1

Graph Theory with its Applications

Bondy & Murty
Chapter 1

Section 1.2

1.2.9.Definition: k – Partite graph: A k – Partite graph is one whose vertex set can
be partitioned into k subsets so that no edge has both end points in any one
subset.

Definition: A complete k – Partite graph: A k -Partite graph is called a complete k

- Partite graph if each vertex is joined to every vertex that is not in the same subset.

1.2.10. Definition : k –cube: The k –cube is the graph whose vertices are the
ordered k –tuples of 0’s and 1’s. Two vertices are adjacent iff they differ in exactly
one coordinate.

1.2.11. Definition: The complement of a graph: Let G be simple a graph. The

complement of G, denoted by Gc, is the simple graph with vertex set V(G) and two
vertices are adjacent in Gc iff they are not adjacent in G.

Self Complementary: A simple graph G if self complementary if G is isomorphic

to Gc.

Show that if G is self complementary then ν 0, 1 (mod 4)

Let G be a self complementary graph with ν vertices and Ɛ edges. Since G Gc

Ɛ(G) = Ɛ(Gc ).
But Ɛ(G) + Ɛ(Gc ) = ν C2 ( since G U Gc is a complete graph)

2 Ɛ(G) = ν(ν-1)/2 4Ɛ(G) = ν(ν-1).

ν is a multiple of 4 or (ν-1) is a multiple of 4.
ν 0, 1 (mod 4).

Section 1.3 The Incidence and Adjacency matrices

The incidence matrix of G:

Let G be a graph with v1, v2, . . . vν vertices and e1, e2, . . . eƐ edges. The incidence
matrix of G is the matrix M(G) = [mij] where mij is the number of times(0,1 or 2)
that vi and ej are incident.

The adjacency matrix of G is the matrix A(G) = [aij] in which aij is the number of
edges joining vi and vj.
 2

Parity: Two integers n, K N(often n = |X|, k = |Y| for any two sets X and Y) have
the same parity if both are even or both are odd, that is if n ≡ k (mod 2). Otherwise
they have opposite parity.

Induced subgraph example:

G is a graph. G’ is a induced subgraph whereas G” is not a induced subgraph.

Section 1.5

k – regular: A graph G is k –regular if d(v) = k for all v .

Example: Kn, Km,n, and k – cubes.

1.5.1. Show that δ ≤ 2Ɛ/ν ≤ Δ.

We know that δ ≤ d(v1) ≤ Δ

δ ≤ d(v2) ≤ Δ

. .
. .
. .
δ ≤ d(vν) ≤ Δ
Adding all the inequalities we get

νδ≤ th ≤ ν Δ which implies ν δ ≤ 2Ɛ ≤ ν Δ. This gives the result.

1.5.5 Definition: Degree sequence of a graph G.

If a graph G has vertices v1, v2, . . . vν, the sequence (d(v1), d(v2), . . . d(vν)) is called a
degree sequence of G.

1.5.6 Definition: Graphic: A sequence d = (d1, d2, . . . dn) is graphic if there is a simple
graph with degree sequence d.
 3

a) Show that the sequences (7, 6, 5, 4, 3, 3, 2) and (6, 6, 5, 4,3, 3, 1) are not graphic.

Consider the sequence (7, 6, 5, 4, 3, 3, 2). Here ν(G) = 7. The fist degree is given as 7.
This implies that there is a loop in the graph which is not possible in a simple
graph. Therefore this sequence is not graphic.

Consider the sequence (6, 6, 5, 4,3, 3, 1). Let d(v1 ) = 6, and d(v2) = 6. This means the
degree of each vertex is minimum two which contradicts the given sequence in which d(v7) =
1. Therefore this sequence is not graphic.

1.5.10. definition: The edge graph of a graph G. This is a graph with vertex set E(G) in
which two vertices are joined if and only if they are adjacent edges in G.

Section 1.6 Paths and Connection

Defintion: Walk: A walk is a finite sequence of vertices and edges occurring alternately, in
which the vertices and edges may be repeated.

Trail: A trail is a walk in which the edges will occur at most once.

Path: A path is a trail in which the vertices are not a repeated.

If W = v0 e1v1 e2. . . ek vk is a walk, then v0 is the origin and vk is the terminus. v1, v2, . . . vk-1
are called the internal vertices.

Connected: Two vertices u and v are said to be connected if there is a (u,v) path in G.

Connection is an equivalence relation in G.

Reflexive: The vertex u is connected to u by a path of length 0.

Symmetry: If u and v are connected by a (u,v) path, then v and u are connectd by the same
path.

Transitive; If u and v are connected by a (u,v) path, v and w are connected by a (v, w) path,
then u and w are connected by a (u,v,w) path.

Connected graph: Let the vertex set V(G) be partitioned into V1, V2, . . ., Vω. Tow vertices u
and v are said to be connected if and only if they are in the same set Vi . The subgraphs
G[V1], G[V2], . . G[Vω] are called the components of G. A graph is a connected graph it has
only one component.

The no. of components of G is denoted by ω(G).

1.6.7. show that if G is disconnected then Gc is connected.

G is disconnected means it has more than one component. Let u, v be any two vertices in G.
If u and v belong to different components of G they are non adjacent in G but they are
adjacent in Gc .
 4

If u and v belong to the same component of G, we choose a vertex w in a different

component of G. then (u, w, v) is a (u, v) path in Gc. Hence Gc is connected.
1.6.11

Let u and v be any two vertices of G. The distance between u and v in G is the length of the
shortest(u,v) path in G. this is denoted by dG(u,v). If there is no path between u and v then
dG(u,v) is infinite.

1.6.12 The diameter of G is the maximum distance between two vertices of G.

Section 1.7.

Closed walk: A walk is closed if it has positive length and it origin and terminus are the
same.

Cycle: A closed path is a called a cycle. A cycle of length k is called k-cycle.

Triangle: A 3 cycle is called as a triangle.

Theorem 1.2: A graph is bipartite if and only if it has no odd cycles.

Proof: Let G be bipartite graph with bipartition (X, Y). Let C: v0, v1, . . ., vk,, v0 be a (k+1)-
cycle in G. Let v0 X without loss of generality. Then v1 Y, v2 X, v3 Y, , . . , vk y
since v0 X ,
In general, v2i X, v2i +1 Y. This implies k = 2i +1. Hence k+1 = 2i, even. The cycle
length is even.

Conversely let G be a cycle with no odd cycles. To prove G is bipartite. Let u V(G) be an
arbitrary vertex . Define

X= { x V(G), d(u, x) is even}

Y= {y V(G), d(u, y) is odd}. We shall show that (X,Y) is a bipartition of G.

Suppose that that v and w are two vertices of X. Let P be a shortest (u,v) path and Q be a
shortest path (u,w) path . Let u1 be the last vertex common to both P and Q. Since P and Q
are of shortest paths, the sections uPu1 and uQu1 are of same length.

If u1 belongs to Y then the sections u1v and u1w are of odd length since v and w are vertices
of X. The length of the pathvu1w is even. If v and w are joined by edge, the length of the
cycle vu1w is of odd length. This is a contradiction to the fact that G has no odd cycles.
Hence v and w of X cannot be adjacent.

Similarly If u1 belongs to X then the sections u1v and u1w are of even length since v and w are
vertices of X. The length of the pathvu1w is even. If v and w are joined by edge, the length of
the cycle vu1w is of odd length. This is a contradiction to the fact that G has no odd cycles.
Hence v and w of X cannot be adjacent.
 5

In a same way we can prove that no two vertices are adjacent in Y also.

Hence the theorem.

1.7.2. Show that if δ(G) ≥2, then G contains a cycle.

Let us assume that G is connected with δ(G) ≥2. Let v0 be an arbitrary vertex and consider a
longest path from v0 to any other vertex say vk. W = v0 e1 v1e2 . . . ek vk. Since δ(G) ≥2, vk
will be incident with some edge other than ek . if it is an edge whose end point in W, then G
contains a cycle. If it is an edge whose end point is not in W, then we get a contradiction to
the fact that W is the longest path in G.
Hence if δ(G) ≥2, then G must contain a cycle.

1.7. 4 Girth of a graph G: is the length of a shortest cycle G. If G has no cycles the girth of
G is infinite.

Chapter 2: Trees

2.1 Trees
An Acyclic graph is a graph with no cycles. An acyclic graph is called a forest.

A tree is a connected acyclic graph.

Theorem: 2.1 In a tree, any two vertices are connected by a unique path.

We prove the theorem by contradiction. Let there be two distinct (u,v) paths P and Q. Let e = xy be
an edge in P and not in Q. Now the subgraph (P U Q) – e is connected. It contains a xPQy path, say
R. Now R +e is a cycle which contradicts that a tree is a acyclic graph.

Theorem: 2.2 : If G is a tree then Ɛ = -1.

We prove the theorem by induction on When = 1, G is K1. In this case Ɛ =0. Therefore Ɛ = -
1.

Suppose that the theorem is true for all graphs with fewer than vertices. Let G be a tree
with t 2. Let uv = e be an edge in G. Then G – e contains no (u,v) path since uv is the unique (u,v)
path. Hence G – e is disconnected, so that ω(G-e) =2. The components G1 and G2 of G –e are
trees, because they are connected and acyclic. G1 and G2 has fewer than vertices. So the
theorem is true for both G1 and G2. That is Ɛ(G1) = t and Ɛ(G2) = tG2) -1.

Now Ɛ(G) = Ɛ(G1) + Ɛ(G2) + 1 = e = tG1) -1 + tG2) -1 + 1 = tG1 U G2) -1 = . tG) -1. Hence
it is true for all the trees.

Corollary: 2.2. Every nontrivial tree has at least two vertices of degree one.

Proof: Let G be a nontrivial tree. Then d(v) ≥ 1 for v V(G). This implies d(v1) ≥ 1, d(v2) ≥ 1 , . . .
d(h ) ≥ 1. Summing up we get h VtG dth ≥ 1+1 + 1 . . . times. This implies 2 Ɛ ≥ .
2( -1) ≥ . t 2. This implies the no. of vertices of degree one is at least 2.
 6

2.1.5: Let G be a graph with Ɛ(G) = -1. Show that the following statements are equivalent.

1. G is connected. 2. G is acyclic 3. G is a tree.

1. 2.
The graph G is connected. Let us prove that G is acyclic. Suppose that G has a cycle. Let ‘e’
be an edge in that cycle. Now G –e is still connected with -2 edges. - Ɛ = – ( – 2) = 2.
This is a contradiction to the fact that every graph with vertices and Ɛ edges has at least -
Ɛ components, since G-e has only one component.

2 3
The graph G is acyclic. Let us prove that G is a tree. In order to prove this, We have to
show G is connected. Let G has ω components G1, G2, . . . G ω. Each component is a tree
since it is connected and acyclic. Therefore Ɛ(Gi) = (Gi) -1. Summing up from i = 1 to ω,
we get Ɛ(G) = - ω. But given in the hypothesis that Ɛ(G) = -1. Therefore ω = 1. This
implies G has one component only. This shows that G is connected. Hence G, being acyclic,
is a tree.

3 1. G is a tree. By definition G is Connected Hence (1).

Corollary 2.4.1.

Every Connected graph contains a spanning tree.

Proof: Let G be connected. Let T be a minimal connected spanning subgraph of G. By

definition ω(T) = 1(Since it is connected) and ω(T - e) > 1. This gives ω(T - e) > 1 > ω(T)
This implies each edge ‘e’ is a cut edge. By theorem 2.4, T is a tree.

Corollary 2.4.2.

Let G be connected. Then Ɛ t

Proof: Let G be connected. By Corollary 2.4.1, G contains a spanning tree. Therefore

Ɛ(G) t Ɛ(T) (Because more edges in G than in T)
t (T) – 1
= (G) – 1.

Theorem 2.5
Let T be a spanning tree of a connected graph G and let e be an edge of G not on T. Then T +
e contains a unique cycle.

Proof:
 7

Each cycle of T + e contains the edge ‘e’. C is a cycle of T + e iff C –e is a path in T

connecting the ends of e. This is a unique path, since any two vertices in a tree are connected
by a unique path. Therefore T + e contains a unique cycle.

Definition: Let S and S’ be two subsets of V(G). The set of edges with one end in S and
another end in S’ is denoted as [S, S’].

Let = V\S where S is a nonempty subset of V. An edge cut of a graph G is a subset of E(G)
of the form [S, ].

Bond: A minimal nonempty edge cut is called a Bond.

Note that A minimal edge cut is not always minimum. But a minimum edged cut is minimal.

Minimum edge cut Minimal edge cut Graph G Total edge cut

Each cut edge gives rise to a bond {e}.

Complement of a subgraph:

Let H be a subgraph of G, Then the complement of H in G is the subgraph G – E(H). It is

denoted by (G).

Cotree of G: Let G be a connected graph. A subgraph of the form , where T is a spanning

tree, is called a cotree of G.

Theorem 2.6: Let T be a spanning tree of a connected graph G and let ‘e’ be any edge of T.
Then
(i) The cotree contains no bond of G.
(ii) + e contains a unique bond of G.

Proof: Let B be a bond of a graph G. Then G –B is disconnected. G - B T. G – T.

. That is the bond B is not contained in any cotree. In other words, the cotree
contains no bond of G.

(iii) T is a spanning tree of a connected graph G and ‘e’ is any edge of T. Since T is a
tree T – e is disconnected and it has two components. Let S be the vertex set of
one of the components.
 8

The edge set [S, ] is a minimal edge cut and therefore a bond B of G and is contained in +e.

B +e, B G – E(T) +e, B+E(T) -e G. This implies T –e+b G. That is T –

e+b is a spanning tree of G for every b B. Therefore every bond of G contained in +e
must include every such element b. It follows that B is the only bond of G contained in +e.

Section 2.3 Cut vertices

Cut vertex: A vertex ‘v’ is said to be a cut vertex of G is the edge E(G) set is partitioned in to
two sets E1(G) and E2(G) such that G[E1] and G[E2] have just the vertex v in common.

If G is nontrival and loopless then v is a cut vertex of G if and only of ω(G-v) >ω(G).

Theorem 2.7: A vertex v of a tree G is a cut vertex of G iff d(v) > 1.

Proof: If d(v) = 0, then, then G K1. In this case v is not a cut vertex.
If d(v) = 1, then G – v is an acyclic graph with (G- v) - 1 edges. This implies G – v is a
tree. Therefore ω(G-v) = 1 = ω(G). Hence v is not a cut vertex.
 9

If d(v) > 1, then there are two distinct vertices u and w adjacent to v. Since G is a tree there
is a unique path uvw in G . Therefore there is no other path (u, w) in G -v. This implies
ω(G-v) >1 = ω(G). Hence v is a cut vertex.

Corollary:

Every nontrivial loopless connected graph has at least two vertices that are not cut vertices.

Proof:

Let G be a nontrivial loopless connected graph. We know that every connected graph
contains a spanning tree. Hence G has a spanning tree T. We know that Every nontrivial tree
has at least two vertices of degree one . Also, a vertex v of a tree G is a cut vertex of G iff d(v) >
1. Hence T has at lest two vertices that are not cut vertices. Let v be such a vertex. ω( T - v) = 1.

T is a spanning subgraph of G. Hence T – v is a spanning subgraph of G – v.

therefore ω(G-v) ≤ ω( T - v) = 1(since more connectivity implies less components). It
follows that ω(G-v) = 1. Hence v is not a cut vertex of G. Since there are two such vertices
G has at least two vertices that are not cut vertices.

Chapter 3

Connectivity
A vertex cut of G is a subset V’ of V such that G – V is disconnected.

A k – vertex cut is vertex cut of k elements.

A complete graph has no vertex cut.

Note that the only graphs which do not have vertex cuts are complete graphs and graphs
which have complete graphs as spanning sub graphs.

Connectivity:

Let a graph G has at least one pair of distinct non adjacent vertices. The connectivity of G,
denoted by κ (G) is the minimum k for which G has a k – vertex cut.

Graph with 1 – vertex cut.

 10

Graph with 2- vertex cut.

Graph with 3- vertex cut.

Otherwise κ (G) = - 1.

κ (G) = 0 if G is trivial or disconnected.

G is said to be k – connected if κ (G)≥ k.

All nontrivial graphs are 1 – connected.

A k –edge cut is a edge cut of k elements,

1- Edge connected

2-edge connected

3 –edge connected

If G is nontrivial, E’ is an edge cut, then G – E’ is disconnected.

Edge connectivity κ’ (G) of G is the minimum k for which G has a k-edge cut.
 11

When G is trivial or disconnected, κ’ (G) =0.

When G is connected with a cut edge κ’ (G) = 1.

k- edge connected: A graph G is said to be k-edge connected if κ’(G)≥ k.

All nontrivial graphs are 1– edge connected.

Theorem 3.1: In a graph G, κ ≤ κ’ ≤ δ.

Proof: First we prove that, κ’ ≤ δ.

If G is trivial then κ’ =0≤ δ. Otherwise, let ‘v’ be a vertex of degree δ. The δ edges of ‘v’
form a δ –edge cut. So at the most κ’ will be δ. This implies κ’ ≤ δ.

To prove κ ≤ κ’

We prove this inequality by induction on κ’, the edge connectivity.

When κ’ = 0, G is either trivial or disconnected. We have κ = 0. So the theorem is

true.

When κ’ = 1, G is K2 or it has a cut edge and hence a cut vertex and so κ = 1. So the
theorem is true.

Now let us assume the theorem is for all κ’ ≤ k -1.

Let us prove the theorem for κ’ = k also.

Let G be a graph with edge connectivity κ’(G) = k. Let ‘e’ be an edge in a k-edge
cut of G.

Let H = G – e. Then κ’(H) = k -1. By assumption κ(H) ≤ κ’(H) = k -1.

 12

We have two cases:

Case 1: When H has a complete graph as a spanning subgraph.

(H does not have a vertex cut)
Case 2: When H does not have a complete graph as a spanning subgraph.
(H has a vertex cut)

Case 1: When H has a complete graph as a spanning subgraph.

(H does not have a vertex cut)

κ(G) = κ(H) ≤ k -1 < k = κ’(G). Hence the theorem.

Case 2: When H does not have a complete graph as a spanning subgraph.

(H has a vertex cut)

Let S be a κ(H)- vertex cut of H. This imples H –S is disconnected. Now G – S may be

connected or not. We have two cases.

Case 2.1: When G – S is not connected and Case 2.2: When G – S is connected.

Case 2.1: When G – S is not connected

In this case κ(G) = κ(H) ≤ k -1 < k = κ’(G). Hence the theorem.

Case 2.2: When G – S is connected.

Let ‘e’ be a cut edge of G – S. In this Case we have two subcases.

Case 2.2.1: When ν(G – S) =2 and Case 2.2.2: When ν(G – S)>2.

Case 2.2.1: When ν(G – S) =2.

κ(G) ≤ ν(G) –1(by definition of connectivity)

≤ ν(G-S+S) –1 = ν(G-S) + ν(S) – 1
 13

= 2+ κ(H) -1 = κ(H)+ 1≤ (k -1) +1 = k = κ’(G). Hence the theorem.

Case 2.2.2: When ν(G – S)>2. In this case

G-S has a 1-vertex cut {v}.
This implies G has a vertex cut {v} + S
κ(G) = |{v} + S| = 1 +|S| = 1+ κ(H) ≤1 + (k -1 ) = k = κ’(G). Hence the theorem.

3.2 Blocks

Definition: A connected graph that has no cut vertices is called a block.

Example: Cn

Every block with at least two vertices is 2 – connected.

A block of a graph is a subgraph that is a block and is maximal with respect to this property.

Every graph is a union of its blocks.

Definition: Internally disjoint paths:

A family of paths in G is said to be internally disjoint paths if no vertex of G is an internal

vertex of more than one path of the family.

Whitney Theorem

A graph G ν ≥ 3 is 2 – connected if and only if any two vertices of G are connected by at

least two internally-disjoint paths.

Proof: if any two vertices of G are connected by at least two internally-disjoint paths
then G is connected and has no 1 – vertex cut. This means that it is 2 – connected.

Conversely, let G be 2 – connected graph. We prove this converse part by induction

hypothesis on the distance between u and v, d(u,v).

Suppose when d(u,v) = 1. Since G is 2 – connected, the edge uv is not a cut edge. We
know that an edge e is a cut edge iff e is not contained in any cycle of G. Here the edge uv is
 14

contained in some cycle of G. it follows that u and v are connected by two internally disjoint
paths. One is the (u,v) path and the other one is the path uPv.

Assume that the theorem is true for any two vertices at distance less than k and let d(u,v)
=k≥2.

Consider a (u,v) path of length k. Let w be the vertex that precedes v on that path. Now
d(u,w) = k -1. Hence by assumption u and w are connected by two internally disjoint paths P
and Q in G.

Since G is 2 – connected, G –w is still connected and there is a path (u,v) path P’ in G –w.
Let x be the last vertex in P’ which is also in PUQ. Note that x can be v also.

Let us assume that x is P. Then G has two internally disjoint paths, one is the section of P
from u to x together with the section P’ from x to v and the other path is Q with the path wv.

Corollary 3.2.1

If G is 2 – connected, then any two vertices of G lie on a common cycle.

By Whitney’s theorem, a graph G is 2-connected if and only if any two vertices are
connected by two internally disjoint paths. That is a graph G is 2 – connected, then any two
vertices of G lie on a common cycle.

Definition: An edge e is subdivided when it is deleted and replaced by a path of

length two connecting the ends, the internal vertex of this path being a new vertex.

Corollary 3.2.2: A G is a block ν ≥ 3 then any two edges of G lie on a common cycle.
Proof: Let G be a block with ν ≥ 3 and let e1, e2 be two edges of G. Form a new
graph G’ by subdividing the edges e1 and e2 and denote the new vertices by v1 and v2. Now
G’ is a block with at least 5 vertices and hence it is 2-connected. From the previous corollary,
v 1 and v2 lie on a common cycle. Thus e1 and e1 lie on a common cycle.

Menger’s theorem: Generalisation of Whitney theorem

 15

A graph with ν ≥ k +1 is k- connected if and only if any two distinct vertices are connected
by at least k – internally disjoint paths.

Edge analogue theorem: A graph with ν ≥ k +1 is k- edge connected if and only if any two
distinct vertices are connected by at least k edge – internally disjoint paths.

Chapter 4 Euler tours and Hamilton cycles

Euler trail: A trail that traverses every edge of a graph is called as an Euler trail.

Tour: A closed walk which traverses each edge of a graph at least once is called a
tour.

Euler Tour: This is a tour which traverses every edge of G exactly once.

Eulerian graph: A graph G is Eulerian if it contains a Euler tour.

Theorem 4.1: A nonempty connected graph is Eulerian if and only it contains no vertices of
odd degree.

Proof: Let G be Eulerian. Let C be an Euler tour in G with origin and terminus ‘u’. When
each vertex v appears as an internal vertex in the Euler tour, two edges will be taken into
account, one is incoming edge and the other one is outgoing edge. Hence d(v) is even for all
the internal vertices. For the origin ‘u’ also, degree is even it is the starting and end point of
the tour C. Hence the degree of u is also even. Therefore it contains no vertices of odd
degree.

Conversely, suppose that G is non Eulerian graph:

Let G be a connected graph with at least one edge and no vertices of odd degree. Choose
such a graph with as few edges as possible. Since the minimum degree of each vertex is two,
G contains a closed trail. Let C be a closed trail of maximum possible length.

Since G is non Eulerian, C is not an Euler tour. So G – E(C) has some component G’ with
some edges, that is Ɛ(G’)> 0. Now C is Eulerian and hence it has no vertices of odd degree.
Therefore the connected graph G’ also has no vertices of odd degree. Since Ɛ(G’) < Ɛ(G), G’
has an euler Tour C’. Since G is a connected graph there is a vertex v in common to both
V(C) and V(C’). Without loss of generality we may assume that v is the origin and terminus
of both C and C’. But then CC’ is a closed trail with the property that Ɛ(CC’) > Ɛ(C). This
is a contradiction to our assumption that C is a closed trail of maximum possible length.

Hence the theorem.

 16

Corollary 4.1:

A connected graph has an Euler trail if and only if it has at most two vertices of odd degree.

Proof: Let G be connected graph. We know that G is Eulerian if and only it contains no
vertices of odd degree. Hence if G has an Euler trail except the origin and terminus all other
vertices are of even degree. In other words G has only two vertices, the origin and the
terminus, of odd degree.

Conversely, suppose that G is a nontrivial connected graph with at most two vertices of odd
degree.

Case 1: If G has no such vertices. Then the degree of all the vertices are even. Hence G is
eulerian. Obviously it has a Euler trail,

Case 2: If G has one vertex of odd degree. This Case is not at all possible. Because we
know that by a theorem, in a graph the no. of vertices of odd degree is even.

Case 3: If G has exactly two vertices of odd degree. Let these vertices be ‘v’ and ‘u’. Join u
and v by an edge ‘e’. Now G +e is graph in which all the vertices are of even degree. Hence
G +e is Eulerian. Hence it has a closed Euler trail C: v0 e1 v1, . . ., ek+1vk+1 v0
where e1 = e. The trail v1, . . ., ek+1vk+1 is a Euler trail.

4.2 Hamilton cycles

Hamilton path: A path which contains every vertex of a graph G is called a Hamilton path.

Hamilton cycle: A Hamilton cycle of a graph G is cycle that every vertex of G.

Hamiltonian graph: A graph which contains a Hamilton cycle is called a Hamiltonian

graph.

Example: The dodecahedron is Hamiltonian. The Hershel graph is non Hamiltonian.

The necessary condition for a graph to be Hamiltonian is given below.

Theorem 4.2: If G is Hamiltonian, then for every nonempty proper subset S of V

ω(G-S) ≤ │S│.

Proof: Let C be a Hamilton cycle of G. Then for every nonempty proper subset S of V
ω(C-S) ≤ │S│. Because if S consists of a single vertex, ω(C-S) =0 since C is a cycle. When
S contains two vertices, ω(C-S) ≤1 and it goes on. Hence ω(C-S) ≤ │S│.
 17

C is spanning subgraph of G. Therefore C- S is spanning subgraph of G -S. ω(G-S) ≤ ω(C-S)

(since more connectivity in G –S). Hence ω(G-S) ≤ │S│.

Theorem 4.3: If G is a simple graph with ν ≥ 3, δ ≥ ν/2, then G is Hamiltonian,

Proof: We prove the theorem by contradiction. Suppose the theorem is false. Let G be a
maximal nonhamiltonian simple graph with ν ≥ 3, δ ≥ ν/2.

Claim: Since ν ≥ 3, G cannot be complete. If G is complete then ν -1= δ≥ ν/2. This implies
2(ν -1) ≥ ν. This in turn implies ν ≥ 2 which contradicts the fact that ν ≥ 3, So G cannot be
complete.

Let u and v be two nonadjacent vertices in G. By the choice of G, G +uv is Hamiltonian.

Since G is non Hamiltonian each cycle of G +uv should contain the edge uv. Thus there is a
Hamilton path v1, v2, . . ., vν in G with orign u = v1 and v = vν

Set S = { vi / u vi +1 E} and T = { vi / vi v E}

Claim: vν SUT

If vν S then u vν +1 E. But there is no such vν +1. If vν T then If vν vν E. This implies

the edge is a loop which is not possible in a simple graph. Hence vν S U T. │SUT│< ν.

Claim: │S ꓵ T│= 0

If S ꓵ T contains some vertex vi then v1, v2, v3, ... vi, vν vν-1 . . . vi+1 v1 is a Hamiltonian cycle
which is a contradiction to our assumption. Hence │S ꓵ T│= 0.

Now, d(u) + d(v) = │S│+ │T│ = │S ꓵ T│+ │S U T│ < ν +0. But d(u) + d(v) ≥ δ+ δ. This
implies 2 δ ≤ d(u) + d(v) < ν. This gives δ< ν/2 which contradicts our hypothesis that δ ≥ ν/2.
Hence G must be a Hamiltonian graph.
 18

Lemma 4.4.1: Let G be a simple graph and let u and v be nonadjacent vertices in G such
that d(u) + d(v) ≥ ν. Then G is Hamiltonian iff G +uv is Hamiltonian.

Proof:
If G is Hamiltonian then obviously G +uv is Hamiltonian.
Conversely suppose G +uv is Hamiltonian and G is not. Then by a theorem(If G is a simple
graph with ν ≥ 3, δ ≥ ν/2, then G is Hamiltonian) we get d(u) + d(v) < ν. But this is a
contradiction to our hypothesis that d(u) + d(v) ≥ ν. Hence when G +uv is Hamiltonian, G is
also Hamiltonian.

Definition

Closure of a graph G: The closure of G is the graph obtained by joining recursively joining
pairs of nonadjacent vertices whose degree sum is at least ν until no such pair remains. We
denote the closure of G by c(G).

Lemma 4.4.2: The closure of G c(G) is well defined.

Proof:

Let G1 and G 2 be two graphs obtained from G by recursively joining pairs of nonadjacent
vertices whose degree sum is at least ν until no such pair remains. Let e1, e2, . . . em be the
edges of G1. Let f1, f2, . . . fn be the edges of G2. Let us prove that each e i is an edge of G2
and each f j is an edge of G1.

Let ek+1 = uv be the first edge in the sequence e1, e2, . . . em that is not an edge in G2.

Set H = G + {e1, e2, . . . ek}. It follows that from the definition of G1 that dH(u) + dH (v) ≥ ν,
because ek+1 =uv is the next edge to be formed. Also by the choice of ek+1 H is a subgraph
of G2 also. Therefore

dG2 (u) + d G2 (v) ≥ ν. This implies ek+1 =uv is an edge of G2. But it is not. This contradiction
proves that all the edges of G1 are the edges of G2. Similarly we can prove that each end
every edge of G2 is an edge of G1.

Hence the lemma.

Theorem 4.3: A simple graph is Hamiltonian if and only if its closure is Hamiltonian.

We know that by a theorem (Let G be a simple graph and let u and v be nonadjacent vertices
in G such that d(u) + d(v) ≥ ν. A graph G is Hamiltonian iff G +uv is Hamiltonian) if G is
Hamiltonian then G+uv is Hamiltonian. Hence when we add each edge uv the new graph is
Hamiltonian. That is, let G1 = G+uv. Now G1 is Hamiltonian. Let G2 = G1+u1v1. By the
above mentioned theorem , G2 is Hamiltonian. Let G3 = G2+u2v2. Now G3 is Hamiltonian.
Thus at the end the closure will become Hamiltonian.
 19

Note: All complete graphs with ν ≥ 3 are Hamiltonian.

Corollary 4.4: Let G be a simple graph with ν ≥ 3. If c(G) is complete, then G is Hamiltonian.

Proof: We know that by a theorem a simple graph is Hamiltonian if and only if its closure is
Hamiltonian. Also all complete graphs are Hamiltonian. Hence c(G) is Hamiltonian . this in
turn says that G is Hamiltonian.

The sufficient condition for a graph to be Hamiltonian is given below.

Theorem 4.5:

Let G be a simple graph with degree sequence (d1, d2, . . .dν) where d1≤ d2≤ . . .≤ dν
and ν ≥ 3. Suppose that there is no value of m less than ν /2, for which dm ≤ m and d ν-m< ν-
m. Then G is Hamiltonian.

Proof: Let G satisfies the hypothesis of the theorem. We know that if c(G) is
complete then G is Hamiltonian. So let us prove that c(G) is complete. Let us denote the
degree of a vertex v in c(G) as d’(v).

Suppose c(G) is not complete. Let u and v be two nonadjacent vertices in c(G) so
that d’(u) ≤ d’(v). ---------------------------------------(1)

Also assume that d’(u) + d’(v) is as large as possible. Since u and v are nonadjacent
vertices d’(u) + d’(v) < ν ------------------(2) ( If the degree sum is ν then uv is an edge)

Let S = {x V/{v}/ x is not adjacent to v in c(G)}. That is S is the set of all vertices
excluding v and vertices adjacent to v. Therefore │S│= ν -1-d’(v)

T= {y V/{u}/ y is not adjacent to u in c(G)}, That is T is the set of all vertices

excluding u and vertices adjacent to u. Therefore │T│= ν -1-d’(u)

S does not contain the vertex v and also because of (1), the degree of all the vertices
of S is at the most d’(u). The degree of each vertex in T is at most d’(v) and the degree of u
is also at most d’(v). Hence TU {u} has degree at most d’(v).

S = u, v1, v2, v5, v8,

T = v, v3, ,v10,v1, v6, v7, v4, v5, v9
 20

Set d’(u)=m. Now c(G) has at least m vertices of degree atmost m(Since d’(u) = m)
and at least ν –m vertices of degree less than ν –m (since by inequality (2) d’(v) < ν –
d’(u)).

The closure c(G) is the spanning subgraph of G. Hence the same is true of G also.
Therefore dm ≤ m and d ν –m < ν –m.

From (2) d’(u) + d’(v) < ν. ν > d’(v) + d’(u)

≥ d’(u) + d’(u) ( from (1))
≥m+m
≥2m
ν/2. This contradicts the hypothesis that there is no value of m less than ν /2,
for which dm ≤ m and d ν-m< ν-m. Hence G must be Hamiltonian.

Definition: A sequence of real numbers (p1,p2, . . pn) is said to be majorised by

another sequence of real numbers ( q1,q2, . .qn) if pi ≤ qi for 1 ≤ i ≤ n.

Definition: A graph G is degree majorised by a graph H if ν(G) = ν(H) and that non
decreasing degree sequence of G is majorised by that of H.

For example C5 is degree majorised by K2,3.

Definition : The join of two disjoint graphs G and H is obtained from G+H by
joining each vertex of G to each vertex of H.

The graph Cm,n: The join of Km and Kmc + Kn -2m, 1≤ m< n/2 is denoted as the graph
Cm,n.

Cm,n = Km ( Kmc + Kn -2m)

C2,5 C1,5

Note: Cm,n is non Hamiltonian.

Let S ={ v V/ d(v) = n-1} and │S│ = m. Now ω(Cm,n - S)= m +1 > m = │S│.
This is a contradiction to the hypothesis given in the necessary condition of a graph G to be a
Hamiltonian graph. Hence Cm,n is non Hamiltonian.

Theorem: 4.6 If G is a non Hamiltonian simple graph with ν ≥ 3, then G is degree-

majorised by some Cm, ν.
 21

Proof: Let G be a non Hamiltonian simple graph with ν ≥ 3. Let (d1, d2, . . ., d ν) be a
degree sequence of G with d1 ≤ d2 ≤ , . . . , ≤ d ν. We know that by a theorem(Let G be a
simple graph with degree sequence (d1, d2, . . .dν) where d1≤ d2≤ . . .≤ dν and ν ≥ 3. Suppose
that there is no value of m less than ν /2, for which dm ≤ m and d ν-m< ν-m. Then G is
Hamiltonian) there exists m less than ν /2, for which dm ≤ m and d ν-m< ν-m. Therefore (d1,
d2, . . ., d ν) is degree majorised by (m,m, . . ., m, ν-m -1, ν-m -1, . . ., ν-m -1, ν -1, ν -1, . . ., ν
-1) with m terms equal to m and ν-2m terms equal to ν-m -1 and m terms equal to ν -1.
[ν = m+ ν-2m + m]
Now the latter sequence is the degree sequence of Cm, ν.

Corollary: 4.6 (Ore and Bondy theorem) If G is a simple graph with ν ≥ 3, and Ɛ >
+ 1, then G is Hamiltonian. Moreover the only non Hamiltonian simple graphs with ν
vertices and + 1 edges are C1, ν and for ν = 5, C2, 5.

Proof: If G is a non Hamiltonian simple graph with ν ≥ 3, then G is degree-majorised

by some Cm, ν for some positive integer for some m less than ν /2. Therefore Ɛ(G)≤ Ɛ(Cm, ν).
We know that h t h . Hence
h t h t m t m t t
Ɛ(Cm, ν) = =
t m t m t t
=
Hence Ɛ(G)≤ + 1. --------------------------(1)

Equality holds in (1) if the degree sequences of G and Cm, ν are the same. Further
more equality will hold in(1) only if wither m =2, ν = 5 or m =1. Hence the the only non
Hamiltonian simple graphs with ν vertices and + 1 edges are C1, ν and for ν = 5, C2, 5.

Chapter 5 Matchings

Definition: Matching: A subset M of the edge set of a graph is called a matching in

G if the edges are links and none of them are adjacent. The two ends of the edge are said to
be matched under M.

If an edge of M is incident with a vertex v, we can say M saturates the vertex v and
the vertex v is said to be M- saturated. Otherwise v is M – unsaturated.

Definition : Perfect Matching: If every vertex v of a graph G is M – saturated, then

the matching is a perfect Matching.
 22

Definition : Maximal Matching: M is a maximum matching of a graph G if G has

no matching M’ with the property │M’│ > │M│.

Definition : Alternating path in G:

Let M be a matching in G. A path in which the edges are alternately in E/M and M is
called as an M - alternating path in G.

Definition : Augmenting path in G: If the origin and terminus of an M -

alternating path in G are M- unsaturated, then it is called M - augmenting path in G.

Theorem 5.1 Berge Theorem: A matching M in G is a maximum matching in G iff

G contains no M – augmenting path.

Proof: Let M be a matching in G and suppose that G contains an M – augmenting

path v0 v1 v2 . . . , v2m+1. Define M’ E by

M’ = (M\{ v1v2, v3v4, . . . v2m-1 v2m}U {v0v1 ,v2 v3, v4 v5, . . . v2m v2m+1}

Now M’ is a matching in G and │M’│=│M│+ 1. This implies that M is not

maximum. Thus if G contains an M – augmenting path then G cannot be maximum.

Conversely, suppose that M is not a maximum matching. Let M’ be a maximum

matching. Obvious that │M’│>│M│.
 23

M’ = { v0v1 , v2 v3, v4 v5, v6 v7, v11 v12} Green

M = { v1v2 , v3v4, v5 v6, v9 v13} Red │M’│ = 5, │M│ =4

Set H = M’ ΔM (the symmetric difference between M and M’)

= (M- M’)U (M’- M)

Each vertex of H has degree one or two since it can be incident with at most end edge
of M and one edge of M’. Thus each component of H is either an even cycle with edges
alternately in M and M’or else a path with edges alternately in M and M’. It is obvious that
the no. of M’ edges is more than that of M. Hence a path P in some component of H must
start and end with M’ edges . This is a path with M unsaturated origin and terminus vertices.
Hence this an M – augmenting path in G.

Section 5.2 Matchings and Coverings in Bipartite graphs

Definition: Neighbour set. Let S be a subset of the vertex set V of G. Then the
neighbour set of S is the set of all the vertices adjacent to the vertices in S.
 24

Suppose G is a bipartite graph with bipartition (X, Y). The necessary and sufficient
conditions of finding a matching of G that saturates every vertex of G is given bu Hall’s in
the following theore.

Theorem: Hall’s theorem: Let G be a bipartite graph with bipartition (X, Y). Then
G contains a matching that saturates every vertex in X iff │N(S)│≥│ S│ for all S

Proof: Suppose that G contains a matching that saturates every vertex in X. Let S
be a subset of X. The vertices in S are matched under M with distinct vertices of N(S).
Hence clearly │N(S)│≥│ S│.

Converse part: Let G be a bipartite graph satisfying the condition │N(S)│≥│ S│ for
all S Suppose that G does not contain any matching that saturates every vertex in X.

Let M* be a maximum matching in G. By our supposition, M* does not saturate all

the vertices of X. Let u be an M*- unsaturated vertex in X. Let Z be the set of all vertices
that are connected to u by M* alternating paths. Since M* is a maximum matching u is the
only one unsaturated vertex in Z. Set S = Z ꓵ X, T = = Z ꓵ Y. Clearly the vertices in S/{u}
are matched under M* with the vertices of T. In fact │T│ =│S│ -1. Also N(S) T. In fact
N(S) T.
Therefore │N(S)│ = │T│ =│S│ -1 < │ S│. This is a contradiction to the hypothesis
that │N(S)│≥│ S│. Therefore our supposition that M* does not saturate all the vertices of X
is wrong.
Hence if G if a bipartite graph satisfying the condition │N(S)│≥│ S│ for all S ,
then G contains some matching that saturates every vertex in X.

Corollary 5.2: If G is a k-regular bipartite graph with k> 0, then G has a perfect
matching.

Proof: Let G be a k-regular bipartite graph with bipartition (X,Y). Since G is k –

regular, k│X│ = │E│ = k│Y│.

k│X│ = k│Y│ │X│ = │Y│since k> 0. Now let S be a subset of X.

Let E1= { ei / ei is incident with vertices of S}, │ E1│= k │ S│

E2= { ej / ej is incident with vertices of N(S)}, │ E2│= k │N( S)│

By definition of N(S), E1 E2. │ E1│ ≤ │ E2│

k │ S│ ≤ k │N( S)│

│ S│ ≤ │N( S)│ By Hall’s theorem, G contains some matching that

saturates every v ertex in X. Since │X│ = │Y│, the other end of the matching saturates
every vertex in Y. This shows that the matching is a perfect one.
 25

Note: Corollary 5.2 is knows as Marriage theorem. We can state this as follows. If
every girl in a village knows exactly ‘k’ girls and every boy knows exactly ‘k’ boys, then
each girl can marry a boy she knows and each boy can marry a girl he knows.

Definition: Covering: A subset K of V is said to be a covering of G if every edges of

G has at least one end in K.

Minimum covering of a graph G is a covering k so that there is no covering K’ such

that │K’│ < │K│.

If K is a covering and M is a matching , then K contains at least one end of a

matching. Therefore │M│≤ │K│. ( Because K has to cover all the edges of G). If M* is a
maximum matching and is a minimum covering then │M*│≤ │ │ -------------(1)

Equality does not hold in (1) except when G is bipartite graph(Konig’s theorem).

Lemma 5.3: Let M be a matching and K be a covering such that │M│= │K│.
Then M is a maximum matching and K is a minimum covering.

Proof: If M* is a maximum matching and is a minimum covering then │M*│≤

│ │ . But │M│ ≤ │M*│ and │ │ ≤ │K│. Therefore

│M│ ≤ │M*│ ≤ │ │ ≤ │K│. Since │M│= │K│, we have │M│= │M*│ and
│ │= │K│ which implies M is a maximum matching and K is a minimum covering.
 26

Theorem 5.3: Konig’s theorem: In a bipartite graph, the no edges in a maximum

matching is equal to the number of vertices in a minimum covering.
Proof: Let G be a bipartite graph with bipartition (X,Y). Let M * be a maximum
matching of G. Set U = {uj / uj are unsaturated vertices in X}. Z = {vi / vi are connected to uj
by M* alternating paths.
Let S = Z , T= Z . Now every vertex in T is M* saturated and N(S) = T.
Define = (X/S)U T. Every edge of G has one of its ends in . Otherwise there would be
one end in S and other end in Y/T contradicting that N(S) = T. Hence a covering
and│M*│=│ │. By a known result(lemma 5.3 Let M be a matching and K be a covering
such that │M│= │K│. Then M is a maximum matching and K is a minimum covering) M*
is a maximum matching and is a minimum covering.

Chapter 6. Edge colourings

Definition: A k-edge colouring ℭ of a loopless graph is an assignment of k colours to

the edges of G.
Definition: Proper k-edge colouring: A k-edge colouring ℭ is said to be proper if
no two adjacent edges have the same colour.

Note: A k-edge colouring is a partition (E1, E2. ... Ek) of E where Ei denotes the
subset of E having the colour i. A proper k-edge colouring is a k-edge colouring (E1, E2. ...
Ek) in which each Ei is a matching.
 27

figure 6.1
This graph has a proper 4 –edge colouring.

Definition: k-edge colourable: If a graph G has a proper k –edge colouring then G is

said to be k-edge colourable.

Every loopless graph is Ɛ-edge colourable.

Definition: The edge chromatic number: This is denoted by χ’(G). The edge
chromatic number is the smallest k for which G is k –edge colourable. If G is k –edge
chromatic then χ’(G) = k. Figure 6.1 depicts a 4 –edge chromatic graph.

Note: In any proper edge colouring, χ’(G)≥ Δ.

When G is bipartite χ’(G) = Δ.
Note : We say that a colour ‘i’ is represented at a vertex ‘v’ if some edge with
colour ‘i; is incident with it.

Lemma 6.1.1. Let G be a connected graph that is not an odd cycle. Then G has a 2-
edge colouring in which both colours are represented at each vertex of degree at least two.
Proof : Let us assume that G is nontrivial. We prove the lemma in two cases, when
G is Eulerian and when G is non Eulerian.
Case 1: when G is Eulerian.
When G is Euleian, all vertices are of even degree. Here we have two cases.
Case 1.1: when G is an even cycle. The proper 2-edge colouring satisfies the
theorem.

Case 1.2: when G is an not even cycle. Let v0 be an arbitrary vertex so that d(v0) ≥ 4.
Let v0 e1 v1 e2 v2 , . . . , v0 be an Euler tour of G. Let E1 = { ei | i odd} and E2 =
{ ei | i even}. The edges of E1 will be given one colour. The edges of E2 will be given a
colour different to E1. Now each vertex of G is an internal vertex of the Euler’s tour.
Therefore the 2- edge colouring (E1 ,E2) satisfies the theorem.
 28

Case 2: when G is non Eulerian.

Let G be a graph which is not Eulerian. In this case, not all the vertices of G are of
even degree. We know that in any graph the number of vertices of odd degree is even. Let
v0 be a new vertex and set G* = (vertices of odd degree in G) ˅ v0 . Now G* is Eulerain (By
joining the vertices of odd degree with v0, those vertices degree will become even. Also d(v0)
is even).
Let v0 e1 v1 e2 v2 , . . . , v0 be an Euler tour of G*. Let E1 = { ei | i odd} and
E2 = { ei | i even}.

The edges of E1 will be given one colour. The edges of E2 will be given a
colour different to E1. Now the 2- edge colouring (E1 ,E2 ) satisfies the theorem.
□

Definition : Improvement on edge colouring:

Let be a k-edge colouring of G. The number of distinct colours at v is denoted by

c(v). It is obvious that c(v) ≤ d(v).
is a proper k-edge colouring of G if and only if c(v) = d(v) for all v.
Let ’ be another k-edge colouring of G. Let the number of distinct colours at v is
denoted by c’(v). ’ is an improvement on if h
h h
th .
 29

An optimal k-edge colouring is one which cannot be improved.

Lemma 6.1.2: Let ₡ = {E1, E2, E3, . . ., Ek} be an optimal k-edge colouring of G.
If there is a vertex ‘u’ in G and colours ‘i’ and ‘j’ such that colour ‘i’ is not represented at u
and ‘j’ is represented at least twice at ‘u’ then the component of G[Ei U Ej] that contains ‘u’
is an odd cycle.

Proof: Let ‘u’ be a vertex of G that satisfies the hypothesis. Let H =

G[Ei U Ej] that contains ‘u’. We have to prove that H is an odd cycle.
Suppose H is not an odd cycle.
We know that if G is a connected graph that is not an odd cycle, then
G has a 2-edge colouring in which both colours are represented at each
vertex of degree at least two. When we recolour the edges of H in this way we
get a new colouring ₡’ = {E1’, E2’, E3’, . . ., Ek’} of G. Let c’(v) be the no .of
distinct colours at v in the colouring ₡’. At ‘u’ c’(u) = c(u) +1. Also

h
h h
th for all the vertices v u. Hence this gives a contradiction
 30

to the fact that ₡ is an optimal k-edge colouring. Therefore H must be an

odd cycle. □

Theorm 6.1: If G is bipartite then χ’ =Δ.

Proof: Let G be a graph with χ’ > Δ. Let

₡ = {E1, E2, E3, . . ., EΔ } be an optimal Δ -edge colouring of G. Let ‘u’ be an

vertex such that c(u) < d(u). Now, at the vertex ‘u’, one colour is represented twice and one
colour is not at all represented. It satisfies the lemma “Let ₡ = {E1, E2, E3, . . ., Ek} be an
optimal k-edge colouring of G. If there is a vertex ‘u’ in G and colours ‘i’ and ‘j’ such that
colour ‘i’ is represented at u and ‘j’ is represented at least twice at ‘u’ then the component of
G[Ei U Ej] that contains ‘u’ is an odd cycle”. Hence the component which contains those two
colours is an odd cycle. In that case the graph cannot be a bipartitie graph. Hence our
supposition that χ’ > Δ is not true. We must have χ’ =Δ in a bipartite graph.
 31

Theorem 6.2: Vizing theorem: If G is simple, then either χ’ = Δ or χ’ =

Δ+1
Proof by Fournier: If G is simple, then to prove χ’ ≤ Δ+1.
Suppose χ’ > Δ+1. Let ₡ = {E1, E2, E3, . . ., EΔ+1 } be an optimal Δ+1 -edge
colouring of G.

(In this graph χ’ = Δ+1= 5)

Let ‘u’ be an vertex such that c(u) < d(u). Now, at the vertex ‘u’, one colour ‘i1’is
represented twice and one colour ‘i0’ is not at all represented. Let uv1 have colour i1.
 32

The maximum degree is Δ. Hence for any vertex d(v) < Δ+1. Consider the vertex v1,
so that d(v1) < Δ+1. Because of this some colour ‘i2’ is not at represented at v1. But colour
‘i2’ should be represe\nted at u. Otherwise recolouring uv1 by i2 we get a improvement on
₡. Thus some edge uv2 has colour i2.

Again for some vertex v2, d(v2) < Δ+1. Because of this some colour ‘i3’ is not at
represented at v2. But colour ‘i3’ should be represented at u. Otherwise recolouring uv1 by
i2 and uv2 by i3 we get a improvement on ₡. Thus some edge uv3 has colour i3.
And so on, we construct a sequence of vertices v1, v2, . . . and a sequence of colours
i1, i2, . . . such that
i) uvj has colour ij
ii) ij+1 is not represented vj
iii) il+1 = ik for some k < l (since the d(u) is finite)
 33

Now for 0≤ j ≤ (k -1) we recolor the edges uvj with ij+1. This gives a new(Δ+1) –edge
colouring ₡’ = {E1’, E2’, E3’, . . ., EΔ+1’ }. Clearly c’(v) ≥ c(v) for all vertices of G. Also
₡’ is also an optimal (Δ+1) –edge colouring of G. By a lemma [Let ₡ = {E1, E2,
E3, . . ., Ek} be an optimal k-edge colouring of G. If there is a vertex ‘u’ in G and colours
‘i’ and ‘j’ such that colour ‘i’ is not represented at u and ‘j’ is represented at least twice at ‘u’
then the component of G[Ei U Ej] that contains ‘u’ is an odd cycle.], the component H’ =
G[Ei0’ U Eik’] that contains the vertex ‘u’ is an odd cycle.

In addition to this colouring let us recolour uvj with colour ij+1, for k ≤ j ≤ l -1 and
uvl with colour ik. This gives a (Δ+1) –edge colouring ₡” = {E1”, E2”, E3”, . . ., EΔ+1” }.
Also, c”(v) ≥ c(v) for all vertices of G. By the above mentioned lemma, the component H”=
G[Ei0” U Eik”] that contains the vertex ‘u’ is an odd cycle.

Note that in H’ , d(vk) = 2 whereas in H” , d(vk) = 1. This is not at all possible in the
same graph G. Hence our supposition that χ’ > Δ+1 is not true. Hence the theorem.
 34

Definition: Multiplicity of G. The maximum number of edges joining two vertices

in G is called the multiplicity of G. This is denoted by μ(G).

In the above figure μ(G) = 4.

Generalized from of Vizing theorem: If G is loopless, then Δ ≤ χ’≤ Δ + μ.

Note : The following is an example in which χ’ = Δ + μ. Here χ’ = 3 μ and Δ =2μ.

 35

Chapter 8 Vertex Colouring

8.1 Chromatic number

Definition: A k- vertex colouring is an assignment of k colours to the vertices of G.

Definiton: Proper k- vertex colouring: A k- vertex colouring is proper if adjacent
vertices have different colours.
Definiton: Independent set:
Let S be a subset of V. The Set S is called an independent set if no two vertices of S
are adjacent in G.

Definiton: Maximum Independent set:

The Set S is called a maximum independent set if there is no set S’ with the
property│S’│ > │S│.

Note: A Proper k- vertex colouring of a loopless graph G is a partition (V1, V2, . . .,

Vk) of V in to k independent sets.
Note: We say a k – vertex colouring as k – colouring.
 36

Definiton: k- vertex colourable: A graph G is k- vertex colourable if G has a proper

k – vertex colouring.
Note: We say a k – vertex colourable as k – colourable.
A simple graph 1- colourable if and only if it is empty.
A simple graph G is 2- colourable iff it is bipartite graph.

Definiton: Chromatic Number: If G is k- vertex colourable, then the minimum k is

known as the Chromatic Number of G. It is denoted by χ(G).

In the above graph χ(G) = 3.

Definiton: If χ(G) = k, then G is said to be k – chromatic. The above graph is a 3-

chroamtic graph.

Definiton: Critical graph: Let H be a proper subgraph of G. The graph G is said to a

critical graph if χ(H) < χ(G) for every proper subgraph H of G.

Definiton: k - Critical graph: The graph G is said to a k- critical graph if G is critical

and k – chromatic.

Note: Every k- chromatic graph has a k-critical subgraph.

 37

4-critical graph Grotzsch graph

Theorm 8.1: If G is k – critical then δ ≥ (k -1)

Proof: We prove the theorem by contradiction. Let G be a critical graph with δ < (k -
1). Let v be a vertex so that d(v) = δ. Since G is k –chromatic, χ(G) = k. Since G is critical,
χ(G-v) = k-1.
Let (V1, V2, . . . , Vk-1) be a (k-1) colouring of G-v. In G, d(v) = δ < (k -1).
Therefore in G v is adjacent to (k-1) vertices. In G, v is non adjacent to every vertex of
Some Vj. Give one colour which is different from the above (k-1) colours to v and all the
vertices of Vj. Then (V1, V2, . , Vj U { v}, . . , Vk-1) becomes (k-1) colouring of (G-v)+ v
= G. This is contradiction to the fact that G is k-colourable. Hence δ < (k -1) cannot be true.
Hence the theorem.
 38

Corollary 8.1.1: Every k- chromatic graph has at least k vertices of degree at least
k-1.

Proof: Let G be a k- chromatic graph. We know that Every k- chromatic graph has a
k-critical sub graph. Let H be a k-critical sub graph of G. By a theorem [If G is k – critical
then δ ≥ (k -1)], δ(H) ≥ (k -1). This means that H has a minimum of (k-1) + 1 vertces. This
is true in G also. Hence G has at least k vertices of degree at least k-1.

Corollary 8.1.2: For any graph G, χ ≤ Δ +1.

We know that for any graph Δ ≥ δ . Hence Δ +1 ≥ δ +1

≥ (k-1) +1 [since δ ≥ (k -1) ]
≥k
≥ χ ( since G is k chromatic χ =k)
Hence the result.

Defintion: The S-components of G:

Let S be a vertex cut of G and let the components of G – S have vertex sets V1,
V2, . . . , Vn. Then the induced sub graphs Gi = G[Vi U S] are called the S-components of G:

Note: We say that colourings of G1, G2. . . . , Gn agree on S if for every v S

vertex is assigned the same colour in each of the colourings.
 39

Definition: Clique: Let G be a simple graph. A subset S of V is said to be a clique

such that G[S] is complete.

Theorem 8.2: In a critical graph, no vertex cut is a clique.

Proof: Let G be a k- critical graph. Suppose that let S be a vertex cut that is a clique.
Denote the S-components of G by G1, G2. . . . , Gn. Since G is k-critical, each Gi is (k-1)
colourable. Since S is a clique, G[S] is complete. Therefore the vertices of S must receive
different colours in any (k-1) colouring of Gi. Hence there are (k-1) colourings of G1,
G2. . . . , Gn which agree on S. But these colouring together yield a (k-1) colouring of G
whereas G is k –chromatic. Therefore our supposition that the vertex cut S is a clique is not
true. Hence the theorem.

Corollary 8.2: Every critical graph is a block.

 40

Proof: If v is a cut vertex then {v} is a vertex cut also. It is a clique. We know that
in a critical graph, no vertex cut is a clique. Hence no critical graph has a vertex cut.
This means critical graphs are block.

Note: If a k-critical graph G has a 2-vertex cut {u,v}

Chapter 9 planar graphs

Section 9.1 Plane and planar graphs

Definition: Embeddable in the plane:
A graph is said to be embeddable in the plane, or planar, if it can be drawn in the
plane so that its edges intersect only at their ends.

Definition: Planar embedding : The drawing of a planar graph without intersection

of edges is known as planar embedding. The planar embedding of G is isomorphic
to the graph G.

Definition: Plane graph: The planar embedding of a planar graph is plane graph.

A graph which is not planar is nonplanar. The complete graph K5 is nonplanar.

Definition: Jordan Curve: A Jordan curve is a continuous non-self- intersecting

curve whose origin and terminus coincide.

The union of the edges in a cycle of a plane curve constitutes a Jordan curve.
 41

Definition: Let J be a Jordan curve in a plane. Then the rest of the plane is
partitioned into two disjoint open sets called interior and exterior of J. The interior of
j is denoted by int J and the exterior of J is denoted by ext J. The open sets along
wth their boundary are denoted by Int J and Ext J respectively. It is obvious that Int
J Ext J = J.

Jordan Curve theorem: Any line joining a point in int J to a point in ext J must
meet J in some point.

Theorem: K5 is non planar.

We prove the theorem by contradiction. Suppose K5 is planar. Let G be a planar

embedding of K5. Let v1, v2, v3, v4, v5 be the vertices of G. Let C = v1, v2, v4, v1 be a
Jordan curve.

Let v3 int C. The edges v3 v1, v3 v2, v3 v31 divide int C into three regions int C1, int
C2 ,int C3 where C1= v1, v2, v3, v1 , C2 = v2, v3, v4, v2 and C3 = v3, v4, v1, v3.

The vertex v5 should lie an any of the regions C1 ,C2 ,C3 or ext C. Suppose that
v5 ext C. Since the graph is complete, the vertices v3 and v5 are to be adjacent.
Therefore the edge must meet C in some point. By Jordan curve theorem that point
must be a vertex which is not true. Hence our supposition that G is a planar
embedding of K5 cannot hold good. This proves the theorem.

Definition: A graph is said to be embeddable on a surface S if it can be drawn

in S so that the edges intersects only at the ends. Such a drawing , if it exists, is called
a embedding of G on S.

The complete graphs K5 and K3,3 can be embeddable on surfaces torus and
Mobius band respectively.
 42

Note: There are graphs which cannot be embeddable on the surfaces also. But
all graphs are embeddable on 3D space 3.

Definition: Stereographic Projection: Consider a sphere S resting on a plane

P. Let z denote the point in the sphere which is diagonally opposite to the point of
contact of S and P. The mapping : S\{z}→P defined by (s) = p if and only if
the points z,s, p are collinear.

Theorem 9.2: A graph is embeddable in the plane if and only if it is

embeddable on the sphere.

Proof: Suppose G has an embedding on the sphere. Let z be a point on the

sphere and not on the embedding . The image of under stereographic
projection from z , : S\{z}→P defined by (s) = p, is an embedding of G on the
plane.
 43

Conversely, suppose that a graph is embeddable in the plane. By using the

inverse map : P → S\{z} defined by (s) , the image of G is
embeddable on the surface S.

Section 9.2: Dual graphs

Definition: Faces of a graph: A plane graph G partitions the rest of the plane into a
number of connected regions. The closures of these regions are called the faces of G. The
set of all faces is denoted by F(G) and the number of faces is denoted by φ(G). Each plane
graph exactly one unbounded face called the exterior face. In the following graph face f6 is
the exterior face.

F(G) = { f1, f2, f3, f4, f5 , f6} and φ(G) =6

Theorem 9.3: Let ‘v’ be a vertex of a planar graph G. Then G can be embedded in
the plane in such a way that ‘v’ is on the exterior.

Proof: Let G be a planar graph. Let be the embedding of G on plane. Then by a

theorem[A graph is embeddable in the plane if and only if it is embeddable on the sphere.]
is the embedding of G on the sphere. Let ‘z’ be a point in the interior of the face containing
the point ‘v’. Let t be the image of t ‫ݎ‬ ‫݋ݎ‬ ‫ݎ‬t from z.
Clearly, is a planar embedding of G in which the point ‘v’ is on the exterior.

Definition: Boundary of a face: It is denoted by b( f). Let G be a connected graph.

Then the boundary of a face f is a closed walk in which each cut edge of G in b( f) is
traversed twice. If b( f) contains no cut edges, then it is a cycle of G.
 44

Here,
b( f2) = v2 e1 v1 e5 v5 e6 v5 e5 v1 e4 v4 e3 v3 e2 v2

b( f3) = v2 e1 v1 e4 v4 e3 v3 e2 v2

b( f1) = v5 e6 v5

Note : A face f is said to be incident with the vertices and edges in its boundary.

If ‘e’ is a cut edge in a plane graph, just one face is incident with ‘e’. Otherwise there
are two faces are incident with it. We say that en edge separates the faces incident with it.
In the above figure edge ‘e3’ separates the faces f2 and f3.

Definition : Degree of a face is defined as the no. of edges it is incident with. It is

denoted as dG( f). For example, d(f2) = 7, d(f3) = 4, d(f1) = 1.

Definition : Dual of a graph G. Let G be a plane graph. The dual of G, denoted by

G* is defined as follows.
(i) The faces, ‘f’, of G are the vertices, v*, of G* .
(ii) Corresponding to each edge ‘e’ of G there is an edge ‘e*’ in G*.
(iii) Two vertices of G* are joined by an edge ‘e*’ if and only if the
corresponding faces are separated by an edge ‘e’ in G.

Note: The dual G* of a plane graph is planar.

If ‘e’ is a loop of G , then‘e*’ is a cut edge in G*.
If ‘e’ is a cut edge of G , then‘e*’ is a loop in G*.
The dual of dual is the original graph. In fact G** is isomorphic to G.
Note that ν(G*) = φ(G)
(G*) = (G)
dG*(f*) = dG(f) for all f in F(G).

Theorem 9.4: If G is a plane graph then .

Proof: Let G* be the dual of a plane graph G. Then

t
=2
=2 t
 45

Definition: A plane graph is self-dual if it is isomorphic to its dual.

Example:

Section 9.3 Euler’s formula

Euler’s formula is a simple formula relating the the number of vertices, edges and
faces in a connected plane graph.

Theorem 9,5: If G is connected plane graph then ν – Ɛ + φ = 2.

Proof: We prove the theorem by induction on the number of faces φ of G.