the partition with the difference d, where

i = 1, 2, . . . , n.
P n n
P n
P nPk,d (i) be the sum of the numbers in Pk,d (i). Evidently, from the definition, Pk,d (i+
Let
1) − Pk,d (i) = d. It is obvious that if there exists a partition of the set {1, 2, . . . , kn} with the
difference d, there also exists a partition with the difference −d. By the notation Pk,d n (i) ⊕ c we

mean that we add the constant c to every number in Pk,d n (i).

If k = 1 then only the following partition of the set {1, 2, . . . , n} is possible

n
P1,1 (i) = {i} for i = 1, 2, . . . , n.

If k = 2 then we have several partitions of the set {1, 2, . . . , 2n}. Let us define the partitions
into 2-tuples in the following way:
n
P2,0 (i) = {i, 2n + 1 − i},
X
n
P2,0 (i) = 2n + 1, for i = 1, 2, . . . , n.
n
P2,2 (i) = {i, n + i},
X
n
P2,2 (i) = n + 2i, for i = 1, 2, . . . , n.
n
P2,4 (i) = {2i − 1, 2i},
X
n
P2,4 (i) = 4i − 1, for i = 1, 2, . . . , n.

Moreover, for 3 ≤ n ≡ 1 (mod 2)

(
n { n+1 i−1 i−1
2 + 2 ,n + 1 + 2 } for i ≡ 1 (mod 2),
P2,1 (i) = i n+1 i
{2, n + 2 + 2} for i ≡ 0 (mod 2),
X
n n+1
P2,1 (i) = n + + i, for i = 1, 2, . . . , n.
2
Note that we are able to obtain the partitions into 2-tuples P2,0 n (i) and P n (i) as P n (i) ∪
2,2 1,s
n


P1,t (i) ⊕ n , where s, t = ±1. We use this idea to construct the other partitions. More precisely,
n n n


Pk,d (i) = Pl,s (i) ∪ Pm,t (i) ⊕ ln ,

where k = l + m.
n (i) from the partitions P n (i), s = ±1 and P n (i),
For example, we are able to obtain P3,d 1,s 2,t
n exists for d = ±1, ±3, ±5 and if
t = 0, ±2, ±4 and also t = ±1 for n odd. This means that P3,d
n in the following way
n ≡ 1 (mod 2) also for d = 0, ±2. Moreover, we are able to construct P3,9
n
P3,9 (i) = {3(i − 1) + 1, 3(i − 1) + 2, 3(i − 1) + 3},
X
n
P3,9 (i) = 9i − 3, for i = 1, 2, . . . , n.

93
 n exists for d = ±1, ±3, ±5, ±9. Note that if n ≡ 1 (mod 2) then also the differences
Thus P3,d
d = 0, ±2 are realizable.
Summarizing the previous fact we get the following theorem.
♦ Theorem 5.2.1. [34] Let n, k, d and i be positive integers. There exists a partition Pk,d n of the

set {1, 2, . . . , kn} into n, n ≥ 2, k-tuples such that the difference between the sum of the numbers
in the (i + 1)th k-tuple and the sum of the numbers in the ith k-tuple is d, i = 1, 2, . . . , n − 1 for

d = k2

or
d = s + t,
n and P n , k = l + m.
where s and t are realizable differences in partitions Pl,s m,t

Moreover, the corresponding ith k-tuple in the partition Pk,d

n can be obtained such that

n
Pk,k 2 (i) = {k(i − 1) + 1, k(i − 1) + 2, . . . , k(i − 1) + k}

or
n n n


Pk,d (i) = Pl,s (i) ∪ Pm,t (i) ⊕ ln ,
where k = l + m, respectively.

n has the property that

Let us note that each of the defined partition Pk,d
X
n n
Pk,d (i) = Ck,d + di,
n is a constant depending on the parameters k and d. Table 5.1 gives the values of
where Ck,d
n for k ≤ 7.
feasible differences for partition Pk,d
It indicates that, for a given k, the number of feasible
values of d is quite big. However,
2 + 1 , ± (k − 1)2 − 1 , . . . , ±1 for k odd


for k ≥ 6 not every number from the set ± (k − 1)
(or ± (k − 1)2 + 1 , ± (k − 1)2 − 1 , . . . , 0 for k even) can be realizable as a difference d in the



n . However, it is a simple observation that for k ≥ 6 all numbers from the set
partition Pk,d

±1, ± 3, . . . , ±(k + 14) for k odd

0, ± 2, . . . , ±(k + 14) for k even (5.1)
n .
are feasible as a difference d in the partition Pk,d

5.3 Counting the Upper Bound of the Difference d

The next theorem gives the upper bound of the difference d if the graph GH is super (a, d)-
H-antimagic.
♦ Theorem 5.3.1. [34] Let G be a (pG , qG )-graph and let H be a connected (pH , qH )-graph.
If GH admits a super (a, d)-H-antimagic labeling and number of subgraphs isomorphic to H in

94
 k d
for every n
moreover for n odd
1 ±1
2 0, ±2, ±4
±1
3 ±1, ±3, ±5, ±9
0, ±2
4 0, ±2, ±4, ±6, ±8, ±10, ±16
±1, ±3, ±5
5 ±1, ±3, ±5, ±7, ±9, ±11, ±13, ±15, ±17, ±25
0, ±2, ±4, ±6, ±8, ±10
6 0, ±2, ±4, ±6, ±8, ±10, ±12, ±14, ±16, ±18, ±20, ±24, ±26, ±36
±1, ±3, ±5, ±7, ±9, ±11, ±15, ±17
7 ±1, ±3, ±5, ±7, ±9, ±11, ±13, ±15, ±17, ±19, ±21, ±23, ±25, ±27, ±29, ±35, ±37, ±49
0, ±2, ±4, ±6, ±8, ±10, ±12, ±14, ±16, ±18, ±24, ±26

n , k ≤ 7.
Table 5.1: The feasible differences d for partition Pk,d

GH is qG then

pH (pG − 2)
d ≤ p2H + qH
2
− 2pH + .
qG − 1

Proof. Let G be an arbitrary (pG , qG )-graph and let H be a connected (pH , qH )-graph. Let GH
contains exactly qG subgraphs isomorphic to H. Let GH admits a super (a, d)-H-antimagic
labeling f ,

f : V (GH ) ∪ E(GH ) → {1, 2, . . . , p + q},

where p = |V (GH )| = (pH − 2)qG + pG and q = |E(GH )| = qH qG .

The smallest possible weight of a subgraph isomorphic to H can be obtained when the
smallest possible numbers are used to label its elements. It means, when the numbers 1, 2, . . . , pH
are used as the vertex labels and the numbers p + 1, p + 2, . . . , p + qH are used as the edge labels.
Thus

a ≥1 + 2 + · · · + pH + (p + 1) + (p + 2) + · · · + (p + qH )
(pH + 1)pH (qH + 1)qH
= + pqH + . (5.2)
2 2
The largest possible weight of a subgraph isomorphic to H can be realizable if the largest possible

95
numbers are used to label vertices as well the edges of this subgraph. Thus

a + (qG − 1)d ≤p + (p − 1) + · · · + (p − pH + 1)
+ (p + q) + (p + q − 1) + · · · + (p + q − qH + 1)
(2p − pH + 1)pH (2p + 2q − qH + 1)qH
= + . (5.3)
2 2
Combining Inequalities (5.2) and (5.3) and after some mathematical manipulations we get the
upper bound for the difference d in the following form.
pH (pG − 2)
d ≤ p2H + qH
2
− 2pH + .
qG − 1

2 −p .
If G is a tree, that is, pG = qG + 1, then from Theorem 5.3.1 it follows that d ≤ p2H + qH H

♦ Corollary 5.3.2. [34] Let G be a tree of order pG and let H be a connected (pH , qH )-graph.
If GH admits a super (a, d)-H-antimagic labeling then

d ≤ p2H + qH
2
− pH .

Carlson [32] defines an amalgamation of graphs as follows. Let G1 , G2 , . . . , Gk be a finite

collection of graphs and let each Gi have a fixed vertex vi called the terminal. The amalgamation
Amal{Gi , vi } is formed by taking all the Gi ’s and identifying their terminals. By amal(H, k) we
denote a graph, where the amalgamation is constructed from k copies of connected graph H.
H is isomorphic to the
If the graph G is isomorphic to a star K1,n , n ≥ 2, then the graph K1,n
amalgamation amal(H, n). Using Corollary 5.3.2 we immediately obtain the following result.
♦ Corollary 5.3.3. [34] Let H be a connected (pH , qH )-graph. If the amalgamation amal(H, n)
admits a super (a, d)-H-antimagic labeling and number of subgraphs isomorphic to H in amal(H, n)
is n then
d ≤ p2H + qH
2
− pH .

A shackle of G1 , G2 , . . . , Gk , denoted by shack(G1 , G2 , . . . , Gk ), is a graph constructed from

non-trivial connected graphs G1 , G2 , . . . , Gk such that for every 1 ≤ i, j ≤ k with |i − j| ≥ 2,
Gi and Gj have no common vertex, and for every 1 ≤ i ≤ k − 1, Gi and Gi+1 share exactly
one common vertex, called a linkage vertex, where the k − 1 linkage vertices are all distinct. In
the case when all Gi ’s are isomorphic to a connected graph H, we call the resulting graph as a
shackle of H denoted by shack(H, k).
If the graph G is isomorphic to a path Pn , n ≥ 2, then the graph PnH is isomorphic to the
shackle shack(H, n − 1) and by Corollary 5.3.2 the upper bound for the difference d is as follows.
♦ Corollary 5.3.4. [34] Let H be a connected (pH , qH )-graph. If the shackle shack(H, n −
1) admits a super (a, d)-H-antimagic labeling and number of subgraphs isomorphic to H in
shack(H, n − 1) is n − 1 then
d ≤ p2H + qH
2
− pH .

96
 Note that this upper bound was proved by Lemma 6 in [61].
On the other hand if the graph H is isomorphic to K2 then from Theorem 5.3.1 it follows.
♦ Corollary 5.3.5. [34] If H is isomorphic to K2 and GH admits a super (a, d)-EAT labeling
then
2pG − 4
d≤1+ .
qG − 1

This upper bound for the difference d was proved in [22].

5.4 Main Result

In this section we show connection between H-antimagic labelings and edge-antimagic total
labelings. We describe a construction how to obtain the H-antimagic graph from a smaller
edge-antimagic total graph G. Note that if H ∼
= K2 then GH ∼= G and the result trivially holds.
The following theorem gives the main result.
♦ Theorem 5.4.1. [34] Let G be a (b, d∗ )-EAT graph and H be a connected graph of order
at least 3. If GH contains exactly qG subgraphs isomorphic to H then GH is super (a, d)-H-
antimagic and d = d∗ +dv +de , where dv and de are feasible values of differences in the partitions
PpqH
G qG
−3,dv and PqH ,de , respectively.

Proof. Let g be a (b, d∗ )-EAT labeling of G. The set of all edge-weights of the edges of G under
the labeling g is
{wtg (e) : e ∈ E(G)} = {b, b + d∗ , . . . , b + (qG − 1)d∗ }.
Denote the edges of G by the symbols e1 , e2 , . . . , eqG such that

wtg (ei ) = b + (i − 1)d∗ ,

where i = 1, 2, . . . , qG .
Let H be a connected (pH , qH )-graph, pH ≥ 3.
Let GH contains exactly qG subgraphs isomorphic to H, say H1 , H2 , . . . , HqG , where the
subgraph Hi replaces the edge ei in G, i = 1, 2, . . . , qG .
Construct a total labeling f , f : V (GH ) ∪ E(GH ) → {1, 2, . . . , qG (pH + qH − 2) + pG } in the
following way:

• f (v) = g(v), if there exist integers t, s, 1 ≤ t < s ≤ qG such that v ∈ V (Ht ) ∩ V (Hs ).
• As pH ≥ 3 then there exists a vertex x, x ∈ V (Hi ) and x 6= v. Then for i = 1, 2, . . . , qG let

f (x) = g(ei ).

• For i = 1, 2, . . . , qG let

{f (y) : y ∈ V (Hi ), y 6= v and y 6= x} = PpqH

G
−3,dv (i) ⊕ (pG + qG ).

97
 • For i = 1, 2, . . . , qG let

{f (e) : e ∈ E(Hi )} = PqqHG ,de (i) ⊕ (pH − 2)qG + pG ,

where dv depends on pH and de depends on qH .

It is not difficult to check that the vertices are labeled with the smallest possible numbers
1, 2, . . . , (pH − 2)qG + pG .
Moreover, for the weight of the subgraph Hi , i = 1, 2, . . . , qG , we obtain
X X
wtf (Hi ) = f (u) + f (e)
u∈V (Hi ) e∈E(Hi )
X X X
= f (v) + f (x) + f (u) + f (e)
v∼ei u∈V (Hi )\{v,x} e∈E(Hi )
ei ∈E(G)
X X q 
= g(v) + g(ei ) + PpH −3,dv (i) ⊕ (pG + qG )
G

v∼ei
ei ∈E(G)
X q


+ PqH ,de (i) ⊕ (pH − 2)qG + pG
G

   
∗ qG
= b + (i − 1)d + CpH −3,dv + dv i + (pH − 3)(pG + qG ))
 
qG

+ CqH ,de + de i + qH (pH − 2)qG + pG

=CpqH
G qG ∗
−3,dv + CqH ,de + b − d + (pH − 3)(pG + qG )
+ qH (pH − 2)qG + pG + (d∗ + dv + de )i.

This concludes the proof.

The largest feasible value of the difference d for a super (a, d)-H-antimagic labeling of GH
is given by the following corollary.
♦ Corollary 5.4.2. [34] Let G be a (super) (b, d∗ )-EAT graph and H be a connected graph
of order at least 3. If GH contains exactly qG subgraphs isomorphic to H then GH is super
(a, d∗ + (pH − 3)2 + qH
2 )-H-antimagic graph.

Proof. From Theorem 5.2.1 it follows that the largest possible value of the difference in the
partition PpqHG 2
−3,dv is (pH − 3) and the largest possible value of the difference in the partition
qG 2 . According to Theorem 5.4.1 the result follows.
PqH ,de is qH

Next corollary gives the formula for other feasible differences of d as a function of pH and
qH .
♦ Corollary 5.4.3. [34] Let G be a (super) (b, d∗ )-EAT graph and H be a connected graph
of order at least 3. If GH contains exactly qG subgraphs isomorphic to H then GH is super

98
(a, d)-H-antimagic, where

d = d∗ + (pH − 3 − t)2 + (qH − s)2 ± t ± s

for every t = 0, 1, . . . , pH − 3 and s = 0, 1, . . . , qH .

5.5 Special Families of Graphs

In this section we consider two special families of graphs, namely amalgamation of graphs
and shackle of graphs.
If the graph G ∼ = K1,n , n ≥ 2, then the graph K1,n H is known as amalgamation of H.
H
According to Corollary 5.3.3, if K1,n admits a super (a, d)-H-antimagic labeling and number of
subgraphs isomorphic to H in K1,nH is n then d ≤ p2 + q 2 − p .
H H H

In [114] authors proved the following result.

Theorem 5.5.1. [114] The star K1,n , n ≥ 2, admits a super (a, d)-EAT labeling for d = 0, 1, 2.

Then we have the following theorem, in [34].

♦ Theorem 5.5.2. [34] Let H be a connected (pH , qH )-graph, pH ≥ 9 and let n be an integer,
H contains exactly n subgraphs isomorphic to H then K H admits a super (a, d)-
n ≥ 2. If K1,n 1,n
H-antimagic labeling for
0 ≤ d ≤ pH + qH + 27.

Proof. It follows from Theorem 5.4.1, Theorem 5.5.1 and Expression (5.1) for partition of num-
bers.

Note that Theorem 5.4.1 gives much more feasible values of the difference d for super (a, d)-
H . Furthermore there exist several feasible differences d which is not
H-antimagic labeling of K1,n
possible to obtain from the proof of Theorem 5.4.1. For these values of difference d we propose
the following.

Open Problem 1. Determine for which values of differences d, 0 ≤ d ≤ p2H + qH 2 − p , not

H
H .
covered by Theorem 5.4.1, there exists a super (a, d)-H-antimagic labeling of K1,n

As we mentioned before, if the graph G ∼ = Pn , n ≥ 2, then the graph PnH is known as

shackle of H. According to Corollary 5.3.4, if PnH admits a super (a, d)-H-antimagic labeling
and number of subgraphs isomorphic to H in PnH is n − 1 then d ≤ p2H + qH 2 −p .
H

For edge-antimagicness of paths in [23] is proved the following.

Theorem 5.5.3. [23] The path Pn , n ≥ 2, admits a super (a, d)-EAT labeling if and only if
d = 0, 1, 2, 3.

Then we get.

99
♦ Theorem 5.5.4. Let H be a connected (pH , qH )-graph, pH ≥ 9 and let n be an integer,
n ≥ 3. If PnH contains exactly n − 1 subgraphs isomorphic to H then PnH admits a super
(a, d)-H-antimagic labeling for
0 ≤ d ≤ pH + qH + 28.

Proof. Using Theorem 5.4.1, Theorem 5.5.3 and Expression (5.1) for partition of numbers we
immediately obtain that 0 ≤ d ≤ pH + qH + 28.

By the same way as for amalgamation we can formulate analogous open problem for shackle
of H.

Open Problem 2. Determine for which values of differences d, 0 ≤ d ≤ p2H + qH 2 − p , not

H
H
covered by Theorem 5.4.1, there exists a super (a, d)-H-antimagic labeling of Pn .

Inayah et al., [61] studied the existence of H-antimagic labeling of shackle of H by using a
different method. Their different approach gives different sets of differences obtained by desired
constructions.
A block of a graph is a maximal subgraph with no cut-vertex. Ngurah et al., [91] defined a
blockcut-vertex graph of a graph G as a graph H where vertices of H are blocks and cut-vertices
in G and two vertices are adjacent in H if and only if one vertex is a block in G and the other is
a cut-vertex in G belonging to the block. Barrientos [28] defined a chain graph as a graph with
blocks B1 , B2 , . . . , Bk such that for every i, i = 1, 2, . . . , k − 1, the blocks Bi and Bi+1 have a
common vertex in such a way that the blockcut-vertex graph is a path.
In [91] Ngurah et al., investigated H-supermagicness of chain graphs consisting of k blocks
where each block is identical and isomorphic to a given cycle. Inspired by their work we decided
to deal with the (super) H-antimagicness of chain graphs in general. Moreover, we also extended
the results for more general cases – the graphs forming the chain need not necessarily to be blocks.
Let H be a connected graph of order at least 2. We denote by kH -path a chain graph with
k graphs H1 , H2 , . . . , Hk where each graph is identical and isomorphic to the given graph H.
Note, that H need not to be a block. Moreover, when H is not isomorphic to K2 then kH -path
s not defined uniquely. There exists many nonisomorphic kH -paths for a given graph H and a
given k.

♦ Theorem 5.5.5. Let G be a kH -path, k ≥ 2 containing exactly k subgraphs isomorphic to H.

Then G is a super (a, d)-H-antimagic graph for d ∈ {0, 1, 2, 3}.

Proof. Let H be a connected graph of order p ≥ 2 and size q. Let G be a kH -path, k ≥ 2,

containing exactly k subgraphs isomorphic to H, say H1 , H2 , . . . , Hk .
Let x2 , x3 , . . . , xk be the vertices of G such that xi ∈ V (Hi−1 )∩V (Hi ) for every i = 2, 3, . . . , k.
Let x1 be a vertex of H1 such that x1 6= x2 and let xk+1 be a vertex of Hk such that xk+1 6= xk .
Let ei , i = 1, 2, . . . , k be an edge of Hi such that ei is adjacent to xi .
Bača et al., [23] proved the path Pk , k ≥ 2, has a super (a, d∗ )-edge-antimagic total labeling
if and only if d∗ ∈ {0, 1, 2, 3}. By fd∗ (Pk ) we denote the super (a, d∗ )-edge-antimagic total

100
labeling of a path Pk . For purposes of this proof we denote the elements of a path Pk such that
Pk = x1 e1 x2 e2 x3 . . . xk−1 ek−1 xk .
Now we define a labeling g of G in the following way.

1. First label the vertices xi , i = 1, 2, . . . , k + 1 of G such that

g(xi ) = fd∗ (xi ).

2. Then label the edges ei , i = 1, 2, . . . , k of G such that

g(ei ) = fd∗ (ei ) + k(p + q − 3).

As fd∗ is an (a, d)-edge-antimagic total labeling then the partial sums wt of the subgraphs
H1 , H2 , . . . , Hk are at this moment

b, b + d∗ , . . . , b + (k − 1)d∗ ,

where b = a + k(p + q − 3). Now we rename the subgraphs Hi , i = 1, 2, . . . , k by the

symbols Hi , i = 1, 2, . . . , k such that the partial sums are now such that

wt(Hi ) = b + (i − 1)d∗ . (5.4)

3. Note that at this moment in every subgraph Hi , i = 1, 2, . . . , k, exactly p + q − 3 are not

labeled. Let U (Hi ) denote the set of all unlabeled vertices and edges in Hi . We label these
elements such that
n l mo
g(U (Hi )) = i + jk : j = 1, 2, . . . , p+q−3
2
n l m l m o
∪ mk + 1 − i : m = p+q−3 2 + 2, p+q−3
2 + 3, . . . , p + q − 3 ,

moreover, use the smallest numbers from this set to label the unlabeled vertices of Hi .

It is easy to see that the labeling g uses the numbers 1, 2, . . . , k(p − 1) + 1 to label the vertices
and the numbers k(p − 1) + 2, k(p − 1) + 3, . . . , k(p + q − 1) + 1 to label edges of G. Moreover,
every number is used once as a label.
Now let us consider the weights of the subgraphs Hi , i = 1, 2, . . . , k, under the labeling g.
First we count the partial sums of Hi when only the elements of U (Hi ) are considered. It is
easy to get that
( k(p+q−3)(p+q−1)
2 if p + q is odd,
wt(U (Hi )) = k((p+q−3) 2 +2(p+q−3)−1 (5.5)
2 +i if p + q is even.

It is easy to get, see (5.4) and (5.5), that the labeling g is super (a, d)-H-antimagic for d ∈
{0, 1, 2, 3}. Moreover, when p + q is even then also d = 4 is feasible.

If a graph H does not contain an articulation we immediately obtain the following corollary.

101
♦ Corollary 5.5.6. [34] If H is a block then kH -path, k ≥ 2, is a super (a, d)-H-antimagic
graph for d ∈ {0, 1, 2, 3}.

Note that as the difference d = 0 is feasible, Corollary 5.5.6 includes the result of Ngurah et
al., [91] about H-supermagicness of chain graphs.
In [16] Bača et al., defined the partitions of a set of integers with determined differences. We
would like to use this concept for construction of H-antimagic labelings of chain graphs also for
another differences. Our goal is also to find an upper bound for feasible values of the difference
d as a function of order and a size of the graph H.

5.6 Conclusion

In this chapter, we examined the existence of super (a, d)-H-antimagic labeling for graph
operation GH , where G is a (b, d∗ )-edge-antimagic total graph and H is a connected graph
of order at least 3. We have found super (a, d)-H-antimagic labeling for all differences d =
d∗ + dv + de , where d∗ is the feasible value of difference in super edge-antimagic graph G and dv
(respectively, de ) are feasible values of differences in the partitions PpqH
G qG
−3,dv (respectively, PqH ,de ).
Additionally, we showed that for a connected (pH , qH )-graph H the graph K1,n H (respectively,

PnH ) admits a super (a, d)-H-antimagic labeling for every difference 0 ≤ d ≤ pH + qH + 27

(respectively, 0 ≤ d ≤ pH + qH + 28).

102
Chapter 6

Applications of Graph Labelings

The field of graph theory has had much influence in various fields because of its applications.
One of the important areas in graph theory is graph labeling whose influence extends to such
applications in the areas as coding theory, X-ray crystallography, radar systems, astronomy,
circuit design, social networking, network security, communication network addressing, solution
of linear congruence systems and database management.
In this chapter we will briefly make references to Bloom and Golomb’s paper [31] on graph
labeling application and follow with some new relevance.

6.1 Application of Graph Labeling

Graph labelings have interesting applications in variety of fields, [31, 43, 96].
Bloom and Golomb explained how graph labeling is applicable to coding theory, ambiguity
in X-ray crystallography, communication network labeling, circuit layouts and finite additive
number theory and ruler problems, [31].
While dealing with X-ray crystallography some times more than one crystal structure has
the same diffraction information. Bloom and Golomb, [31] showed that this is equivalent to
determining all labelings of appropriate graphs which produce a pre-specified set of edge labels.
Similarly in coding theory, the design of certain important classes of good non periodic codes for
pulse radar and missile guidance is equivalent to labeling the complete graph in such a way that
all the edge labels are distinct. The vertex labels then determine the time positions at which
pulses are transmitted.
Graceful labeling plays a significant role in communication networks. A communication
network is composed devices represented as vertices. Every vertex has computing power and can
exchange messages over communication links (edges). It is useful for identification to assign each
user terminal (vertex) a vertex label, subject to the constraint that all connecting communication
links receive distinct labels. In this way, the numbers of any two communicating terminals
automatically specify (by simple subtraction) the edge label of the connecting edge.

103
6.1.1 Linear Congruence

An equation of the form ax ≡ b (mod m) where 0 ≤ x ≤ m holds true for some x is said to
be linear congruence equation. In this section we will address the problem of solving a special
type of linear congruence equation.
Given n, k ∈ N and a1 , a2 , . . . , ak ∈ Zn , it is known classically [Uspensky and Heaslet 1939;
Vandiver 1924], [120, 121] that the linear congruence a1 x1 + a2 x2 + · · · + ak xk = 1(in Zn )
has a solution if and only if gcd(a1 , a2 , . . . , ak ) ∈ ZX
n , the group of units of Zn . Adams and
Ponomarenko, [2] asked when such a solution exists with distinct xi ∈ Zn , a question that
appears to have been overlooked in the literature. In general, some additional conditions are
necessary; for example, 1x1 + 1x2 + 1x3 = 1 does not have a solution with distinct xi ∈ Z3 .
Their partial solution has a stronger coefficient condition and another restriction involving
φ(n), the Euler’s totient function. The general case remains open.
Given a composite n, k < n, and gcd(a1 , a2 , . . . , an ) ∈ ZX
n (group of units in Zn ), then there
exist distinct xi ∈ Zn (xi 6= xj , 1 ≤ i, j ≤ k) such that a1 x1 + a2 x2 + · · · + an xn ≡ 1 has a
solution in Zn .
Adams and Ponomarenko [2] proved that if k ≤ φ(n) and ai ∈ ZX n for 1 ≤ i ≤ k, where
Euler’s totient function, φ(n) represents number of positive integers less than or equal to n which
are relatively prime to n, then there exists distinct xi ∈ Zn satisfying

a1 x1 + a2 x2 + · · · + an xn ≡ 1.

In [55], the authors study a special case of finding the solution of linear congruence when

a1 + a2 + · · · + ak = n − 1

using super edge-antimagic labeling of trees.

In [112], Sugeng and Miller proved that every caterpillar has a b-edge consecutive edge magic
graph for every b and super edge magic labeling is a special case of b-edge consecutive edge magic
labeling when b = n.
Consider the sequence of length n + 1 as (a1 , a2 , . . . , ak , 0, 0, . . . , 0) such that
| {z }
n+1−k

a1 + a2 + · · · + ak = n − 1.

Adding 1 to each element of the sequence we get another sequence (a1 + 1, a2 + 1, . . . , ak +

1, 1, 1, . . . , 1). When this sequence is considered as the degree sequence of a graph, the summation
| {z }
n+1−k
satisfies the Handshaking Lemma (every finite undirected graph has an even number of vertices
with odd degree consequently there is even number of odd degree vertices), that is
k
X
(ai + 1) + 1, 1, . . . , 1 = 2n.
| {z }
i=1 n+1−k

104
Although it is quite easy to determine the degree sequence of a given graph, the converse
procedure that is to determine whether a given degree sequence is graphical (that is if there
exists some finite simple graph having this degree sequence) is not straight forward. Authors in
[30], have developed algorithm to determine whether a degree sequence is graphic.
As explained in [55] we construct a caterpillar of order n + 1 and size n with the sequence
(a1 + 1, a2 + 1, . . . , ak + 1, 1, 1, . . . , 1) as follows.
| {z }
n+1−k

Let the vertex set V = A ∪ B, v2i−1 ∈ A and deg(v2i−1 ) = a2i−1 + 1, v2i ∈ B and deg(v2i ) =
a2i+1 when 2i ≤ k, where 1 ≤ i ≤ (k + 1)/2 when k is odd, 1 ≤ i ≤ k/2 when k is even.
Let uji denote the vertices of degree 1 in A and B.
Then edge set E = {vi vi+1 : 1 ≤ i ≤ k} ∪ {vi uji : 1 ≤ i < k, 1 ≤ j ≤ ai } ∪ {vi uji : 1 < i <
k, 1 ≤ j ≤ ai − 1} and |E| = n.
For convenience, we denote plane representation of caterpillar by Ca,b with bipartition A0 , B 0
of its vertex set V (Ca,b ). The a-vertices of A0 are labeled ui , 1 ≤ i ≤ a and the b-vertices of B 0
are labeled wi , 1 ≤ i ≤ b. The super edge-magic labeling of the caterpillar Ca,b is defined by
function f as shown below
Define
f : V (Ca,b ) ∪ E(Ca,b ) → {1, 2, . . . , p + q}
in such a way that

f (ui ) =i i ∈ {1, 2, . . . , a},

f (wj ) =a + j j ∈ 1, 2, . . . , b,
f (ui wj ) =2(a + b) − (i + j − 1) for all i, j, where ui wj is an edge.

Thus,
f (ui wj ) + f (ui ) + f (wj ) = 3a + 2b + 1, ui wj ∈ E(Ca,b ),
the above labeling is super edge-magic labeling.
We know that, X X
f (v)deg(v) + f (e) = qc(f )
v∈V (G) e∈E(G)

Since f is a super edge-magic labeling, we get the following congruences.

a
X b
X 2n+1
X
((deg(uj ) − 1)f (uj ) + ((deg(wi ) − 1)f (wi ) + m = 0 (in Zn ),
j=1 i=1 m=1
a b
X X (2n + 1)(2n + 2)
((deg(uj ) − 1)f (uj ) + ((deg(wi ) − 1)f (wi ) + = 0 (in Zn ),
2
j=1 i=1
a
X b
X
((deg(uj ) − 1)f (uj ) + ((deg(wi ) − 1)f (wi ) + 1 = 0 (in Zn ).
j=1 i=1

105
 Let k be the number of vertices in V (Ca,b ) whose degree is greater than one. For brevity, we
write the above equation as
k
X
(di − 1)f (vi ) + 1 = 0 (in Zn ), di − 1 > 0,
i=1
k
X
(di − 1)f (vi ) = −1 (in Zn ),
i=1
k
X
(di − 1)f (vi ) = n − 1 (in Zn ).
i=1

Therefore, f (vi ) ∈ {1, 2, . . . , n}, 1 ≤ i ≤ k and are distinct in Zn .

Now, ai = di − 1 for 1 ≤ i ≤ k so
k
X
ai f (vi ) = n − 1 (in Zn ).
i=1

Let xi = n − f (vi ), 1 ≤ i ≤ k.
We can observe that xi = n − f (vi ) for (1 ≤ i ≤ k) will be less than 0 when n < f (vi ).
To get a distinct solution, we will have to label one vertex of degree 1 with label (a + b) while
constructing the super edge-magic labeling.
Finally, we will verify that xi = n − f (vi ) is a solution of

a1 x1 + a2 x2 + · · · + ak xk ≡ 1.

So,

a1 x1 + a2 x2 + · · · + ak xk =a1 (n − f (v1 )) + a2 (n − f (v2 )) + · · · + ak (n − f (vk ))

=a1 n + a2 n + · · · + ak n − (a1 f (v1 ) + a2 f (v2 ) + · · · + ak f (vk ))
=0 − (n − 1) = 1 in Zn .

Thus, the solution xi = n − f (vi ) are distinct in Zn .

The process is explained with the help of following example.
Example 1: Find a solution to 3x1 + 2x2 ≡ 1 in Z6 .
Here n = 6 and k = 2. In order to find a solution we need to check that is 2 ≤ φ(n) = φ(6)?
Obviously 2 ≤ φ(6) = 2. So we can find solution of above equation.
Form a (n + 1) degree sequence (3, 2, 0, 0, 0, 0, 0). Now add 1 to all of them, so we have
(4, 3, 1, 1, 1, 1, 1). We can form a caterpillar with this degree sequence since it satisfies the
Handshaking Lemma. We will form a caterpillar with order 7 and size 6 whose degree sequence
is (4, 3, 1, 1, 1, 1, 1) and it has super edge magic labeling. One such possible way is shown in
Figure 6.1.

106
 7 6 5

8 9 10 11 12 13

4 3 2 1

Figure 6.1: Supermagic labeling of caterpillar of order 7 and size 6

Now as explained in [55], to find a solution we will find the non pendant vertices that is with
label 5 and 4 in our case.
Finally the solution will be x1 = 6 − 5 = 1 and x2 = 6 − 4 = 2.
Check:
3 × 1 + 2 × 2 = 7 ≡ 1 in Z6 .

So solution x1 = 1 and x2 = 2 is possible with the help of super edge magic labeling of
graphs.
Example 2: Find a solution to 5x1 + 2x2 + 2x3 ≡ 1 in Z10 .
Here n = 12 and k = 3. In order to find a solution we need to check that is 3 ≤ φ(n) = φ(12)?
Obviously 3 < φ(12) = 4. So we can find solution of above equation.
Form a (n + 1) degree sequence (6, 2, 2, 0, 0, 0, 0, 0, 0, 0, 0). Now add 1 to all of them, so we
have (6, 3, 3, 1, 1, 1, 1, 1, 1, 1, 1). We need to form a caterpillar graph with order 11 and size 10
whose degree sequence is (6, 3, 3, 1, 1, 1, 1, 1, 1, 1, 1) and it posses a super edge magic labeling.
One such possible way is shown in Figure 6.2.

4 5 6 7 8 9 10 11

21 20 19 18 17 16 15 14 13 12

1 2 3

Figure 6.2: Supermagic labeling of caterpillar of order 11 and size 10

Now to find a solution we will find the non pendant vertices that is with label 1, 9 and 3 in
our case.

107
 Finally the solution will be x1 = 10 − 1 = 9, x2 = 10 − 9 = 1, x3 = 10 − 3 = 7.
Check:
5 × 9 + 2 × 1 + 2 × 7 = 61 ≡ 1 in Z10 .

So solution x1 = 9, x2 = 1 and x3 = 7 is possible with the help of super edge magic labeling
of graphs.

6.1.2 Irregular Reflexive Labeling

As mentioned in Section 3.1, irregular reflexive labeling is the return to the original spirit of
irregular labeling, restricting the vertex labels to be even numbers to indicate the presence of
loops and allowing 0 as a possible vertex label, indicating a vertex with no loop.
In this section we will explain how irregular reflexive labeling can be applied to network
analysis.
In real world, there is a variety of network. One side of real world networks is regular
networks (same weights of graph elements) while the other side of networks is totally irregular
(different weights of graph elements). Most of the networks in real world are combinations of
both where some part of network is regular and other is irregular.
The irregular (or irregular reflexive) networks are mostly used in the analysis of networks.
What happens when networks are completely irregular or what happens when they are com-
pletely regular? One can study and analyse networks in this regard. Since most of real world
networks are in between so having identified both extremes - one can use this to estimate what
happens within real world networks.
It will be very difficult to apply irregular reflexive labeling as direct application to some
networks which are very dynamic like social networks. By the time we set up a network, the
network configuration would have changed and it will be hard to accommodate those changes.
Rather it would be easy to form application of irregular reflexive labeling to communication
networks. It could be constructed in a way to make sure that it is an irregular network. This
identification helps to uniquely identifying each terminals and thus making sure that no two
vertices have same weights. Also if the network is not irregular then it could have happened
that there is some fault in network or could be that a link is broken between two vertices.
Some recent applications of graph labeling can be found in secret sharing [109] and relational
databases [115].
Graph labeling presents a common context for many applied and theoretical problems. This
has been illustrated in the previous section, in which we narrated some brief applications in
variety of areas of science. The reward of such efforts is obvious and often immediate. This
principle is apparent in the many examples which are drawn from these labeling techniques.
Thus there could be possible more application of labeling. Readers are invited to explore more
applications.
In the following section we will state some conjectures and open problems.

108
6.2 Conjectures and Open Problems

Graph labeling field has many interesting open problems and conjectures. Some of them are
open for many years and only partial success has been achieved.
Following conjecture by Ringel and Kotzig, also known as Graceful Tree Conjecture has been
the focus of many papers. The motivation for this conjecture was the actual conjecture by Ringel
that complete graph K2n+1 can be decomposed into 2n + 1 subgraphs that are isomorphic to a
given tree of order n.

Conjecture 6.2.1. [99] All trees are graceful.

Among the trees known to be graceful are: caterpillars [99], trees with at most 4 end-vertices
[58], [125] and [69]; trees with diameter at most 5, see [125] and [57]; trees with n(V ) ≤ 35 are
graceful, symmetrical trees [29], [95] and many more, refer [47].
We now describe a fundamental conjecture about harmonious labeling, which has been open
since its origin in 1980.

Conjecture 6.2.2. [51] All trees are harmonious.

Many authors including Graham and Sloane tried to prove this conjecture, see [47]. The
recent success is “All trees with at most 31 vertices are harmonious”. This result is proved by
Fang [41] in 2011 using probabilistic backtracking, tabu search, two-stage constraint solving and
hybrid algorithm.
Hartsfield and Ringel gave the following conjecture.

Conjecture 6.2.3. [54] Every connected graph except K2 is antimagic.

As discussed in Chapter 2, this conjecture is still open for graphs in general. The weaker
version of this conjecture is that every tree except K2 is antimagic.
Kotzig, Rosa [74] and Ringel, Lladó [98] posed the following conjecture about edge magic
total labeling.

Conjecture 6.2.4. [74, 79, 98] Every tree admits edge magic total labeling.

The notion of super edge-magic total labeling was introduced by Enomoto et al., [38]. They
proposed following conjecture.

Conjecture 6.2.5. [38] Every tree admits super edge magic total labeling.

Sugeng et al., [113] posed following conjecture for ladders graphs.

Conjecture 6.2.6. [113] The ladder L2 = Pn 2P2 is super (a, d)-EAT if n is even and d ∈ {0, 2}.

Tanna et al., [118] gave the following conjecture for vertex irregular reflexive labeling for
generalised Petersen graph.

109
Conjecture 6.2.7. [118] Prove that RVS(GP (n, k)) = d(n/2)e + 1.

The following are some of the open problems in graph labeling.

Open Problem 6.2.1. [27] Find (a, d)-EAT labelings for even cycles with d ∈ {4, 5} and for
odd cycle with d = 5.

Open Problem 6.2.2. [27] Find (a, 5)-EAT labelings for paths Pn , for the feasible values of a.

Many researchers would solve existing labeling conjectures or open problems for some graphs
but so far there is no great progress in generalising a set of graphs that adheres a property
of particular labeling since it has certain characteristics so Tanna et al., gave following open
problem. Solving such a categorization would be great achievement in graph labeling and will
lead to further classification of any new labeling scheme.
Further, it could be extended that why a particular graph G will admit labeling scheme x
and will not adhere labeling scheme y since it does not fall in that category.
This categorization of graphs admitting a set of labeling may generate big results in graph
labeling.

Open Problem 6.2.3. [116] Categorise graphs in general that admits a set of labeling.

There is great potential of developing applications of irregular labeling or reflexive irregular

labeling to network analysis or social network analysis.

Open Problem 6.2.4. [116] Develop an application of irregular labeling in Social Networks.

Many times a graphs is very close to attend a particular labeling but only a few or one
element does not match the labeling pattern. For example, when trying to prove that the graph
is magic, it might happen that only one element of graph does not adhere the magic constant
but it has some different magic constant that is bimagic labeling. So this might be of interest
to know the type of graphs that misses the magicness by just one.

Open Problem 6.2.5. [82] Develop a classification of graphs admitting a bimagic labeling with
two constants k1 and k2 but one of them being used only once.

Similar open problem could be developed for irregular reflexive labeling.

Open Problem 6.2.6. [118] Categorize graphs admitting irregular reflexive labeling with strength
s but using strength s only once.

6.2.1 Future Work

Irregular reflexive labeling is a recent concept and only a little work has been done in this
field. So there a vast development scope in this area. In particular, once we have labeled graph G
using irregular reflexive labeling, a big question to be address here “Is such a labeling unique?”.
If not, how many ways a graph G could be labeled differently.

110
 Further, if there is more than one way of labeling a graph G, which of these labeling patterns
is the best? For example G has rvs(G) = s, once s is achieved it really does not matter how
many times you have used s in the labeling of G.
One possible answer could be usage of s optimal times, that is using in such a way that the
total weight is as least as possible. If someone is using labeling to write up a code or algorithm,
it occupies the least store space. Thus least time to compilation and run time. Consider the
case when weights represent cost of vertices in a network then the lower the cost, the better the
efficiency. So we must use the maximum strength in such a way that the cost is minimized.
Another future work could be development of an algorithm, where the input is one vertex
configuration at a time and its incident edges. Next input will have another one vertex and
again incident edges. Every stage one should make sure that the graph is irregular reflexive.
This could be very difficult since there is no way to predict the next input or its degree. Or it
could be impossible but either way, it will be a great work.
One more dimension to future work would be distance reflexive labeling. This concept will
be similar to those of distance magic labeling, [10, 44, 45, 72].
Miller et al., in [89], introduced an new labeling labeling technique based on neighbourhood
of a vertex. They defined function f : V (G) → {1, 2, . . . , v} to be 1-vertex magic labeling of a
graph with v vertices as a bijective function with the property that there is a constant k such
that for any vertex x, the sum of labels of all vertices in the neighbourhood of x is a k, that is,
X
f (y) = k,
y∈N (x)

where N (x) is the set of all vertices adjacent to x.

One could develop existence of distance reflexive labeling for regular graphs.

6.3 Conclusion

In this chapter, we discussed various application of graph labeling and posed some conjectures
and open problems. We have also discussed some possibilities of future work.

111
112
Bibliography

[1] Adamaszek, A., Allen, P., Grosu, C., Hladký, J., Almost all trees are almost graceful,
arXiv:1608.01577, (2016), 1–40.

[2] Adams, D., Ponomarenko, V., Distinct solution to a linear congruence, Involve - J. Math.,
3(3), (2010), 341–344.

[3] Ahmad, A., Al-Mushayt, O.B.S., Bača, M., On edge irregularity strength of graphs, Appl.
Math. Comput., 243, (2014), 607–610.

[4] Ahmed, A., Awan, K., Javid, I., Slamin, Total vertex irregularity strength of wheel related
graphs, Australas. J. Combin., 51, (2011), 147–156.

[5] Aldred, R., McKay, B., Graceful and harmonious labelings of trees, Bull. Inst. Combin.
Appl., 23, (1998), 69–72.

[6] Alon, N., Combinatorial nullstellensatz, Combi. Probab. Comput., 8, (1999), 7–29.

[7] Alon, N., Kaplan, G., Lev, A., Roditty, Y., Yuster, R., Dense graphs are antimagic, J. Graph
Theory, 47, (2004), 297–309.

[8] Amar, D., Togni, O., Irregularity strength of trees, Discrete Math., 190, (1998), 15–38.

[9] Anholcer M., Palmer, C., Irregular labelings of circulant graphs, Discrete Math., 312(23),
(2012), 3461–3466.

[10] , Arumugam, S., Kamatchi, N., Kovář, P., Distance Magic Graphs, Utilitas Mathematica,
99, (2016).

[11] Ayel, J., Favaron, O., Helms are graceful, Progress in Graph Theory, Academic Press,
(1984), 89–92.

[12] Babujee, J.B., Bimagic labeling in path graphs, The Mathematics Education, 38, (2004),
12–16.

[13] Babujee, J.B., On edge bimagic labeling, J. Combin. Inf. Syst. Sci., 28, (2004), 239–244.

[14] Bača, M., On certain properties of magic graphs, Utilitas Math., 37, (1990), 259–264.

[15] Bača, M., Bertault, F., MacDougall, J.A., Miller, M., Vertex-antimagic total labelings of
graphs, Discuss. Math. Graph Theory, 23, (2003), 67–83.

113
[16] Bača, M., Brankovic, L., Lascsáková, M., Phanalasy, O., Semaničová-Feňovčı́ková, A., On
d-antimagic labelings of plane graphs, Electron. J. Graph Theory Appl., 1(1), (2013), 28–39.

[17] Bača, M., Brankovic, L., Semaničová-Feňovčı́ková, A., Labelings of plane graphs containing
Hamilton path, Acta Math. Sinica - English Series, 27(4), (2011), 701–714.

[18] Bača, M., Irfan, M., Ryan, J., Semaničová-Feňovčı́ková, A., Tanna, D., Note on Reflexive
Edge Irregular Labelings of Graphs, AKCE International Journal of Graphs and Combina-
torics, 2017.

[19] Bača, M., Irfan, M., Ryan, J., Semaničová-Feňovčı́ková, A., Tanna, D., On reflexive edge
irregular labelings for the generalized friendship graphs, Turkish Journal of Mathematics,
2017.

[20] Bača, M., Jendrol’, S., Miller M., Ryan, J., Total irregular labelings, Discrete Math., 307,
(2007), 1378–1388.

[21] Bača, M., Kimáková, Z., Semaničová-Feňovčı́ková, A., Umar, M.A., Tree-antimagicness of
disconnected graphs, Mathematical Problems in Engineering, Article ID 504251, 2015 (2015),
4 pages, doi:10.1155/2015/504251.

[22] Bača, M., Lin, Y., Miller, M., Simanjuntak, R., New constructions of magic and antimagic
graph labelings, Utilitas Math., 60, (2001), 229–239.

[23] Bača, M., Lin, Y., Muntaner-Batle, F.A., Super edge-antimagic labelings of the path-like
trees, Utilitas Math., 73, (2007), 117–128.

[24] Bača, M., Miller, M., Super Edge-antimagic Graphs: A Wealth of Problems and Some
Solutions, Brown Walker Press., Boca Raton, Florida, 2008.

[25] Bača, M., Miller, M., Phanalasy, O., Ryan, J., Semaničová-Feňovčı́ková, A., Sillasen, A.A.,
Totally antimagic total graphs, Australasian J. Combin., 61(1), (2015), 42–56.

[26] Bača, M., Miller, M., Phanalasy, O., Semaničová-Feňovčı́ková, A., Super d-antimagic la-
belings of disconnected plane graphs, Acta Math. Sinica - English Series, 26(12), (2010),
2283–2294.

[27] Bača, M., Youssef, M. Z., Further results on antimagic graph labeling, Australas. J. Com-
bin., 38, (2007), 163–172.

[28] Barrientos, C., Graceful labelings of chain and corona graphs, Bull. Inst. Combin. Appl.,
34, (2002), 17–26.

[29] Bermond J., Sotteau, D., Graph decompositions and G-design, Proc. 5th British Combin.
Conf., 15, (1976), 53–72.

[30] Biswas, P., Paul, A., Bhattacharya, P., Analyzing the realization of degree sequence by
constructing orthogonally diagonalizable adjacency matrix, Procedia Computer Science, 57,
(2015), 885–889.

114
[31] Bloom, G., Golomb, S., Applications of numbered undirected graphs, Proc. of the IEEE,
65(4), (1977), 562–570.

[32] Carlson, K., Generalized books and Cm -snakes are prime graphs, Ars Combin., 80, (2006),
215–221.

[33] Chartrand, G., Jacobson, M., Lehel, J., Oellermann, O., Ruiz, S., Saba, F., Irregular
networks, Cong. Numer., 64, (1988), 187–192.

[34] Dafik, Slamin, Tanna, D., Semaničová-Feňovčı́ková, A., Bača, M., Constructions of H-
antimagic graphs using smaller edge-antimagic graphs, ARS Combinatoria, 2017.

[35] Delorme, C., Maheo, M., Thuillier, H., Koh K., Teo, H., Cycles with a chord are graceful,
J. Graph Theory, 4, (1980), 409–415.

[36] Dimitz, J., Garnick, D., Gyárfás, A., On the irregularity strength of the m × n grid, J.
Graph Theory, 16, (1992), 355–374.

[37] Doob, M., Characterizations of regular magic graphs, J. Combin. Theory, Ser. B, 25,
(1978), 94–104.

[38] Enomoto, H., Lladó, A. S., Nakamigawa, T., Ringle, G., Super edge-magic graphs, SUT J.
Math., 34 (1998), 105–109.

[39] Exoo, G., Ling, A., McSorley, J., Phillips, N., Wallis, W., Totally magic graphs, Discrete
Math., 254, (2002), 103–113.

[40] Fang, W., A computational approach to the graceful tree conjecture, arXiv:1003.3045,
(2010), 1–7.

[41] Fang, W., New computational results on harmonious trees, arXiv:1106.3490, (2012), 1–5.

[42] Faudree R., Lehel, J., Bounds on the irregularity strength of regular graphs, Combinatorics.
Collog. Math. Soc. János Bobyai, 52, (1987), 247–256.

[43] Franklin, J. N., Ambiguities in the X-ray analysis of crystal structures, Acta Cryst., A30,
(1974), 698–702.

[44] , Fronček, D., Kovář, P., Kovářová, T., Constructing distance magic graphs from regular
graphs, J. Combin. Math. Combin. Comput., 78, (2011), 349 – 354.

[45] , Fronček, D., Kovář, P., Kovářová, T., A note on 4-regular distance magic graphs, Aus-
tralasian Journal of Combinatorics, (2012), 127 – 132.

[46] Frucht, R., Graceful numbering of wheels and related graphs, Ann. N. Y. Acad. Sci., 319,
(1979), 219–229.

[47] Gallian, J.A., A dynamic survey of graph labeling, Electron. J. Combin., 19, (2016), #DS6.

[48] Gallian, J.A., Prout, J., Winters, S., Graceful and harmonious labelings of prisms and
related graphs, Ars Combin., 34, (1992), 213–222.

115
[49] Gnanajothi, R., Topics in graph theory, Ph.D. Thesis, Madurai Kamaraj University, India,
(1991).

[50] Golomb, S., How to number a graph, Graph Theory and Computing, (ed. R.C. Read),
Academic Press, New York, (1972), 23–37.

[51] Graham R., Sloane, N., On additive bases and harmonious graphs, SIAM J. Alg. Discrete
Math., 1, (1980), 382–404.

[52] Gutiérrez A., Lladó, A., Magic coverings, J. Combin. Math. Combin. Comput., 55, (2005),
43–56.

[53] Gyárfás, A., The irregularity strength of Km,m is 4 for odd m, Discrete Math., 71, (1998),
273–274.

[54] Hartsfield, N., Ringel, G., Pearls in graph theory: A comprehensive introduction, Academic
Press, Boston - San Diego - New York - London, 1990.

[55] Hegde, S.M., Murthy, T.S, A partial solution to linear congruence conjecture, National
Academy Science Letters, 39, (2016), 451–453.

[56] Hőede, C., Kuiper, H., All wheels are graceful, Utilitas Math., 14, (1987), 311.

[57] Hrnčiar, P., Haviar, A., All trees of diameter five are graceful, Discrete Math., 233, (2001),
133–150.

[58] Huang, C., Kotzig A., Rosa, A., Further results on tree labelings, Utilitas Math., 21c,
(1982), 85–87.

[59] Inayah, N., Lladó, A., Moragas, J., Magic and antimagic H-decompositions, Discrete Math.,
312, (2012), 1367–1371.

[60] Inayah, N., Salman, A.N.M., Simanjuntak, R., On (a, d)-H-antimagic coverings of graphs,
J. Combin. Math. Combin. Comput., 71, (2009), 273–281.

[61] Inayah, N., Simanjuntak, R., Salman, A.N.M., Syuhada, K.I.A., On (a, d)-H-antimagic
total labelings for shackles of a connected graph H, Australasian J. Combin., 57, (2013),
127–138.

[62] Ivančo, J., On supermagic regular graphs, Math. Bohemica, 125, (2000), 99–114.

[63] Ivančo, J., Jendrol’, S., Total edge irregularity strength of trees, Discuss. Math. Graph
Theory, 26, (2006), 449–456.

[64] Jendrol’, S., Miškuf, J., Soták, R., Total edge irregularity strength of complete graphs and
complete bipartite graphs, Discrete Math., 310, (2010), 400–407.

[65] Jendrol’, S., Tkáč, M., The irregularity strength of tKp , Discrete Math., 145, (1995), 301–
305.

[66] Jeurissen, R.H., Magic graphs, a characterization, Europ. J. Combin., 9, (1988), 363–368.

116
[67] Jeyanthi, P., Selvagopal, P., More classes of H-supermagic graphs, Internat. J. Algor.
Comput. Math., 3, (2010), 93–108.

[68] Jezný, S., Trenkler, M., Characterization of magic graphs, Czechoslovak Math. J., 33,
(1983), 435–438.

[69] Jin, D.J., Meng, F.H., Wang, J.G., The gracefulness of trees with diameter 4, Acta Sci.
Natur. Univ. Jilin., (1993), 17–22.

[70] Jungreis, D., Reid, M., Labeling grids, Ars Combin., 34, (1992), 167–182.

[71] Kang, Q., Liang, Z., Gao, Y., Yang, G., On the labeling of some graphs, J. Combin. Math.
Combin. Comput., 22, (1996), 193–210.

[72] , Kovář, P., Silber, A., Distance magic graphs of high regularity, AKCE International
Journal of Graphs and Combinatorics,, 9(2), (2012), 213?219

[73] Kotzig, A., Rosa, A., Magic valuations of finite graphs, Canad. Math. Bull., 13, (1970),
451–461.

[74] Kotzig, A., Rosa, A., Magic valuations of complete graphs, Publ. CRM, 175, (1972).

[75] Lladó, A., Moragas, J., Cycle-magic graphs, Discrete Math., 307, (2007), 2925–2933.

[76] Lih, K.W., On magic and consecutive labelings of plane graphs, Utilitas Math., 24, (1983),
165–197.

[77] Liu, Y., All crowns and helms are harmonious, manuscipt.

[78] MacDougall, J.A., Vertex-magic labeling of regular graphs, Lecture, DIMACS Connect
Institute, July 18, 2002.

[79] MacDougall, J.A., Miller, M., Slamin, Wallis, W.D., Problems in magic total labellings,
Proceedings of Tenth Australasian Workshop on Combinatorial Algorithms, (1999), 19 – 25.

[80] MacDougall, J.A., Miller, M., Wallis, W.D., Vertex-magic total labelings of graphs, Utilitas
Math., 61, (2002), 3–21.

[81] MacDougall, J.A., Miller, M., Wallis, W.D., Vertex-magic total labelings of wheels and
related graphs, Utilitas Math., 62, (2002), 175–183.

[82] Marr, A.M., Phillips, N. C. K., Wallis, W. D., Bimagic Labeling, AKCE J. Graphs. Combin.,
6(1), (2009), 155–160.

[83] Maryati, T.K., Baskoro, E.T., Salman, A.N.M., Ph -(super)magic labelings of some trees, J.
Combin. Math. Combin. Comput., 65, (2008), 198–204.

[84] Maryati, T.K., Salman, A.N.M., Baskoro, E.T., Supermagic coverings of the disjoint union
of graphs and amalgamations, Discrete Math., 313, (2013), 397–405.

117
[85] Maryati, T.K., Salman, A.N.M., Baskoro, E.T., Ryan, J., Miller, M., On H-supermagic
labelings for certain shackles and amalgamations of a connected graph, Utilitas Math., 83,
(2010), 333–342.

[86] Mata-Montero, M., Normore, S., Shalaby, N., Generalized Langford sequences: New results
and algorithms, International Mathematical Forum, 9(4), (2014), 155–181.

[87] Miller, M., Phanalasy, O., Ryan, J., All graphs have antimagic total labelings, Electronics
Notes in Discrete Mathematics, 38, (2011), 645–650.

[88] Miller, M., Phanalasy, O., Ryan, J., Rylands, L., A note on antimagic labelings of trees,
Bulletin of the ICA, 72, (2014), 94–101.

[89] Miller, M., Rodger, C., Simanjuntak, R., Distance magic labeling of graphs, Australaasian
Journal of Combinatorics, 28, (2003), 305 – 315.

[90] Mühlbacher, J., Magische Quadrate und ihre Verallgemeinerung: ein Graphentheoretisches
Problem, Graph, Data Structures, Algorithms, Hansen Verlag, 1979, München.

[91] Ngurah, A.A.G., Salman, A.N.M., Susilowati, L., H-supermagic labelings of graphs, Dis-
crete Math., 310, (2010), 1293–1300.

[92] Nierhoff, T., A tight bound on the irregularity strength of graphs, SIAM J. Discrete Math.,
13, (2000), 313–323.

[93] Nowakowski, R., Generalizations of the Langford-Skolem problem, M.Sc. Thesis, University
of Calgary, Canada, (1975).

[94] Nurdin, S., Baskoro, E., Salman, A., Gaos, N., On the total vertex irregularity strength of
trees, Discrete Math., 310, (2010), 3043–3048.

[95] Poljak, S., Sǔra, M., An algorithm for graceful labeling of a class of symmetrical trees, Ars
Combin., 14, (1982), 57–66.

[96] Prasanna, N.L., Sravanthi, K., Sudhakar, N., Applications of Graph Labeling in Commu-
nication Networks, Oriental Journal of Computer Science and Technology, 7(1), (2014),
139–145.

[97] Ringel, G., Problem 25, Theory and Its Applications, Proc. Symp. Smolenice, (1963), 162.

[98] Ringel, G., Llado, A., Another tree conjecture, Bull. Inst. Combin. Appl., 18, (1996), 83–85.

[99] Rosa, A., On certain valuations of the vertices of a graph, Theory of Graphs, Internat.
Symposium, Rome, July 1966, Gordon and Breach, N.Y. and Dunod Paris, (1967), 349–355.

[100] Ryan, J., Munushinge, B., Tanna, D., Reflexive irregular labelings, Preprint, 2014.

[101] Salman, A.N.M., Ngurah, A.A.G., Izzati, N., On (super)-edge-magic total labelings of
subdivision of stars Sn , Utilitas Math., 81, (2010), 275-284.

118
[102] Sedláček, J., Problem 27, Theory and Its Applications, Proc. Symp. Smolenice, (1963),
163–164.

[103] Sedláček, J., On magic graphs, Math. Slovaca, 26, (1976), 329–335.

[104] Semaničová-Feňovčı́ková, A., Bača, M., Lascsáková, M., Miller, M., Ryan, J., Wheels are
cycle-antimagic, Electronic Notes Discrete Math., 48, (2015), 11–18.

[105] Seoud, M., Youssef, M., Harmonious labeling of helms and related graphs, manuscript.

[106] Shiu, W.C., Lam, P.C.B., Lee, S.M., On construction of supermagic graphs, J. Combin.
Math. Combin. Comput., 42, (2002), 147–160.

[107] Siddiqui, M., On the total edge irregularity strength of a categorical product of a cycle
and a path, AKCE Int. J. Graphs Combin., 9(1), (2012), 43–52.

[108] Simanjuntak, R., Miller, M., Bertault, F., Two new (a, d)-antimagic graph labelings, Proc.
Eleventh Australas. Workshop Combin. Alg. (AWOCA), (2000), 179–189.

[109] Slamet, S., Sugeng, K.A., MILLER, M., Sum graph based access structure in a secret
sharing scheme, Journal of Prime Research in Mathematics, 2, (2006), 113–119.

[110] Stewart, B., Magic graphs, Canad. J. Math., 18, (1966), 1031–1059.

[111] Stewart, B., Super magic complete graphs, Canad. J. Math., 19, (1967), 427–438.

[112] Sugeng, K.A., Miller, M., On consecutive edge magic total labeling of graphs, Journal of
Discrete Algorithms, 6, (2008), 59–65.

[113] Sugeng, K.A., Miller, M., Bača, M., Super edge-antimagic total labeling, Utilitas Math.
71, (2006), 131–141.

[114] Sugeng, K.A., Miller, M., Slamin, Bača, M., (a, d)-edge-antimagic total labelings of cater-
pillars, Lecture Notes in Computer Science, LNSC 3330, (2005), 169–180.

[115] Sutton, M., Miller, M., Using graph labeling to store linkages in relational databases,
Proceedings of the Eleventh Australasian Workshop on Combinatorial Algorithms (AWOCA),
(2000), 125–138.

[116] Tanna, D., Ryan, J., Semaničová-Feňovčı́ková, A., Edge irregular reflexive labeling of
prisms and wheels, Electronic Notes in Discrete Mathematics (Elsevier), (2016).

[117] Tanna, D., Ryan, J., Semaničová-Feňovčı́ková, A., Bača, M., Vertex irregular reflexive
labeling of prisms and wheels, Electronic Journal of Graph Theory and Applications, (2017).

[118] Tanna, D., Ryan, J., Semaničová-Feňovčı́ková, A., Vertex irregular reflexive labeling of
ladders and generalised petersen graph, Preprint, (2017).

[119] Truszczynski, M., Graceful unicyclic graphs, Demonstratio Math., 17, (1984), 377–387.

[120] Uspensky, J.V., Heaslet, M.A., Elementary number theory, McGraw- Hill Book Company,
Inc., New York, (1939).

119
[121] Vandiver, H.S., Questions and discussions: On algorithms for the solution of the linear
congruence, Amer. Math. Monthly, 31(3), (1924), 137–140.

[122] Wallis, W., Magic graphs, Birkhäuser, Boston - Basel - Berlin, (2001).

[123] Wallis, W.D., Baskoro, E.T., Miller, M., Slamin, Edge-magic total labelings, Austral. J.
Combin., 22, (2000), 177–190.

[124] Wijaya, K., Slamin, Total vertex irregular labelings of wheels, fans, suns and friendship
graphs, J. Combin. Math. Combin. Comput., 65, (2008), 103–112.

[125] Zhao, S.L., All trees of diameter four are graceful, Graph Theory and its Applications:
East and West (Jinan, 1986), 700–706, Ann. New York Acad. Sci., 576, New York Acad.
Sci., New York, 1989

120