A study on connectivity in graph theory june 18 pdf

CONNECTIVITY OF
GRAPHS
ASWATHY.A

2
CONNECTIVITY IN GRAPHS
ASWATHY.A
DEPARTMENT OF MATHEMATICS
MOUNT TABOR TRAINING COLLEGE
PATHANAPURAM

3
CONTENTS
PAGE NO
PREFACE 3
INTRODUCTION 4
Chapter 1 CONNECTIVITY OF GRAPHS 6 
Chapter2 APPLICATIONS 27
REFERENCE 34

4
PREFACE
In the present times Mathematics occupies an important place in
curriculum. The “Connectivity in graphs” is one of the important subject at
MSc level. In this book an attempt has been made to cover up most of topics
included in the MSc syllabus by Indian Universities. This book covers the
syllabi of Regular and Correspondence courses of Indian Universities.
The book has been written in a simple and lucid manner and is
up-to-date in its contents. To illustrate theory some examples have been given.
It is hoped that the book will be appreciated by teachers and
students alike. While preparing the book, material has been drawn from works of
different authors, periodicals and journals and the author is indebted to all such
persons and their publishers.
All suggestions for improvement of the book shall be thankfully
accepted.
Author

5
INTRODUCTION
Graph Theory is regarded as one of the areas of Applied Mathematics. Graph Theory
has been independently discovered many times. Leonhard Euler is known as “Father of Graph
Theory”. Subsequent rediscoveries of Graph Theory has been made by Gustav Kirchhoff and
Arthur Cayley. Another approach was of Hamilton’s. The origin of graph theory can be traced
back to Euler's work on the Konigsberg bridges problem (1735), which subsequently led to the
concept of an Eulerian graph. The study of cycles on polyhedra by the Thomas P. Kirkman (1806
- 95) and William R. Hamilton (1805-65) led to the concept of a Hamiltonian graph. The concept
of a tree, a connected graph without cycles, appeared implicitly in the work of Gustav Kirchhoff
(1824-87), who employed graph-theoretical ideas in the calculation of currents in electrical
networks or circuits. Later, Arthur Cayley (1821-95), James J. Sylvester (1806-97), George
Polya (1887-1985), and others use 'tree' to enumerate chemical molecules. The study of planar
graphs originated in two recreational problems involving the complete graph 5K and the
complete bipartite graph 3,3K . These graphs proved to be planarity, as was subsequently
demonstrated by Kuratowski. First problem was presented by A. F. Mobius around the year
1840.
The development of graph theory is very similar the development of probability theory,
where much of the original work was motivated by efforts to understand games of chance. The
large portions of graph theory have been motivated by the study of games and recreational
mathematics. Generally speaking, we use graphs in two situations. Firstly, since a graph is a very
convenient and natural way of representing the relationships between objects we represent
objects by vertices and the relationship between them by lines. In many situations (problems)
such a pictorial representation may be all that is needed. Secondly, we take the graph as
mathematical model, solve the appropriate graph-theoretic problem, and then interpret the
solution in terms of the original problem.
In the past few years, Graph Theory has been established as an important mathematical tool
in wide variety of subjects. Graph Theory has application to some areas of Physics, Chemistry,
Communicative Science, Computer Technology, Electrical and Civil Engineering, Architecture,
Operation Research, Genetics, Psychology, Sociology, Economics, Anthropology and
Linguistics. The theory had also emerged as a worthwhile mathematical discipline in its own
right. Graph Theory is intimately related to many branches of Mathematics including Graph
Theory, Matrix Theory, Numerical Analysis, Probability and Topology. In fact Graph Theory
serves as a mathematical model for any system involving a binary relation.
In this text book, discuss the connectivity of graphs. This textbook is divided into two
chapters. In the first chapter, discussed about definitions and theorems on connectivity. The
second chapter deals with some applications of connectivity.

6
CHAPTER –1
CONNECTIVITY OF GRAPHS
Definition (2.1)
An edge of a graph 𝐺 is called a bridge or a cut edge if the subgraph 𝐺 − 𝑒 has
more connected components than 𝐺 has.
Figure (2.1)

7
Figure (2.2)
A graph and two bridge deletions
Theorem (2.1)
An edge 𝑒 of a graph 𝐺 is a bridge if and only if 𝑒 is not part of any cycle in 𝐺.
Proof:
Let 𝑒 have end vertices 𝑢 and 𝑣. If 𝑒 is not a bridge then, it is either a loop or
there is a path 𝑃 = 𝑢𝑢1 𝑢2 . . .𝑢n 𝑣, from 𝑢 to 𝑣, different from the edge 𝑒. If it is a
loop then it forms a cycle. If there is such a path 𝑃 then 𝐶 = 𝑢𝑢1 𝑢2…𝑢n 𝑣𝑢, the
contradiction of 𝑃 with , is a cycle in 𝐺.This shows that if 𝑒 is not a bridge then it
is part of a cycle. This is equivalent to saying that if 𝑒 is not part of any cycle then
𝑒 must be a bridge.

8
Conversely, suppose that e is part of some cycle 𝐶 = 𝑢0 𝑢1 𝑢2...𝑢m in 𝐺. Let 𝑒 =
𝑢i 𝑢i+1. In the case where 𝑚 = 1, 𝐶 = 𝑢0 𝑢1, and so 𝐶 is just the edge 𝑒 and 𝑒 is a
loop. On the other hand , 𝑖𝑓 𝑚 > 1 then 𝑃 = 𝑢i 𝑢i-1….𝑢0 𝑢m-1…𝑢i+1 is a path
from 𝑢 to 𝑣 different from 𝑒. Thus 𝑒 is not a bridge .This shows that if 𝑒 is a bridge
then it is not part of any cycle in 𝐺 , completing the proof.
Theorem (2.2)
Let 𝐺 be a connected graph .Then 𝐺 is a tree if and only if every edge of 𝐺 is a
bridge i.e., if and only if for every edge 𝑒 of 𝐺 the subgraph 𝐺 − 𝑒 has two
components.
Proof:-
Suppose that 𝐺 is a tree. Then 𝐺 is acyclic i.e., it has no cycle, and so no edge
of 𝐺 belongs to a cycle. In other words, if 𝑒 is any edge of 𝐺 then it is a bridge.
Conversely, suppose that 𝐺 is connected and that every edge 𝑒 of 𝐺 is a bridge.
Then 𝐺 can have no cycles since any edge belonging to a cycle is not a bridge.
Hence 𝐺 is acyclic and so is a tree.
Problem (2.1)
Find all bridges in the graph.

9
Figure (2.3)
Solution :-
Bridges in the graph are 𝑒2,𝑒3, 𝑒4, 𝑒5,𝑒6,𝑒7,𝑒13and 𝑒14.
Definition (2.2)
A vertex 𝑣 of a graph 𝐺 is called a cut vertex of 𝐺 𝑖𝑓  (𝐺 − 𝑣) >  (𝐺).In
other words , a vertex 𝑣 is a cut vertex of 𝐺 if its deletion disconnects some
connected components of 𝐺 , there by producing a subgraph having more
connected components than 𝐺 has.

10
Figure (2.4)
𝑉 is a cut vertex of 𝐺1,since  (𝐺1) = 1 while  1G −𝑣) = 3.The graph 𝐺2
has no cut vertices.
Theorem (2.3)
Let 𝑣 be a vertex of the connected graph 𝐺. Then 𝑣 is a cut vertex of 𝐺 if and
only if there are two vertices 𝑢 and 𝑤 of 𝐺, both different from 𝑣, such that 𝑣 is on
every 𝑢 − 𝑤 path in 𝐺.
Proof:-
First let 𝑣 be a cut vertex of 𝐺 .Then 𝐺 − 𝑣 is disconnected and so there are
vertices 𝑢 and 𝑤 of 𝐺 which lie in different components of 𝐺 – 𝑣 .Thus, although
there is a path in 𝐺 from 𝑢 and 𝑤, there is no such path in 𝐺 − 𝑣.This implies that
every path in 𝐺 from 𝑢 to 𝑤 contains the vertex 𝑣.
Conversely , suppose that 𝑢 and 𝑤 are two vertices of 𝐺 , different from 𝑣,
such that every path in 𝐺 from 𝑢 to 𝑤 contains 𝑣.Then there can be no path from 𝑢

11
to 𝑤 in 𝐺 − 𝑣.Thus 𝐺 − 𝑣 disconnected (with 𝑢 and 𝑤 lying in different
components).hence 𝑣 is a cut vertex.
Note:
No vertex of a complete graph is a cut vertex.
Theorem (2.4)
Let 𝐺 be a graph with 𝑛 vertices, where 𝑛 ≥ 2.Then 𝐺 has at least two vertices
which are not cut vertices.
Proof:-
We may suppose that 𝐺 is a connected graph. We proceed by assuming the
result is false for our 𝐺 and so the proof will be complete if we derive a
contradiction from this.
Thus we are assuming that there is at most one vertex in 𝐺 which is not cut
vertex. Now let 𝑢 , 𝑣 be vertices in 𝐺 such that the distance 𝑑 (𝑢 , 𝑣)between them
is the greatest of distances between pairs of vertices in 𝐺. i.e., 𝑑(𝑢 , 𝑣) =
𝑑𝑖𝑎𝑚(𝐺).since 𝐺 is connected and has at least two vertices, 𝑢 ≠ 𝑣. Thus by our
assumption, one of these vertices must be a cut vertex, say 𝑣. Then 𝐺 − 𝑣 is
disconnected and so there is a vertex 𝑤 in 𝐺 which does not belong to the same
component as 𝑢 does in 𝐺 − 𝑣. This implies that every 𝑢𝑤 path in 𝐺 contains the
vertex 𝑣. It follows from this that the shortest path in 𝐺 from 𝑢 to 𝑤 contains the
shortest path from 𝑢 to 𝑣 and this contradiction completes our proof.

12
Definition (2.3)
A vertex cut of 𝐺 is a subset V of 𝑉 such that 𝐺 − V is disconnected.
A k-vertex cut is a vertex cut of k elements.
Definition (2.4)
Let 𝐺 be a simple graph .The connectivity (vertex connectivity) of 𝐺, denoted
by 𝑘(𝐺),is the minimum number of vertices in 𝐺 whose deletion from 𝐺 leaves
either a disconnected graph or trivial graph.
Note:-
For 𝑛 ≥ 2, the deletion of any vertex from 𝐾n,results in 𝐾n-1 and in general the
deletion of 𝑡 vertices (𝑡 < 𝑛)results in 𝐾𝑛 − 𝑡.
This shows that 𝑘(𝐾𝑛) = 𝑛 − 1.
A connected graph 𝐺 has 𝑘(𝐺) = 1 if and only if either 𝐺 = 2K or 𝐺 has a
cut vertex.𝑘(𝐺) = 0 if and only if either 𝐺=𝐾1 or 𝐺 is disconnected.
Figure (2.5)

13
𝑘(𝐺1) = 3 and 𝑘(𝐺2) = 4.
Theorem (2.5)
The vertex connectivity of any graph G can never exceed the edge
connectivity of G.
Proof:-
Let α denote the edge connectivity of G. Therefore, there exists a cut set S
in G with α edges. Let S partition the vertices of G into subsets 1V and 2V . By
removing at most α vertices from 1V ( or 2V ) on which the edges in S are incident,
we can effect the removal of S (together with all other edges incident on these
vertices) from G. Hence the theorem.
Theorem (2.6)
A graph of order n is k-connected (where 11  nk ) if the degree of each vertex
is at least 2/)2(  kn .
Proof:-
If the graph is complete, it is k-connected for  1 nk . Assume that the graph is
not complete and not k-connected. So there is a disconnecting set S of s vertices
such that G-S is a disconnected graph. Let H be a component of G-S with as few
vertices as possible. If the order of H is r,  rsnr  , which gives the upper
bound   2/sn  for the order r. If v is any vertex of H, the degree of v in the graph

14
G cannot exceed   sr 1 . Thus   2/22/)2(12/)(deg  knsnssnv ,
violating the given inequality.
Theorem (2.7)
A graph is n-edge connected if and only if the number of edges in any cut is
at least k.
Theorem (2.8)
A vertex of a tree is a cut vertex if and only if it is not a pendant vertex.
Proof:-
Let v be a pendant vertex of a tree T. The removal of v results in the removal
of v and the only edge incident on it so that T-v is still connected. Therefore v is
not a cut vertex of T.
Figure (2.6)
Let u be a non pendant vertex of a tree T so that the degree of u is greater than 1.
Therefore there exists at least two edges uv and uw incident on u. Since T is a tree
vuw is the unique path connecting v and w. The removal of u deletes the edges uv
and uw. Therefore there is no path connecting the vertices v and w in T-u and so T-
u is disconnected. Hence u is a cut vertex of the tree T.

15
Definition (2.5)
A simple graph 𝐺 is called 𝑛-connected if 𝑘(𝐺) ≥ 𝑛.
Note:-
𝐺 is 1-connected if and only if 𝐺 is connected and has atleast two vertices.
𝐺 is 2- connected if and only if 𝐺 is connected with atleast 3 vertices but
no cut vertices.
In figure (2.4), 𝐺1 is 3-connected and 𝐺2 is 4-connected.
Definition (2.6)
An edge cut of 𝐺 is a subset of 𝑒 of the form [𝑆 , 𝑆] where 𝑆 is a non-empty
proper subset of 𝑉.
An 𝑘 −edge cut is an edge cut of k elements.
If 𝐺 is non-trivial and 𝐸 is an edge cut of 𝐺, then 𝐺 − 𝐸 is disconnected.
Definition (2.7)
Let 𝐺 be a simple graph. The edge connectivity of 𝐺, denoted by k(G), is the
smallest number of edges in 𝐺 whose deletion from 𝐺 either leaves a disconnected
graph or an empty graph.
Any non empty simple graph 𝐺 with a bridge has k(G) =1.
k(G)=0 if and only if either 𝐺 is disconnected or an empty graph.

16
All non-trivial connected graphs are 1-edge connected.
Figure (2.7)
Here k(𝐺) =2.
Theorem (2.9)
The edge connectivity of a graph 𝐺 cannot exceed the degree of the vertex with
the smallest degree in 𝐺.
Proof:-
Let vertex iv be the vertex with the smallest degree in 𝐺. Let  ivd be the
degree of iv . Vertex iv can be separated from 𝐺 by removing the  ivd edges
incident on vertex iv . Hence the theorem.
Definition (2.8)
A simple graph 𝐺 is called n-edge connected if   nGk  .

17
Theorem (2.10)
A simple graph 𝐺 is n-edge connected if and only if given any pair of distinct
vertices 𝑢 and 𝑣 of 𝐺, there are atleast 𝑛 edge disjoint paths from 𝑢 to 𝑣.
Proof :-
Suppose that 𝐺 is 𝑛-edge connected and 𝑢 and 𝑣 are two distinct vertices of 𝐺.
Then any 𝑢 − 𝑣 separating set of edges must be atleast 𝑛 members. Thus, there
must be atleast 𝑛-edge disjoint 𝑢 − 𝑣 paths, as required.
Conversely, suppose that given any pair of distinct vertices 𝑢 and 𝑣 of 𝐺 ,
there are atleast 𝑛 −edge disjoint paths from 𝑢 to 𝑣. Then, for each pair of vertices
𝑢 and 𝑣 every 𝑢 − 𝑣 separating set of edges must have atleast 𝑛 members. Thus it
requires the deletion of atleast 𝑛 edges from 𝐺 inorder to produce a disconnected
graph or an empty graph. In other words, 𝐺 is 𝑛-edge connected, as required .
Theorem (2.11) (Whitney’s Theorem)
For any graph ,  kk .
Proof :-
Let 𝐺 be a connected graph.
If 𝐺 has no edges, then k=0. Otherwise a disconnected graph results when
all the edges incident with a vertex of minimum degree are removed. In either case,
k .

18
To show kk  .
If 𝐺 is disconnected or trivial, then 0 kk . If 𝐺 is connected and has a
bridge 𝑥, then k=1. In this case, 𝑘 = 1 since either 𝐺 has a cut vertex incident
with 𝑥 or 𝐺 is 𝐾2.
Finally, suppose 𝐺 has 2k edges whose removal disconnects it. Clearly,
the removal of 1k of these lines produces a graph with a bridge uvx  . For
each of these 1k edges, select an incident vertex different from 𝑢 or 𝑣. The
removal of these vertices also removes the 1k edges and quite possibly more. If
the resulting graph is disconnected, then kk  , if not 𝑥 is a bridge, and hence the
removal of 𝑢 or 𝑣 will result in either a disconnected or a trivial graph, so kk 
in every case.
Thus for any graph 𝐺,  kk .
Theorem (2.12)
For any three integers 𝑟, 𝑠, 𝑡 such that tsr 0 there is a graph 𝐺 with
skrk  , and t .
Proof :-
Take two disjoint copies of 𝐾𝑡+1 . Let 𝐴 be a set of 𝑟 vertices in one of them
and 𝐵 be a set of 𝑠 vertices in the other. Join the vertices of 𝐴 and 𝐵 by 𝑠 edges
utilizing all the vertices of 𝐵 and all the vertices of 𝐴. Since 𝐴 is a vertex cut and
the set of these 𝑠 edges is an edge cut of the resulting graph 𝐺, it is clear that

19
  rGk  and   sGk  . Also there is atleast one vertex which is not in 𝐴 ∪ 𝐵 and
it has degree 𝑡, so that 𝛿(𝐺) = 𝑡 .
Theorem (2.13)
 ≥
𝑛
2
ensures k in a graph.
Proof :-
Since it is enough to rule out k , suppose 𝐺 is a graph with
2
n
 and k . Let 𝐹 be a set of k edges disconnecting 𝐺, and 𝐶1 and 𝐶2 be the
components of 𝐺 − 𝐹 and 𝐴1 and 𝐴2 the end vertices of the edges of 𝐹 in 𝐶1 and
𝐶2 respectively. Let rA 1 and sA 2 and suppose 𝑉 (𝐶1 ) = 𝐴1 . Then each
vertex of 𝐶1 is adjacent with atleast one edge of 𝐹. So the number 𝑚1 of edges in
𝐶1 satisfies the following inequality.
 krm  
2
1
1
   r
2
1
since k .
 1
2
1
 r
 rr 1
2
1
 since  Fr k .

20
But a graph on r vertices cannot have more than  rr 1
2
1
 edges. Thus,
  11 ACV  and similarly,   22 ACV  .
Thus each of 1C and 2C contains atleast 𝛿 + 1 vertices.
 GVn 
 12  






 1
2
2
n
= 𝑛 + 2
This contradiction establishes the theorem.
Problem (2.2)
Show that if 𝑎 and 𝑏 are the only two odd degree vertices of a graph 𝐺, then
𝑎 and 𝑏 are connected in 𝐺.
Solution:
If 𝐺 is connected, nothing to prove.
Let 𝐺 be disconnected.
If possible assume that 𝑎 and 𝑏 are not connected.
Then 𝑎 and 𝑏 lie in the different components of 𝐺. Hence the component of 𝐺
containing 𝑎 (similarly containing 𝑏) contains only one odd degree vertex 𝑎, which

21
is not possible as each component of 𝐺 is itself a connected graph and in a graph
number of odd degree vertices should be even.
Therefore 𝑎 and 𝑏 lie in the same component of 𝐺.
Hence they are connected.
Problem (2.3)
Prove that a connected graph 𝐺 remains connected after removing an edge 𝑒 from
𝐺 if and only if 𝑒 lie in some circuit in 𝐺.
Solution:
If an edge 𝑒 lies in a circuit 𝐶 of the graph 𝐺 then between the end vertices of 𝑒,
there exist at least two paths in 𝐺.
e
Figure (2.8)
Hence removal of such an edge e from the connected graph G will not effect
the connectivity of G. Conversely, if e does not lies in any circuit of G then
removal of e disconnects the end vertices of e.
Hence G is disconnected.

22
Definition (2.9)
A connected graph that has no cut vertices is called a block.
Every block with atleast 3 vertices is 2-connected.
Definition (2.10)
A block of a graph 𝐺 is a maximal subgraph 𝐻 of 𝐺 such that 𝐻 is a block.
Every graph is the union of its blocks.
Figure (2.9)

23
The blocks of 𝐺 are
Definition (2.11)
Let 𝑢 and 𝑣 be two vertices of a graph 𝐺. A collection {𝑃(1), 𝑃(2), … , 𝑃(𝑛)}
of 𝑢 − 𝑣 paths is said to be internally disjoint if , given any distinct pair 𝑃(𝑖) and
𝑃(𝑗) in the collection, 𝑢 and 𝑣 are the only vertices 𝑃(𝑖) and 𝑃(𝑗)in common.
Theorem (2.14)
Let 𝐺 be a simple graph with atleast 3 vertices. Then 𝐺 is 2-connected if and
only if for each pair of distinct vertices 𝑢 and 𝑣 of 𝐺 there are two internally
disjoint 𝑢 − 𝑣 paths in 𝐺.
Proof :-
Suppose that any pair of distinct vertices is connected by a pair of internally
disjoint paths. Then clearly 𝐺 is connected so it remains to prove that 𝐺 has no cut

24
vertices. Assume to the contrary, that 𝑣 is a cut vertex of 𝐺. Then there are two
vertices 𝑢 and 𝑤 of 𝐺, both different from 𝑣, such that 𝑣 is on every 𝑢 − 𝑤 path in
𝐺. However, by the hypothesis there are two internally disjoint 𝑢 − 𝑤 paths in
𝐺 and atmost one of these can then pass through 𝑣. Thus 𝑣 is not on every 𝑢 −
𝑤 path, a contradiction. Hence 𝐺 has no cut vertices.
Conversly, suppose that 𝐺 is 2-connected. Let 𝑢 and 𝑣 be a pair of distinct
vertices of 𝐺. We use induction on 𝑑(𝑢, 𝑣), the distance between 𝑢 and 𝑣, to show
that there is a pair of internally disjoint 𝑢 − 𝑣 paths. First, if 𝑑(𝑢, 𝑣) = 1 then
𝑢 and 𝑣 are joined by an edge, say 𝑒. It follows that 𝑒 is not a bridge, since 𝐺 has
no cut vertices. Hence , there is a 𝑢 − 𝑣 path 𝑃 different and so internally disjoint
from the 𝑢 − 𝑣 path 𝑄 given by the single edge 𝑒.
We now assume that   2,  kvud and that if 𝑥 and 𝑦 are any pair of
vertices with 𝑑(𝑥, 𝑦) < 𝑘 then there are two internally disjoint 𝑥 − 𝑦 paths. Let
𝑃 be a path of length 𝑘 from 𝑢 to 𝑣 and let 𝑤 be the second last vertex of 𝑃. Then
𝑑(𝑢, 𝑤) = 𝑘 − 1 and there are two internally disjoint 𝑢 − 𝑤 paths say 𝑄1 and 𝑄2 .
Since 𝐺 is 2-connected, 𝑤 is not a cut vertex, i.e., 𝐺 − 𝑤 is connected and so
there is a 𝑢 − 𝑣 path P which does not pass through 𝑤. Let 𝑥 be the last vertex of
p which is also a vertex of either 𝑄1 or 𝑄2 .

25
Figure (2.10)
Suppose that 1Qx . Let 𝑃1 be the 𝑢 − 𝑣 path given by the 𝑢 − 𝑥 section of 𝑄1
followed by the 𝑥 − 𝑣 section of P . Let 𝑃2 be the 𝑢 − 𝑣 path given by the path
𝑄2 followed by the edge 𝑤𝑣. Then 𝑃1 and 𝑃2 are internally disjoint by the
definition of 𝑄1, 𝑄2, 𝑥 and P . The proof now follows by induction and is
complete.
Corollary (2.15)
If 𝐺 is 2-connected, then any two vertices of 𝐺 lie on a common cycle.
Definition (2.12)
An edge 𝑒 is said to be subdivided when it is deleted and replaced by a path of
length two connecting its ends, the internal vertex of this path being a new vertex.

26
Figure (2.11)
Subdivision of an edge
Note :-
The class of blocks with atleast 3 vertices is closed under the operation of
subdivision.
Corollary (2.16)
If 𝐺 is a block with 𝑛 ≥ 3, then any two edges of 𝐺 lie on a common cycle.
Proof :-
Let 𝐺 be a block with 𝑛 ≥ 3, and let 𝑒1 and 𝑒2 be two edges of 𝐺. Form a new
graph G by subdividing 𝑒1 and 𝑒2 and denote the new vertices by 𝑣1 and 𝑣2.
Clearly G is a block with atleast five vertices, and hence is 2-connected. It follows
that 𝑣1 and 𝑣2 lie on a common cycle of G . Thus 𝑒1 and 𝑒2 lie on a common cycle
of 𝐺.

27
CHAPTER-2
APPLICATIONS
Application (3.1)
If we think of graph as representing a communication network, the
connectivity (or edge connectivity) becomes the smallest number of
communication stations (or communication links) whose break down would
jeopardize communication in the system. The higher the connectivity and edge
connectivity, the more reliable the network. From this point of view, a tree
network, such as the one obtained by Kruskal’s Algorithm, is not very reliable, and
one is led to consider the following generalization of the connector problem.
Let k be a given positive integer and let 𝐺 be a weighted graph. Determine a
minimum-weight k-connected spanning subgraph of 𝐺.
For k=1, this problem reduces to the connector problem, which can be solved
by Kruskal’s Algorithm. For values of k greater than one, the problem is unsolved
and is known to be difficult. However if 𝐺 is a complete graph in which each edge
is assigned unit weight, then the problem has a simple solution which we now
present.
Observe that, for a weighted complete graph on 𝑛-vertices in which edge is
assigned unit weight, a minimum-weighted 𝑚-connected spanning subgraph is
simply an m-connected graph on 𝑛 vertices with as few edges as possible. We shall
denote by 𝑓(𝑚, 𝑛) the least number of edges that an 𝑚-connected graph on 𝑛
vertices can have (It is, of course, assumed that 𝑚 < 𝑛).
𝑓(𝑚, 𝑛) ≥ [
𝑚𝑛
2
] (1)

28
We shall show that equality holds in (1) by constructing an 𝑚-connected graph
Hm ,n on 𝑛 vertices that has exactly [
𝑚𝑛
2
] edges. The structure of Hm ,n depends on
the pattern of 𝑚 and 𝑛, there are 3 cases.
Case 1 : 𝑚 even. Let 𝑚 = 2𝑟. Then H2r,m is constructed as follows. It has vertices
0,1, … . , 𝑛 − 1 and two vertices 𝑖 and 𝑗 are joined if 𝑖 − 𝑟 ≤ 𝑗 ≤ 𝑖 + 𝑟 . H8,n is
shown in figure 3.1 (a).
Case 2 : 𝑚 odd, 𝑛 even. Let 𝑚 = 2𝑟 + 1. Then H2r+1,n is constructed by first
drawing H2r,n and then adding edges joining vertex 𝑖 to vertex 𝑖 +
𝑛
2
for 1 ≤ 𝑖 ≤
𝑛
2
.
H9,8 is shown in figure 3.1(b).
Case 3 : 𝑚 odd, 𝑛 odd. Let = 2𝑟 + 1 . Then H2r+1, n is constructed by first drawing
H2r,n and then adding edges joining vertex 0 to vertices
𝑛−1
2
and
𝑛+1
2
and vertex 𝑖 to
vertex 𝑖 +
𝑛+1
2
for 1 ≤ 𝑖 <
𝑛−1
2
. H9,9 is shown in figure 3.1 (c).

29
Figure (3.1)
Theorem (3.1) (Harary Theorem)
The graph nmH , is 𝑚-connected.
Proof :-
Consider the case 𝑚 − 2𝑟. We shall show that nmH , has no vertex cut of
fewer than 2𝑟 vertices. If possible, let V  be a vertex cut with |V | < 2𝑟 . Let 𝑖 and
𝑗 be vertices belonging to different components of VH nr
,2 . Consider the two
sets of vertices 𝑆 = {𝑖, 𝑖 + 1, … , 𝑗 − 1, 𝑗} and 𝑇 = {𝑗, 𝑗 + 1, … , 𝑖 − 1, 𝑖} where

30
addition is taken modulo 𝑛. Since |V | < 2𝑟, we may assume, without loss of
generality, that |V ∩ 𝑠| < 𝑟. Then there is clearly a sequence of distinct vertices in
𝑆| V which starts with 𝑖, ends with 𝑗, and is such that the difference between any
two consecutive terms is atmost 𝑖. But such a sequence is an (𝑖, 𝑗) path in
VH nr
,2 , a contradiction. Hence nrH ,2 is 2𝑟-connected. The case 𝑚 = 2𝑟 + 1 is
also 2𝑟 + 1 -connected.
It is easy to see that  




2
,
mn
nmf (2)
It now follows from (1) and (2) that  




2
,
mn
nmf and that nmH , is an 𝑚-
connected graph on 𝑛-vertices with a few edges as possible.
Application (3.2)
Suppose we are given n stations that are to be connected by means of e lines
(telephone lines, bridges, rail roads, tunnels, or highways) where 1 ne . What is
the best way of connecting? By “best” we mean that the network should be as
invulnerable to destruction of individual stations and individual lines as possible.
In other words, construct a graph with n vertices and e edges that has the maximum
possible edge connectivity and vertex connectivity.

31
Figure (3.2)
For example, the graph in figure (3.2) has n=8, e=16, and has vertex
connectivity of one and edge connectivity of three. Another graph with the same
number of vertices and edges (8 and 16, respectively) can be drawn as shown in
figure (3.3).
Figure (3.3)
Graph with 8 vertices and 16 edges
It can easily be seen that the edge connectivity as well as the vertex connectivity
of this graph is four. Consequently, even after any three stations are bombed, or

32
any three lines destroyed, the remaining stations can still continue to
“communicate” with each other. Thus the network of figure (3.3) is better
connected than that of figure (3.2) (although both consist of the same number of
lines-16).
Theorem (3.2)
The maximum vertex connectivity one can achieve with a graph G of n
vertices and e edges  1 ne is the integral part of the number n
e2
Proof :-
Every edge in G contributes two degrees. The total (2e degrees) is divided
among n vertices. Therefore, there must be at least one vertex in G whose degree is
equal to or less than the number n
e2 . The vertex connectivity of G cannot exceed
this number.
To show that this value can actually be achieved, one can first construct an n-
vertex regular graph of degree equal to the integral part of the number n
e2 and
then add the remaining  2
ne  .(integral part of the number n
e2 ) edges
arbitrarily.
Thus we can summarize as follows:
vertex connectivity ≤ edge connectivity ≤
n
e2
,
and

33
maximum vertex connectivity possible = integral part of the number n
e2 .
Thus for a graph with 8 vertices and 16 edges, we can achieve a vertex
connectivity(and therefore edge connectivity) s high as four ( 8
162  ).

34
REFERENCES
1.Diestel, R., Graph Theory, Electronic Edition,
2. Chartrand, Gary (1985), Introductory Graph Theory, Dover.
3. Bondy, J.A.; Murty, U.S.R. (2008), Graph Theory, Springer.
4. Biggs, N.; Lloyd, E.; Wilson, R. (1986), Graph Theory, 1736–1936, Oxford
University Press.
5. Harary, Frank (1969), Graph Theory, Reading, MA: Addison-Wesley.

A study on connectivity in graph theory june 18 pdf

More Related Content

What's hot (20)

Viewers also liked (6)

Similar to A study on connectivity in graph theory june 18 pdf (20)

Recently uploaded (20)

A study on connectivity in graph theory june 18 pdf