Background History

Graph theory is an integral branch of the discrete mathematics that deals with the way objects
are connected. Graph theory is one of the few disciplines of the mathematics that has a precise
origin and it finds its roots in the 18th century when Leonhard Euler (1736) offered a solution to
the famous Konigsberg seven bridge problem by studying the problem using nodes and lines
connecting them in an abstract way, see Figure (1.1).

Each chemical formula defines a compound from the different atoms it contains. The French
chemist Gay-Lussac (1811) observed that different substance had the same chemical
composition, so that their atoms would have to be linked in different arrangement. Such
substances are known as chemical isomers and they always differ from one another in at least
one physico-chemical property. Later on, a British polymath Arthur Cayley (1857) established a
method for counting the constitutional isomers of alkanes via trees, which gave
incentive to George Polya (1937) towards the discovery of his prominent counting theorem.
Historically, in essence, this was the very first encounter between graph theory and chemistry
that gave rise to chemical graph theory.

Kirchhoff (1847) stumble upon graphs while solving a problem related to electrical networks.

Harary rightly pointed out that "graph theory was independently discovered on several
occasions and three names deserve special mention-Euler, Kirchhoff, and Cayley". It is worth
mentioning that the term Graph was first time coined by Joseph Sylvester (1878) due to its
geometrical realization but graphs are now considered to be a topological object rather than
geometrical as no specific length or shape of lines are necessary between nodes.

The colorful nature of graph theory can be observed by the fact that, since its inception, it has
been employed to pursue for the solution of diverse problems. Consequently, graph theory
proved to be a multidisciplinary discipline that has enormous applications in areas like
chemistry, sociology, computer science, engineering, biology, transportation and optimization.
Graph
A graph consists of a set of elements called vertices and a set of elements called edges. Each
edge joins two vertices.

Examples: Here in fig(1) A={a,b,c,d} is the set of vertices and B={ab,ac,bc,cd} is the set of edges.

Fig(1)

Here in fig(2) A={a,b,c,d} is the set vertices and B={ab,ad,bd,bc,cd,cc} is the set of edges.

Fig(2)

Here in fig(3) A={0,1,2,3,4,5,6,7,8,9} is the set vertices and

B={01,05,04,12,16,23,27,38,34,49,57,58,68,69,79} is the set of edges.

Fig(3)

Graph (Network):
Let V(G) be a set of vertices (nodes) and E(G) be a set of edges such that

E(G) ⊆ V(G) x V(G)

Then G= (V(G), E(G))

is called a graph (network).

As graphs are discrete in structure. So, they are also called network.

Adjacent and Incident Vertices

The vertices v and w of a graph are adjacent vertices if they are joined by an edge
e. the vertices v and w are incident with the edge e, and the edge e is incident
with the vertices v and w.

Example:

Fig(1)

Degree of a vertex

In a graph, the degree of a vertex v is the number of edges incident with v, with each loop
counted twice, and is denoted by deg v.

Example

Fig(1)
Pendant Vertex
Let G be a graph, A vertex v of G is called a pendant vertex if and only if v has degree 1. In
other words, pendant vertices are the vertices that have degree 1, also called pendant
vertex.
In the case of trees, a pendant vertex is known as a terminal node or leaf node, or leaf since it
has only 1 degree. Remember leaf nodes have only 1 degree so pendant vertex is called a leaf
node in the case of trees.
Example:

Fig(1)

In the given diagram A and B are pendant vertices because each of them has degree 1.
Multiple edges

In a graph, two or more edges joining the same pair of vertices are multiple edges.

Example:

Fig(1) Fig(2)

Null Graph
A null graph has no edges. The null graph of n vertices is denoted by Nn.
Example:

Fig(1)

Empty graph
If set of vertices = φ

Set of edges = φ

Then the graph is called empty graph.

Note: we cannot draw the empty graph.

Multi-Graph:

A graph having multi-edges is called a multi-graph.

Example:

Fig(1)

The graph above is a multigraph because of the double edge between and and
the triple edge between and

Simple Graph

A graph is called simple graph if it does not contain any loops or multi-edges.

Example:

Fig(1)
Fig( 2)
Order of a graph
The number of vertices in a graph is called order of graph

Fig(1)

Order of graph(1) = 3

Size of a graph
The number of edges in a graph is called size of a graph.

Fig(1)
Size of graph (1) is = 4

Isolated vertex
A vertex of degree zero is called isolated vertex.

Fig(1)
In Fig(1) 4 is isolated vertex

Girth of a graph
The girth of a graph is the length of shortest cycle contained in the graph. If the graph does not
contain any cycle its girth is defined to be infinity.

The Petersen graph has a girth of 5

Wheel graph
We obtain a wheel graph wn when we add an addition as vertex to a cycle for n≥3 and
connect the new vertex to each of (n-1) vertices of by new edges.

Fig(1)

|E( )|=2(n-1)

Tadpole graph ( )

If one of vertex of a path pk is joined with a cycle cn, as shown in figure is called tadpole.
By handshaking lemma

Fig(1)

V( )= n+k

E( )= n+k

Prism graph( )
Prism graph consists of two cycle cn cn’ with addition of n edges obtained after joining their
corresponding vertices of both the cycles.

|E( )|= 3n

Fig(1)

Minimum and maximum degree of G

The minimum degree of G is the minimum among the vertices of G and is denoted by δ(G).
similarly, the maximum degree of G is denoted by ∆(G)

For v Є V(G)

0≤δ(G) ≤ degree v ≤ ∆(G)

If G is simple graph of order n, then degree v≠n

Fig(1)

δ(G) = min{2,2,2,6,1,1,1} = 1

∆(G) = max{2,2,2,6,1,1,1} = 6
Neighbourhood set of a vertex
Let G = (V,E) be a graph and v Є V. The neighbourhood set of vertex "v" is

the set of vertices that are adjacent to v denoted by N (v).

(v) = ,uЄ V(G): "u" is linked (adjacent) with "v"- OR

N(v) = ,uЄ V/uv Є E-

Odd Graph
A graph is called odd graph if the degree of each vertex is an odd number.

Fig(1)

Even Graph

A graph is called even graph if degree of each vertex is an even number.

Fig(1)

Lemma
Every graph has an even number of vertices of odd degree.

Proof

Let be the set of vertices of even degree and be the set of vertices of odd degree. Let size
of G is m, then by handshaking lemma.

2m = ∑
=∑ + ∑

∑ =2m - ∑

Since sum of even degrees is an even. So R.H.S is even. → The number of odd vertices of G is
even. | | is the even number.

Loop and Loop Graph

An edge joining a vertex to itself is a loop and a graph having at least one such loop is called a
loop graph.

Example:

Fig(1)

Pseudo graph
A graph which contains both loop and multiple edge is called pseudo graph.

Example:

Fig(1) Fig(2)

Directed Graph

A directed graph is a graph made up of a set of vertices connected by edges, in

which the edges have a direction associated with them.

Example:

Fig(1)

Undirected Graph
The undirected graph is defined as a graph where the set of nodes are connected together, in
which all the edges are bidirectional. Sometimes, this type of graph is known as the undirected
network.

Example:

Fig(1)

Connected and Disconnected Graph

A graph is connected if any two vertices of the graph are connected by a path,

Example:

Fig(1)

while a graph is disconnected if at least two vertices of the graph are not connected by a path.

Fig(2)

Bridge

An edge in a connected graph is a bridge if its removal leaves a disconnected graph.

Example:

Fig(1)

Regular Graph
A graph is regular if all the vertices of the graph have the same degree. In a regular graph G of
degree r, the degree of each vertex of G is r.

Example:

Fig(1)

a,b,c,d vertices have same degree(3)

Complete Graph

A graph is called complete graph if every two vertices pair are joined by exactly one edge. The
complete graph with n vertices is denoted by Kn.

Example:

Fig(1)

Cycle Graph

If a graph consists of a single cycle or a path which has same initial and final vertex it is called
cycle graph. The cycle graph with n vertices is denoted by .

For cycle , n≥3

Examle:

Fig(1)

Petersen graph
Petersen graph is an undirected graph with 10 vertices and 15 edges.
Example

Fig(1)
Bipartite Graph

If the vertex-set of a graph G can be split into two disjoint sets, V1 and V2, in such a way that
each edge in the graph joins a vertex in V1 to a vertex in V2, and there are no edges in G that
connect two vertices in V1 or two vertices in V2, then the graph G is called a bipartite graph.

Example:

Fig(1)

Complete Bipartite Graph

A complete bipartite graph is a bipartite graph in which each vertex in the first set is joined to
every single vertex in the second set. The complete bipartite graph is denoted by Kx, y where
the graph G contains x vertices in the first set and y vertices in the second set.

Example:

Fig(1)

Subgraphs

A subgraph of a graph G is a graph all of whose vertices are vertices of G and all of whose edges
are edges of G.

Example:

Fig(1) Fig(2)
H is subgraph of graph G.

Proper Sub-graph
If H⊆G and either V(H) is a proper subset of V(G) or E(H) is a proper subset of E(G), then H is a
proper sub-graph of G (H≠G).

Spanning Sub-graph
If a sub-graph of G has the same vertex set as G, then it is a spanning sub-graph of G.

V(H) = V(G)

Fig(1)

Spanning Sub-graph:
If we remove an edge from a sub-graph, then it is called spanning sub-graph. [V(H)| = |V(G)| G-
, G- are the spanning sub-graph.

Note: Null graph is a spanning sub-graph .

Induced Sub graph
A sub-graph F of a graph G is called an induced sub-graph of G if whenever u and v are vertices
of F and uv is an edge of G, then uv is an edge of F as well.

Fig(1)

Cartesian Product (G × H)
For two graphs G and H, then Cartesian product G x H has vertex set V(G) x V(H).
Two distinct vertices (u, v) and (x, y) are adjacent in G × H, if either

(i) u = x and vy ∈ E (H)

ог

(ii ) v = y and vx ∈ E (G)

Addition of Two Graphs:

Let G₁ and G₂ be two graphs, then G₁ + G2 is such a graph in which each vertex of G₁ is joined to
all the vertices of G2.

Complement of a Graph:
Let G be a graph with vertex set V(G) and edge set E(G), then G is the complement of graph G
having vertex set V(G) and edge set E(G) such that V(G) = V(G)

E(G) = {xy: x, y ∈ V(G) ^ x, y ɇ E(G)-

Fig(1)
Self-Complementary Graph:

If G is isomorphic to ̅ , i.e.,

G̃ ̅

Then the complement of G is itself is called the self-complementary graph.

G= ̅

No. of edges = No. of edges

No. of vertices = No. of vertices

Degree of vertices = Degree of vertices

Fig(1)

Line Graph: L(G)

Let G be a graph with vertex set V(G) and edge set E(G), then its line graph is a graph with
vertex set

V (L(G)) = E(G)

and edge set

E(L(G)) = {uv: u, v ∈ V(L(G)), u & v are incident in G}

Let G be a graph.
Then its line graph L(G) is

Handshaking lemma
Hand shaking lemma states that the sum of degree of all the vertices in a graph is equal to twice
the number of edges in a graph. Mathematically, we can write as follows:

Fig(1)

The sum of degrees of all six vertices is 2 + 3 + 2 + 3 + 3 + 1 = 14, twice the number of edges.

Walk

A walk of length k in a graph is a succession of k edges of the for

Example:

Fig(1)

Uv ,vw ,wx, …, yz

This walk is denoted by uvwx…yz, and is referred to as a walk between u and z.

Trails

A trail is a walk in which all the edges, but not necessarily all vertices, are different.

Example:

Fig(1)

Walk vzzywxy is a trail

Path

A path is a walk in which all the edges and all the vertices are different.

Example:

Fig(1)

Walk vwxyz is path

Closed walk

Fig(1)

A closed walk in a graph is a succession of edges of form

Uv,vw,wx,…,yz,zu

That starts and ends at same vertex.

Closed trail

A closed trail is a closed walk in which all the edges are different.

Example:

Fig(1)

The closed walk vywxyzv is closed trail.

Cycle

A cycle is a closed walk in which all the edges are different and all the intermediate vertices are
different.

Example:

Fig(1)

The closed trail zz, vwxyv are cycles.

Open walk or open trail

A walk or trail is open if it starts and finishes at different vertices.

Example:

Fig(1)

vzyx is open walk or trail.

Isomorphism
If two graphs G and H contain the same number of vertices connected in the same way, they
are called isomorphic graphs (denoted by G≅H).
It is easier to check non-isomorphism than isomorphism.
If any of these following conditions occurs, then two graphs are non-isomorphic

The number of connected components are different

Vertex-set cardinalities are different
Edge-set cardinalities are different
Degree sequences are different

Example
The following graphs are isomorphic

Fig(1) Fig(2) Fig(3)

Tree
A Tree is a connected acyclic undirected graph. There is a unique path between every pair of
vertices in G. A tree with N number of vertices contains (N−1) number of edges. The vertex
which is of 0 degree is called root of the tree. The vertex which is of 1 degree is called leaf node
of the tree and the degree of an internal node is at least 2.

The following are examples of tree –

Labeled Trees
Definition − A labeled tree is a tree the vertices of which are assigned unique numbers from 1
to n. We can count such trees for small values of n by hand so as to conjecture a general
formula. The number of labeled trees of n number of vertices is nn−2nn−2. Two labeled trees
are isomorphic if their graphs are isomorphic and the corresponding points of the two trees
have the same labels.
Example

Fig(1)

Fig(2)

Unlabeled Trees
An unlabeled tree is a tree the vertices of which are not assigned any numbers. The number of
labeled trees of n number of vertices is (2n)!(n+1)!n! (nth Catalan number)
Example

Fig(1) Fig(2)
 Fig(3)

Rooted Tree
A rooted tree G is a connected acyclic graph with a special node that is called the root of the
tree and every edge directly or indirectly originates from the root. An ordered rooted tree is a
rooted tree where the children of each internal vertex are ordered. If every internal vertex of a
rooted tree has not more than m children, it is called an m-ary tree. If every internal vertex of a
rooted tree has exactly m children, it is called a full m-ary tree. If m=2m=2, the rooted tree is
called a binary tree.

Fig(1)

Fig(2)

Binary Trees:
A tree-like structure that is rooted and in which each vertex has at most two children and each
child of a vertex is designated as its left or right child.
Example:

Fig(1) Full binary tree

1 Introduction
All graphs are finite, simple and undirected. The graph G has vertex set V=V(G) and edge set
E=E(G) and we let e= |E| and v = V. A general reference for graph theoretic notions is [8]. We
will deal only with connected graphs, although the concepts apply equally to graphs with more
than one connected component.

A labeling for a graph is a map that takes graph elements to numbers (usually positive or non-
negative integers). In this paper the domain is the set of all vertices and edges, while other
labelings have used the vertex set alone, or more rarely the edge set alone. The most complete
recent survey of graph labelings is [2].

Various authors have introduced labelings that generalize the idea of a magic square. Kotzig
and Rosa [5], for example, defined a magic labeling to be a labeling on the vertices and edges in
which the labels are the integers from 1 to e+v and where the sum of labels on an edge and its
two endpoints is constant. Related labelings have been studied by other authors and there are
numerous variations in the terminology used. Readers are referred to [6] for a discussion of
these matters and a standardization of the terminology. That paper studies in detail the kind of
labeling described above, which we will now refer to as an edge-magic total labeling.

In this paper we introduce the notion of a vertex-magic total labeling. This will be an
assignment of the integers from 1 to e+ v to the vertices and edges of G so that at each vertex
the vertex label and the labels on the edges incident at that vertex add to a fixed constant.
More formally, a one-to-one map A from EUV onto the integers {1, 2,..., e + v) is a vertex-magic
total labeling if there is a constant k so that for every vertex x,

λ(x) + Σλ(xy) = k (1)

where the sum is over all vertices y adjacent to z. Let us call the sum of labels at vertex z the
weight of the vertex; so we require wt(x)= k for all x. The constant k is called the magic constant
for λ.

The notion of vertex-magic labeling was at least partially suggested by the following question
which appeared on a set of mathematical enrichment problems for high school students [4]
(The problem appeared at the time when corruption charges against International Olympic
Committee members were dominating the news):

The Olympic emblem consists of 5 overlapping rings containing 9 regions. In order to contribute
to a pension fund for a retiring IOC delegate, people are asked to deposit money into each
region. The guidelines allow the delegate to take all the money in any one of the rings. Place $1,
$2, $9 in the 9 regions so that the amount in each ring is the same.

In all that follows we will use the term labeling to mean vertex-magic total labeling. It is not
hard to find examples of labelings for some graphs. One labeling for the graph K4e is shown in
Figure 1. On the other hand, not every graph has a labeling. For the graph K2, since λ(x) ≠ λ(y),
then λ(x) + λ(xy) ≠ λ(y) + λ(xy), and so no labeling is possible.

In subsequent sections, we show how to construct labelings for several families of graphs and in
some cases carry out complete enumerations of the admissible labelings. In addition, we
determine several classes of graphs which cannot be labeled. The general flavour of the results
that we have discovered so far can be summed up very roughly as follows: if there is not much
variation among the degrees of the vertices then the graph possesses a labeling; if there is, then
it does not.

2 Basic Counting
Set M= e+ v and let S, be the sum of the vertex labels and Se the sum of the edge labels. Clearly,
since the labels are the numbers 1,2,..., M, we have as the sum of all labels

+ =∑ =( )

At each vertex we have λ( ) + Σ λ( y) = k. We sum this over all v vertices . This a dds each
vertex label once and each edge label twice, so that

+2 = vk. (2)

Combining these two equations gives us

+( ) = vk. (3)
The edge labels are all distinct (as are all the vertex labels). The edges could conceivably receive
the e smallest labels or, at the other extreme, the e largest labels, or anything between.
Consequently we have

∑ ≤ ≤∑ (4)

A similar result holds for . Combining (3) and (4), we get

( )+ ( ) ≤ vk ≤ 2 ( )− ( )

which will give the range of feasible values for k.

It is clear from (1) that when k is given and the edge labels are known, then the vertex labels
are determined. So the labeling is completely described by the edge labels. Surprisingly,
however, the vertex labels do not completely determine the labeling. Having assigned the
vertex labels to a graph, it may be possible to assign the edge labels to the graph in several
different ways. Figure 2 shows two labelings of W4 which have the same vertex labels. but a
different edge labeling.

3 Regular Graphs

If a regular graph possesses a labeling, we can create a new labeling from it. Given a labeling λ
for any graph G, define the map λ' on EUV by

λ'( )= M +1-λ( )

for any vertex , and

λ'(xy)= M+1-λ(xy)

for any edge xy. Clearly X' is also a one-to-one map from the set EUV to ,1,2,…..,e+v) and we will
call λ' the dual of λ. We have the following:
Theorem 1
The dual of a labeling for a graph G is a labeling if and only if G is regular.

Proof. Suppose λ is a labeling for G with wt( ) = k. Then

where

wt'(x)= λ'(x)+ ∑ λ(xy)

=M+1-λ(x)+∑(M + 1-λ(xy))

=(r+ 1)(M + 1) – k

Where r is the number of edges incident at x. Clearly this is constant if and only if r is constant;
in other words, if and only if G is regular. Then (r+ 1)(M + 1)-k is the magic sum for the dual.

The general problem of whether one can use a labeling of a graph G to produce a labeling of
some subgraph or supergraph of G appears to be very difficult. The next theorem answers a
very special case of this question for regular graphs.

Theorem 2
Let G be a regular graph having a labeling in which the label 1 is assigned to some edge e'. Then
the graph G-e' has a labeling.

Proof. Suppose G is an r-regular graph. Define a new mapping Λ by Λ( ) =λ( )-1 and Λ(xy) =
λ(xy) — 1. Now A is a one-to-one mapping from EUV to {0,1,2,….,e+v-1}. If we now delete the
edge e' (which is labeled 0 by Λ) from G, then Λ is a one-to-one map from (E-e') U V to
{0,1,2,….,e+v-1). Also

w ( ) =λ( )-1+∑(λ(xy)) - 1)

=w ( ) -r-1

=k-r-1

Thus w ( ) is a constant k' and so Λ is a labeling for G-e'.

Theorem 3
If G is a regular graph and e an edge such that G -e has a labeling, then that labeling is derived
from a labeling of G by the process described in Theorem 2.
Proof. Suppose G is an r-regular graph and let λ be any labeling for G-e where e is the edge
. Adjoin the edge to G-e and define λ( )= 0. Now simply adding 1 to all the labels
defines a new mapping λ' which is easily seen to be a labeling for G having the label 1 appearing
on the edge . If the magic sum of λ is k, then the magic sum of λ' is k+r+ 1.

4 Cycles and Paths

The easiest regular graphs to deal with are the cycles. For cycles (and only for cycles) a vertex-
magic total labeling is equivalent to an edge-magic total labeling, and the edge-magic labelings
have already received some attention (see [6] and the other papers cited there).

Theorem 4
The n-cycle has a labeling for any n≥3.

Proof. The constructions of Theorems 4, 5, 6, and 7 of [6] provide an edge-magic total labeling
λ' for for every n ≥ 3. In other words there is a constant k so that

λ'( )+ λ'( ) + λ'( )=k

for all vertices of . If we define a new mapping λ by λ( ) = λ'( ) and λ ( )=

λ'( ), then we clearly have k as the weight at each vertex, and so λ is a vertex-magic total
labeling of G.

Corollary 1
, the path with n vertices, has a labeling for any n≥ 3.

Proof. For each n ≥ 3 at least one of the edge-magic labelings λ' referred to in the proof of the
theorem assigns the label 1 to a vertex. Then the corresponding vertex-magic labeling λ will
assign the label 1 to some edge e. By Theorem 2, - e will have a vertex-magic labeling.

Corollary 2
Every labeling of is derived from a labeling of .

Proof. This follows immediately from Theorem 3.

Using equations (3) and (4), we can readily determine the feasible values. of k for the n-cycle.
We find

≤k≤
A systematic search has found exactly 4 labelings for the 3-cycle; one for each feasible value of
k. As mentioned above, once k is given, the edge labels completely determine the labeling, so
we list only the edge labels.

k= 9 1,2,3

k= 10 1,3,5

k= 11 2,4,6

k=12 4,5,6

Since the cycles are regular graphs, the duality described in Section 2 applies. The labeling with
k = 12 is dual to the labeling with k = 9, and the labeling with k = 11 is dual to that with k = 10.

There are exactly 6 labelings for the 4-cycle and again they come in dual pairs. Every feasible
value of k admits a labeling. They are listed below, with the edge labels given in cyclic order:

k= 12 1,3,2,6

k= 13 1,4,6,5

k=13 1,5,2,8

k= 14 3,4,8,5

k= 14 1,7,4,8

k = 15 3,7,6,8

For the 5-cycle, we find 14 ≤ k ≤19. There are 6 labelings. None exists corresponding to k =15 or
k= 18. For k= 14 there is a unique solution (1, 4, 2, 5, 3), and for k 16 we find (1,5,9,3,7) and also
(1,7,3,4,10).

Each of these has a dual.

For , , , , , Godbold and Slater [3] have found the number of non-isomorphic
labelings to be 20, 118, 282, 1540, 7092 respectively.

The Olympic rings problem referred to in the Introduction can be interpreted as asking for a
labeling on the path of 5 vertices; one of the solutions the students found is shown in Figure 3.
Figure 3: One labeling of .

In the case of a path with v vertices, we have e=v−1 and M = v+e = 2v−1 and therefore

+ v(2v- 1)= vk

So ≡ 0 (modv) and consequently, from (2), ≡ 0 (modv) also. Equations (3) and (4) then give
us

≤ ≤3

from which it follows that

≤k≤ (5)

For a path with 3 vertices, (5) implies that 6 ≤ k ≤8. According to Theorem 5, the labelings of
will be derived from those of which assign the label 1 to an edge. There are two such listed
previously and they produce the labelings for shown in Figure 4. There is no labeling of
with magic constant k = 8.

Figure 4: The two labelings of the path .

For the path with 4 vertices, equation (5) implies that 9≤ k ≤ 11. Four of the six labelings of the
4-cycle listed previously yield a derived labeling for the 4-path. This time there is a labeling for
all feasible values of k (we list the edge labels only):

k= 9 2,1.5

k=10 4,5,3

k=10 4.1.7

k=11 6,3,7
For v=5, equation (5) implies that 11 ≤ k ≤ 15. There are 4 labelings; they correspond to k = 11,
13 and 14. There is no labeling with k=12 or k = 15.

k= 11 2, 4, 1, 3

k= 12 none

k= 13 4,8,2,6

k= 13 9,3,2,6

k= 14 6,7,3,9

k= 15 none

Further systematic search discovered 10 labelings for v=6; there are 66 for v = 7, and for v = 8
there are 131 labelings.

Graphs5 Complete Bipartite

The complete bipartite graphs provide us with interesting and surprising. results. Labelings for
and exist and have been dealt with above as a path and a cycle respectively. The first
general result shows that for a labeling to exist, the parts must be nearly the same size.

Theorem 5 If n> m+1, then has no labeling.

Proof.

We prove that has no labeling. Let the vertices of one part of be { ,……., }
and the vertices in the other part be ( ,……, ) Then v = 2m+ 2 and e= +2m so that the
label set is {1,2,……, +4m+2). Then the sum of the weights on { ,…., } is at least the sum
of all but the largest m + 2 labels, so

wt( )+ wt( )+··· wt( )

≥ 1+2+…..+ +3m

Consequently

K≥

= (6)
On the other hand, the sum of the weights on {y1,, ym+2} is at most. the total of all but the m
smallest labels:

Thus

wt (1)+wt (2) +···· wt (ym+2)

≤ (m+1)+(m+2)+…..+ ( + 4m+2)

( )( )
=

Thus

( )( )
K≤ (7)

However, a straightforward calculation shows that (7) is always less than (6). So no labeling can
exist for and, a fortiori, for with n> m + 2.

In contrast with the result of Theorem 5, we show that a labeling can always be found when
m=n. Curiously enough, the proof uses a magic square construction.

Theorem 6
There is a labeling for for every m> 1.

Proof. Let S = S(i, j) be any magic square of order m+1 on the numbers {1,…….., ). Each
row and column sums to the magic square constant (m + 1)( + 2m+ 2). Form the matrix M
where M (i, j) = S(i, j) - 1. Since S is magic, the rows and columns of M will each sum to the
constant

k = (m + 1)( + 2m+ 2) − (m + 1) (8)

and the entries of M will be the numbers in (0,…., -1), once each. There are standard
constructions for magic squares ([1],[7]) that will guarantee that 1 occurs in the first (or last)
row or column. We will assume that the rows and columns of M are permuted so that M(1, 1) =
0 (This means the diagonals may no longer sum to the magic constant, but they are not
required for the construction.)

Let the vertices of be { ,…..

, ,….., } with the u; in one part

and the in the other. We then define a labeling λ by

λ( )=M(i + 1, 1)

λ( )=M(1, i + 1)

λ( )=M(i + 1, j +1)

Now the sum of any row or column corresponds to the sum of labels on a vertex and its
incident edges, so A is a labeling with k as in equation (8).

The labeling constructed in the proof of Theorem 8 satisfies an additional property beyond that
required by the definition. The entries in the first row of M add to the magic constant k, as do
the entries in the first column. This means that the labels on the vertices in each part-set also
sum to the magic constant k.

Corollary 3
There is a labeling for the graph - e for every m > 2.

Proof. In the standard constructions for magic squares referred to above, the entry 2 always
appears in a different row and column from the entry 1. Consequently in the labeling defined in
the proof of the theorem, the label 1 is always assigned to some edge e. Then by Theorem 2,
the graph - e has a labeling as well.

Using equations (3) and (4), we can determine the range of feasible values of k for a labeling of
. In this case, we have M = + 2m and this leads to the range

[ - ]≤k≤ [ + ]

The magic sum from the construction in the theorem simplifies to

[ -(m+1)]

which is seen to be within the range above.

The only other complete bipartite graphs which might admit labelings are those in which n = m
+ 1. We have not yet found a construction which produces a labeling for for all m. A
systematic search has revealed. 4 distinct labelings for and they are listed in the table
below. The numbers are the labels on the edges and the groupings represent the edges

meeting each of the vertices in the 2-part of . The first labeling listed is displayed in Figure
5.

k= 18 {4, 6, 1}, {3, 2,8}

k=19 {5, 6, 1}, {3, 4, 10}

k=19 {6, 7, 1}, {2, 3, 10}

k= 20 {7, 8, 4}, {2, 3, 10}

Again, the ease of finding small examples leads us to believe that it can always be done.

Figure 5: A labeling of .

Graphs with Vertices of Degree One

For a graph to have a labeling, the presence of degree one vertices turns out to create a
restriction on both the number of edges and on the number of vertices of higher degree in the
graph. As an illustration, we first examine a family of graphs in which the number of vertices of
degree one is the same as the number of vertices of higher degree.

Let G be any graph of order n and size e. We define a G-sun to be a graph G* of order 2n
formed from G by adjoining n new vertices of degree one to the vertices of G. We have v* =
|V(G*)| = 2n and e* = |E(G*)| = e + n. When G is the cycle and one new vertex is adjoined to
each vertex of G, the resulting G* has been commonly referred to as a sun.

Let ,……, be the vertices of G and ,……., the degree one vertices. Let us call the edges
of G the inner edges of the G-sun and the the inner vertices; the others will be the outer
edges and vertices. The following theorem shows that for a labeling to exist, the number of
edges in G must be bounded above by a function of n which is essentially linear.

Theorem 7
Let G be any graph of order n. If G has e edges, then a G-sun G* has no labeling whenever

√
e>
Proof. The label set for G* is {1, 2,..., 3n+e}. We calculate the minimum possible sum of weights
on the inner vertices; this is achieved by putting the e smallest labels on the inner edges and
the next 2n smallest labels on the inner vertices and outer edges. Remembering that the inner
edge labels will each be added twice, this gives us

∑wt( ) ≥ ∑ +∑

= [(2n+e)(2n + e + 1) +e(e + 1)] .

Since there are n inner vertices, we must therefore have:

k ≥ [2 + n(2e + 1) +e(e + 1)] .

Calculating the maximum possible sum of weights on the outer vertices, we find (taking the
sum of the 2n largest labels):

∑wt( )≤ ∑ -∑

= *(3n+e)(3n + e + 1) − (n + e) (n + e + 1)+

4 + 2ne+n.

Since there are n outer vertices, we have

K ≤ (4 + 2ne+n)

= 4n+ 1 + 2e.

Consequently, a labeling cannot exist whenever

[2 + n(2e + 1) + e(e + 1)] > 4n+1 + 2e.

This simplifies to

and the theorem follows.

+e-2 >0

A second illustration of the impact of degree one vertices is found by examining trees. This time
the existence of a labeling forces a lower bound on the number of internal vertices in the graph.
We have the following:

Theorem 8
Let T be a tree with n internal vertices. If T has more than 2n leaves, then T does not admit a
labeling.

Proof. The proof is similar to that of the previous theorem. Suppose T has 2n+1 leaves, ,···,
(and n internal vertices ,…, ) so that v=3n+ 1 and e = 3n. Then the label set is {1, 2, . .
., 6n+ 1}.

The maximum possible sum of weights on the leaves is the sum of the 4n+2 largest labels, i.e.

Σwt( ) ≤ ∑ -∑

=(8n+1)(2n+1)

and since there are 2n + 1 leaves, we get

k≤ 8n+1.

The minimum possible sum of weights on the internal vertices is the sum of all but the 2n+1
largest labels. Therefore

Σwt( ) ≥ ∑

= 2n(4n+1).

Since there are n internal vertices, we get

K ≥ 8n +2

and these two inequalities show that a labeling is impossible. It is easily checked that a higher
proportion of leaves only worsens the inequalities.

The labeling of the tree with 2 internal vertices and 4 leaves displayed in Figure 6 shows that
labelings for a tree with n internal vertices and 2n leaves are possible.

Figure 6: A labeling of .
7 The Complete Graphs
A systematic search for labelings on has revealed interesting results. In this case Є
{1,2,….,10) and equations (3) and (4) imply that 19 ≤ k ≤ 25. The table shows the number of
distinct labelings for each value of k.

k= 19 none

k=20 2

k=21 5

k=22 none

k=23 5

k=24 2

k=25 none

As described in Section 2, these come in dual pairs also; those labelings with k=24 are the duals
of those with k = 20 and those with k = 23 are duals of those with k = 21. The graph pictured in
Figure 1 is -e and its labeling was obtained from one of the above labelings with k=20 by
deleting an edge labeled with 1, as described in Theorem 2.

We can apply equations (3) and (4) for arbitrary n to find the range of feasible values of k for
. Using v = n and e= , we find:

≤k≤

Theorem 9
There is a labeling for for all odd n.

Proof. Let , ,…., be the vertices of the complete graph, where n is odd. Let A and B be
orthogonal Latin squares of side n, each based on {1, 2,..., n}, having the following properties:

[A1] A is symmetric, and its diagonal entries are all different;

[B1] B has every diagonal entry 1;

[B2] + =n+2 for every i and j.

Suitable Latin quares are easy to construct. For example, one can take the first row of each as
1,2,…,n; A is found by back-circulating, and B by circulating.

Now define an array C by = min{ , - − 1)) and an array D by = +n . Then set

λ( )= , λ( ) = . We claim that λ is a vertex-magic total labeling. This is because:

[1] By orthogonality, the elements of A+n(B-1) are all different thus providing the required
labels.

[2] In the array C there are exactly two copies of each label from n + 1 to (1) and because of B2,
exactly one of each lies above the diagonal.

[3] The sum of each row of D is a constant, namely

(1 + 2 + ... + n) + 2(1 + 2 + ··· + )

where the first sum results from the elements of A and the second from B-1. This guarantees
the magic property.

The magic constant k for the labeling produced by the above construction is which is the
maximum possible. By permuting the entries of C, it is possible to construct labelings
corresponding to other values of k.

We have not yet found a convenient construction that will produce labelings for , for all even
n. However, the ease with which examples have been found for small values of n suggests the
following

Theorem 10
There is at least one labeling for and for for every feasible value of k.

Proof. Labelings for and have been found by brute force. For , we must have 35 ≤ k ≤
45. In Table 1 we exhibit one solution for each k≤40. For , we get 59 ≤ k ≤ 73. Shown in Table
2 is one solution for
Table 1: Labelings of the Complete Graph on 5 vertices

each k≤ 66. In both cases the duals provide the remaining examples.

The fact that labelings seem to be plentiful for n = 5 and n = 6 suggests the following
conjecture. The absence of labelings for some values of k in the case of is probably just the
Law of Small Numbers in operation.
Table 2: labeling of the complete graph on vertices