Cut-sets and cut-vertices

Cut-set
• In a connected graph G, a cut-set (cocycle) is a set of edges whose
removal from G leaves G disconnected.
Cut-sets are
{a,c,d,f}, {a,b,g},
{a,b,e,f} and
{d,h,f}. What
about edge {k} ?
{a,c,h,d} is not a
cut-set because
one of its proper
subset {a,c,h} is a
cut-set.
• To emphasize the fact that no proper subset of a cut-set can be a
cut-set, the cut-set is referred as a minimal cut-set, a proper cut-set
or a simple cut-set.
• The cut-set always cuts the graph into two. It is defined as a
minimal set of edges in a connected graph whose removal reduces
the rank of the graph by one.
• Rank of the graph in Fig. 4.1(b) is _________ which is less than that of
graph in 4.1 (a) _________ .
• Another way of looking at a cut-set is : if we partition all the
vertices of a connected graph G into two mutually exclusive
subsets, a cut-set is a minimal number of edges whose removal
from G destroys all paths between these two sets of vertices.
• True or False: every edge of a tree is a cut-set.
• Applications: communication and transportation networks, which
help to identify the bottleneck of the network.
Properties of a cut-set
• Consider a spanning tree T in a connected graph G and an
arbitrary cut-set S in G.
• Is it possible for S not to have any edge in common with T?
• No, otherwise removal of the cut-set S from G would not
disconnect the graph.
• Theorem 4.1: Every cut-set in a connected graph G must contain
at least one branch of every spanning tree of G.
• Will the converse also be true? That is, will any minimal set of
edges containing at least one branch of every spanning tree be a
cut-set?
• Answer is: Yes
• Prove.
• Theorem 4.2: In a connected graph G, any minimal set of edges
containing at least one branch of every spanning tree of G is a cut-
set.
• Theorem 4.3: Every circuit has an even number of edges in
common with any cut-set.
All cut-sets in a graph
• Cut-sets are used to identify weak spots in a communication net.
• This requires to list all cut-sets of the corresponding graph, and
find which ones have the smallest number of edges.
• There are large number of cut-sets in a small graph, like 4.1, which
requires a systematic method of generating all relevant cut-sets.
• How to find the circuits in a graph?

• The strategy to find cut-sets is same.

• Just as a spanning tree is essential for defining a set of
fundamental circuits, it is also essential for a set of fundamental
cut-sets.
Fundamental cut-sets
• Consider a spanning tree T of a connected graph G. Take any
branch b in T. Since {b} is a cut-set in T, {b} partitions all vertices of
T into two disjoint sets – one at each end of b. Consider the same
partition of vertices in G, and the cut-set S in G that corresponds to
this partition.
• Cut-set S will contain only one branch b of T, and the rest (if any) of
the edges in S are chords with respect to T. Such a cut-set S
containing exactly one branch of a tree T is called a fundamental
cut-set (basic cut-set) with respect to T.
• Just as every chord of a spanning tree defines a unique fundamental circuit,
every branch of a spanning tree defines a unique fundamental cut-set.
• The fundamental cut-set has meaning only with respect to a given spanning
tree.
How to obtain other cut-sets from a given
set of cut-sets.
• Theorem 4.4: The ring sum of any two cut-sets in a graph is either
a third cut-set or an edge disjoint union of cut-sets.
Example
• We cannot start with any two cut-sets in a given graph and obtain
all its cut-sets by this method.
• What is a minimal set of cut-sets from which we can obtain every
cut-set of G by taking ring sum? – Answer: set of all fundamental
cut-sets with respect to a given spanning tree.
Fundamental Circuits and Cut-sets
• Consider a spanning tree T in a given connected graph G. Let ci be a
chord with respect to T, and let the fundamental circuit made by ci
be called Γ, consisting of k branches b1, b2, …, bk in addition to the
chord ci, that is
Γ = {ci, b1, b2, …, bk } is a fundamental circuit with respect to T.
• Every branch of any spanning tree has a fundamental cut-set
associated with it. Let S1, be the fundamental cut-set associated
with b1, consisting of q chords in addition to the branch b1, that is
S1 = {b1, c1, c2, … , Cq} is a fundamental cut-set with respect to T.
• According to theorem 4.3, there must be an even number of edges
common to Γ and S1, and there is only one other edge in Γ (which
is ci) that can possibly also be in S1.
• Therefore, there must be two edges b1 and ci common to Γ and S1.
Thus, the chord ci is one of the chords c1, c2, … , cq.
• The same argument holds for fundamental cut-sets associated
with b2, b3,…, bk. Therefore, the chord ci is contained in every
fundamental cut-set associated with branches of Γ.
• Is it possible for the chord ci to be in any other fundamental cut-set
S’ ( with respect to T) besides those associated with b1, b2, …, bk?
• Answer: No
• Otherwise (since none of the branches in Γ are in S’), there would
be only one edge ci common in S1 and Γ, a contradiction to
Theorem 4.3.
• Theorem 4.5: With respect to a given spanning tree T, a chord ci,
that determines a fundamental circuit Γ occurs in every
fundamental cut-set associated with the branches in Γ and in no
other.
 Example
• Consider the spanning tree
{b,c,e,h,k} and the
fundamental circuit made by
chord f is {f,e,h,k}.
• The three fundamental cut-
sets determined by the three
branches e, h and k are:

Chord f occurs in each of these three fundamental cut-sets, and there is

no other fundamental cut-set that contains f.
• The converse of Theorem 4.5 is also true.
• Theorem 4.6: With respect to a given spanning tree T, a branch bi
that determines a fundamental cut-set S is contained in every
fundamental circuits associated with the chords in S, and in no
other.
• Self study: Proof of Theorem 4.6.
List all the fundamental cut-sets w.r.t any
spanning tree and generate other cut-sets.
Connectivity and Separability
• Edge Connectivity: Each cut-set of a connected graph G consists of
a certain number of edges. The number of edges in the smallest
cut-set (cut-set with fewest number of edges) is defined as the
edge connectivity of G. In other words, the edge connectivity of a
connected graph can be defined as the minimum number of edges
whose removal reduces the rank of the graph by one.
• The edge connectivity of a tree is one.
• Vertex connectivity of a connected graph G is defined as the
minimum number of vertices whose removal from G leaves the
remaining graph disconnected.
• The vertex connectivity of a tree is one.
• The vertex connectivity is meaningful only for graphs that have
three or more vertices and are not complete.
• Note: The edge and vertex connectivity of a disconnected graph is
zero.
• Does the vertex connectivity of a complete graph with n vertices
exist?
Edge and vertex connectivity Examples
• Karger’s algorithm to find min-cut edge connectivity
Separable Graph:
• A connected graph is said to be separable if its vertex connectivity
is one. All other connected graphs are called nonseparable.
• In a separable graph there exists a subgraph g in G such that g and
it’s complement have only one vertex in common.
• In a separable graph a vertex whose removal disconnects the
graph is called a cut-vertex, a cut-node or an articulation point.
• In a tree every vertex with degree greater than one is a cut-vertex.
• Theorem 4-7: A vertex v in a connected graph G is a cut-vertex if
and only if there exist two vertices x and y in G such that every
path between x and y passes through v.
• The theorem states that v is a crucial vertex in the sense that any
communication between x and y.
• Suppose we are given n stations that are to be connected by means
of e lines (telephone lines, bridges, railroads, tunnels, or highways)
where e ≥ n − 1.
• 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.
Two graphs with 8 vertices and 16 edges
• What is the highest vertex and edge connectivity we can achieve
for a graph with n vertices and e edges?
• Theorem 4-8: The edge connectivity of a graph G cannot exceed
the degree of the vertex with the smallest degree in G.
• Theorem 4-9: The vertex connectivity of any graph G can never
exceed the edge connectivity of G.
• Theorem 4.10: The maximum vertex connectivity one can achieve
with a graph G on n vertices and e edges (e≥ n-1) is the integral
part of the number 2e/n ;
What is the edge and vertex connectivity of
the graph?
What is the edge and vertex connectivity of
the graph?
• Edge connectivity = 3
• Vertex connectivity = 3
• K-connected graph: A graph G is said to be k-connected if the vertex
connectivity of G is k.
• Is 1-connected graph is the same as a separable graph?
• Theorem 4.11: A graph G is k-connected if and only if every pair of
vertices in G is joined by k or more paths that do not intersect and at
least one pair of vertices is joined by exactly k nonintersecting paths.
• Theorem 4.12: The edge connectivity of a graph G is k if and only if
every pair of vertices in G is joined by k or more edge-disjoint paths,
and at least one pair of vertices is joined by exactly k edge disjoint
paths.
Is the given graph 2-connected? If so,
identify the articulation points.
Is the given graph 2-connected? If so,
identify the articulation points.
• Yes, the graph is 2-
connected and
articulation points are 3
and 4.
What is the edge connectivity of the graph?
1-Isomorphism
• A separable graph consists of two or more nonseparable
subgraphs. Each of the largest nonseparable subgraphs is called a
block.
• The above two graphs are not isomorphic (as they do not have same number of
vertices), but they are related by the fact that the blocks of the graph 1 are
isomorphic to the components of the graph 2. Such graphs are said to be 1-
isomorphic.
• Two graphs G1 and G2 are said to be 1-isomorphic if they become isomorphic
to each other under repeated application of the following operation:
Operation 1: “Split” a cut-vertex into two vertices to produce two disjoint
subgraphs.
Two nonseparable graphs are 1-isomorphic if and only if they are isomporphic.
• Theorem 4-14: If G1 and G2 are 1-isomorphic graphs, the rank of G1
equals the rank of G2 and the nullity of G1 equals the nullity of G2.
 1-isomorphic graphs

Can we join two In that case, all the

blocks of fig. 4.9 by graphs are 1-
gluing together isomorphic to each
two vertices? - Yes other.
• 1-isomorphism for nonseparable graphs is the same as
isomorphism, however for separable graph, 1-isomorphism is
different from isomorphism.
• Graphs that are isomorphic are also 1-isomorphic, but 1-
isomorphic graphs may not be isomorphic.
• This finding help in extending the concept to 2 – isomorphic
graphs for 2-connected graphs (graphs with vertex connectivity of
two).
• In a 2-connected graph G let vertices x and y be a pair of vertices
whose removal from G will leave the remaining graph disconnected.
• G consists of a subgraph g1 and its complement g1’, such that g1
and g1’ have exactly two vertices x and y, in common. Suppose that
we perform the following operation 2 on G (G will not remain
original after this operation):
• Operation 2: “Split” the vertex x in x1 and x2 and the vertex y into y1
and y2 such that G is split into g1 and g1’. x1 and y1 go with g1 and x2
and y2 will go with g1’. Now rejoin the graphs g1 and g1’ by merging
x1 with y2 and x2 with y1.
• Two graphs are said to be 2-isomorphic if they become isomorphic
after undergoing operation 1 or operation 2, or both operations
any number of times.

Note that in (a) the

degree of vertex x is
four, but in (d) no
vertex is of degree four.
• Note: Isomorphic graphs are always 1-isomorphic, and 1-
isomorphic graphs are always 2-isomorphic, but 2-isomorphic
graphs are not necessarily 1-isomorphic, and 1-isomorphic graphs
are not necessarily isomorphic. However, for graphs with
connectivity three or more, isomorphism, 1-isomorphism, and 2-
isomorphish are synonymous.
• No edges or vertices are created or destroyed under operation 2.
Therefore, the rank and nullity of a graph remain unchanged
under operation 2. Therefore, 2-isomorphic graphs are equal in
rank and equal in nullity.
• Circuit Correspondence: Two graphs G1 and G2 are said to have a
circuit correspondence if the following condition is satisfied:
There is a one-to-one correspondence between the edges of G1 and G2, such
that a circuit in G1 formed by certain edges of G1 has a corresponding
circuit in G2 formed by the corresponding edges of G2, and vice versa.
Isomorphic graphs always have circuit correspondence.
• Since in a separable graph G every circuit is confined to a
particular block, every circuit in G retains its edges as G
undergoes operation 1. Hence 1-isomorphic graphs have circuit
correspondence.
• What happens to a circuit in a graph G when it undergoes operation 2?
• A circuit Γ in G will fall in one of three categories:
1. Γ is made of edges all in g1, or
2. Γ is made of edges all in g1’, or
3. Γ is made of edges from both g1 and g1’, and in that case Γ must include
both vertices x and y.
• In case 1 and 2, Γ is unaffected by operation 2. In case 3, Γ still has the
original edges, except that the path between vertices x and y in g1,
which constituted a part of Γ, is “flipped around”. Thus every circuit in
a graph undergoing operation 2 retains its original edges. Therefore,
2-isomorphic graphs also have circuit correspondence.
• Theorem 4-15: Two graphs are 2-iomorphic if and only if they
have circuit correspondence. (H. Whitney).

• 2-isomorphism and circuit correspondence have applications in

theory of contact networks, electrical networks and in duality of
graphs.
Network Flow
• Assumptions
• At each intermediate vertex the total rate of commodity entering is equal
to the rate of leaving.
• The flow through a vertex is limited only by the capacities of the edges
incident on it.
• Lines(edges) are lossless.
• Questions
• What is the maximum flow possible through the network between a
specified pair of vertices?
• How do we achieve this flow?
• A cut-set with respect to a pair of vertices a and b in a connected
graph G puts a and b into two different components.
• The capacity of a cut-set S in a weighted connected graph G (in
which the weight of each edge represents its flow capacity) is
defined as the sum of the weights of all the edges in S.
• Theorem 4.13: The maximum flow possible between two vertices
a and b in a network is equal to the minimum of the capacities of
all cut-sets with respect to a and b.