Graph Theory Class Note 2
Graph Theory Class Note 2
In this section we discuss two important concepts in graph theory – planarity and duality.
First we consider planar graphs and derived some properties of these graphs then the dual
graphs and related properties. Duality has been of considerable interest to electrical
network theories. This interest is due to the fact that the voltages and currents in an
electrical network are dual variables.
Can you find a path from each dog to each house such that no two paths intersect?
A quadapus is a little-know animal similar to an octopus, but with four arms. Here are
five quadapi resting on the sea floor.
Can each quadapus simultaneously shake with others in such a way that no arms cross?
Informally, a planner graph is a graph that can be drawn in the plane so that no edges
cross. Thus above two examples are asking whether two graphs below are planar, i.e.
whether they can be redrawn so that no edges cross.
K3,3 K5
Here, Fig.-1is not planar but Fig.-2 and Fig.-3are planer graphs.
Theorem: A connected planar graph with n vertices and e edges has e – n +2 regions.
Proof: Here it is sufficient to prove the theorem for a simple graph, because adding a
self-loop or a parallel edge simple adds one region to the graph and simultaneously
increases the value of e by one. We also remove the edges that do not form the
boundaries of any region, because addition of any such edge increase by one and
increases n by one, keeping the quantity e – n fixed.
A polygonal net can represent any planer graph, because for simple graphs the plane
representation are made by the straight lines. Let the polygonal net representing the
given graph consist of r region or faces and let Kp be the p-sided regions.
Since each edge is on the boundary of exactly two regions, we have
3.K3 + 4.K4 + 5.K5 + ……+ Ks = 2e
and K3 + K4 + K5 + ……+ Ks = r
where Ks is the number of polygons with maximum edges.
Now sum of all angles subtended at each vertex in the polygonal net is 2n .
We know that sum of all interiors and exterior angles of a s-sided region are respectively
(s-2) and (s+2). Therefore
(3 – 2)K3 + (4 – 2)K4 + …. + (s – 1)Ks + 4 = 2n [ since grand sum is the sum of
interior angles of r – 1 regions and exterior angles of unbounded region].
(2e – 2r) + 4 = 2n
e–r+2=n
r = e - n +2.
Proof: Let G1, G2, ….., Gk be the components of G each of which is a planar (nI, eI)
graph. Let rI denote the number of regions of GI ( i = 1, 2, .. k). Then rI = eI – nI + 2 for
each i = 1, 2, ….., k. Summing up, we get
k k
ri = (e i ni 2)
i 1 i 1
r + (k – 1) = e - n +2k [since unbounded region is counted k – 1 times in the sum]
n – e + r = k + 1.
Proof: Let r be the number of regions of G. Since every region is bounded by p edges
(including the unbounded region) and each edge lies on the boundary of exactly two
regions we have
r
2e = the number of edges in the boundary of ith region = rp
i 1
2e
=r=e–n+2
p
2e = p(e – n + 2)
(p – 2)e = p( n - 2)
p(n 2)
e= .
p 2
3
Corollary3: If G be a simple planar (n, e) with regions then e r and e 3n – 6,
2
where n 3.
Fig. – 4 Fig. – 5.
n (n 1)
Proof: K5 is the complete graph of five vertices. Here, n = 5 and e = = 10.
2
Therefore, 3n – 6 = 15 – 6 = 9 < e(= 10).
Thus the graph violates the second inequality of Corollary-3 and hence it is non-planar.
Similarly for K3,3 n = 6 and e = 3.3 = 9
Therefore, 3n – 6 = 16 – 6 = 12 > e (= 9).
Thus K3,3 satisfies the above inequalities, but the graph is non- planar.
To prove the non-planarity of Kuratowski‟s second graph, we use the additional
condition that no region in K3,3 can be bounded with fewer than four edges. Hence, if this
graph is planar then 2e 4r and substituting for r from Euler‟s formula
e 4(e n 2)
2.9 4(9 – 6 + 2)
18 20, which is a contradiction.
Duality:
In this section we characterized the class of graphs which have duals. First we
prove that every planner graph has a dual. The proof is based on a procedure for
constructing a dual of a given planer graph.
Geometric Dual:
Given a planar graph G, its geometric dual G* is constructed as follows: Place a
vertex in each region of G including the unbounded region. For each edge e of G, draw
an edge e* joining the vertices representing the face or faces for which e is a boundary
such that e* crosses only the edge e. The new graph G* is called the geometric dual of G.
G* is always a plane connected graph.
The following observations can be made about the relationship between a planar graph
and its dual:
Theorem: If G is a planar graph, then d(f ) = 2e where F(G) is the set of all faces of G.
f F( G )
Exercise:
Graph Coloring
The intuitive idea behind the colouring of a graph is simply paint its vertices such
that no two adjacent vertices have same color. What is the minimum number of colors
necessary? This constitutes a coloring problem.
After the vertices have been painted, they can be categorized into the groups or
its; for example, one set having all the blue colors, another of red and so forth. This is a
partitioning problem. The colouring and partitioning can either be carried out on vertices
or on edges of the graph and for planer graphs; these can be performed in the regions.
Colouring and petitioning have several important applications such as; coding theory,
partitioning of logic in digital computer, scheduling and assignments etc.
A graph can be coloured by assigning a different colour to each of its vertices. However,
for most graphs, a colouring can be found that uses fewer colours than the number of
vertices in the graph. A colour with colours such that no two adjacent vertices have the
same colour is called properly coloured graph.
R G B B Y
R R R
R
Y B Y B R B
Fig.-1
Definition 2: A graph G is said to be k-colourable if each vertex can be assigned one of k
colours so that adjacent vertices get different colours.
Here, we are interested with the proper colouring that requires the minimum number of
colours. The minimum colouring of a graph is generally difficult to compute, but the
following theorem provides an upper bound.
Proof: We use the induction principle on the number of vertex n of G. Let P(n) be the
proposition that an n-vertex graph with maximum degree at most k is (k+1)-colorable. A
1-vertex graph G has maximum degree 0 and is 1-colorable, so P(1) is true.
Let us assume that P(n) is true and G be an (n+1) vertex graph with maximum degree at
most k. Remove a vertex v from G. The maximum degree of G-v is at most k and G-v is
(k+1)-colorable by our assumption P(n). Now add back the vertex v to G. We can assign
v a different color from all adjacent vertices, since v has degree at most k and k+1 color
are available. Therefore, G is (k+1)-colorable. The theorem follows by induction.
Definition 3: The chromatic number of a graph G denoted by (G) is the least number of
colors needed for its proper coloring. Such a graph is called -chromatic.
It is easily verified that the graph in Fig.(i),(ii) and (iii) are all 3-chromatic.
In coloring graphs we are concentrates with colouring of connected graphs only (as in
disconnected graphs, colouring of the vertices in one component does not effect the
other). Self-loops are also not considered for the colouring. Parallel edges between two
vertices can be replaced by a simple edge without effecting adjacency of vertices. Thus,
for colouring problems we need to consider only simple connected graphs.
(ii) A graph with one or more edges (without a self-loop) is at least 2-chromatic.
(iv) (Km. n) = 2 where Km,n is the complete bipartite graph with m, n vertices.
2 if n is even
(v) (Cn) = where Cn is the cycle graph with n vertices.
3 if n is odd
3 if n is even
(vi) (Wn) = where Wn is the wheel graph with n vertices.
4 if n is odd
v2 v3 v4
v1
v5 v6
v7 v8
Fig.-2
Theorem 4: For any graph G, (G) 1 + (G) where (G) is the maximum degree of a
vertex in G.
Proof: Let H be the smallest induced subgraph of G such that (H) = n. Then (H - v) =
n – 1 for any vertex v H. It follows that d(v) n – 1 so that n - 1 dmax(H) dmax(G) =
(G).
Proof: Let G be a simple connected planer graph with n-vertices. If n 5, the result is
obvious. Let n 5. We complete the proof by induction hypothesis. Since G is a
connected planer graph. It must have a vertex of degree less than or equal to 5. Let v be a
vertex of G such that d(v) 5. Let G/ = G - v. Then by induction hypothesis G/ is 5-
colorable. When v has degrees 1, 2, 3 and 4 there is no difficulty, since we can give to v
one more color.
We are left with the case when d(v) = 5 and all the five colors are used in coloring
vertices v1, v2, v3, v4, v5 which are adjacent to v in G. Let vi be colored by ci (say) for i
=1, 2, 3, 4 and 5.
v5 v1 c3
v2 c1
v4 v
c3
v3
c1
c1
c3
Fig.-3
1 1
Note: Every graph with e edges satisfies (G) 2e .
2 4
a b
e
g c
f d
Fig.-4
Example:
v2
e1 e2
e3
v1
e4 e5
v3 e6 v4
Fig.- 5
v3 v4
v1 v2
We may color the vertex v1 with any one of colors the second vertex v2 can be color
properly in -1 ways. Father we may color the third vertex v3 using any of - 1 color
which has not been used in v2 similarly in v4.
Theorem 3: If G is a disconnected graph with components G1, G2, ……….., Gi then the
product PG1 (λ) , PG 2 (λ) ,…… PG r (λ) , of there respect polynomials equal to PG ( ).
Proof: Left as an exercise.
Theorem 4: If v1 and v2 be the two nonadjacent vertices of a graph G then its polynomial
PG ( ) = PG1 (λ) + PG 2 (λ) . Where G1 and G2 are obtain from G by adding an edge
between v1, v2 and by fusion of v1, v2 respectively.
Proof: PG ( ) = the number of different colorings from colors = (the number of
colorings of G in which u and v receive different colors) + (the number of coloring in
which u and v receive the same color)
= (number of coloring of G1) + (number of coloring of G2)
= PG1 (λ) + PG 2 (λ) .
Example: the following figure illustrates how the chromatic polynomial of a graph G is
as a sum of chromatic polynomial of complete graphs.
= +
v
+ + +
Exercise:
1. Prove that chromatic polynomial of any tree with n vertices is k(k – 1)n -1.
2. Find the chromatic polynomials of the following graphs and hence determine their
chromatic numbers.
Within a class of students (containing some girls and some boys), each girl likes some
boys and does not like others. Under what conditions can each girl are paired up with a
boy that she likes? Many applications of graphs involve such pairings. We can model the
situation with a bipartite graph, since it has a vertex partition into two sets. If a girl likes a
boy, put an edge between them. Here our goal is to find a matching (or assignment) for
the girls. That is, a subset of edges such that exactly one edge is incident to each girl and
at most one edge is incident to each boy.
A complete matching (if it exists) is the largest matching. Whereas the converse is not
necessary true.
An alternating path P, with respect to M is termed as M-augmenting if its end vertices are
not incident with any edge of M (that is end vertices are M-unsaturated).
Proof: Assume, for contradiction, that there exists an M-augmenting path P. Then by
deleting the edges of E(P) M and adding the edges of E(P) M we get a new matching
Given a finite set of girl each of whom knows several boys, under what condition
can we marry of the girls in such a way that each girl marries a boy he knows. The
problem was solved by P. Hall in 1935, and knows as Hall‟s marriage theorem.
Hall‟s marriage theorem states necessary and sufficient conditions for the existence of a
matching in a bipartite graph. Hall‟s, theorem is a very useful mathematical tool in graph
theory.
Here we proof the Hall‟s theorem using girl marry-boy terminology. For example the
relation between the girls and boys as shown below:
For us to have any chance at all of matching up the girls, the following „marriage
condition‟ must hold:
„Every sub set of girls marry at least as large a set of boys‟.
For example, we cannot find a matching if some 4 girls like only 3 boys. Hall‟s theorem
says that this necessary condition is actually sufficient, if the marriage condition holds,
then a matching exist.
Hall’s theorem: Let G= (V1 V2, E) be a bipartite graph such that V1 V2 . Ten G has
a complete matching saturating every vertex of V1 if an only if S N(S) for every set
S V1, where N(S) is the neighborhood subset of V2 that are adjacent to some vertex in
S, i.e.
N(S) = v V2 : there exist u S, {u, v} E .
A graph has many minimal coverings and they may be different sizes. The number of
vertices in a minimal covering of the smallest size is called vertex covering number and
denoted by V(G).
Example 1: In fig-(a) we mark a vertex covering of size 2 and in Fig-(b) show a vertex
covering of size 4. Hence V(G) = 2.
Since no vertex can cover two edges of a matching, the size of every vertex covering is at
least the size of every matching. Therefore, a matching and a vertex covering of the same
size prove that each is optimal. But such proofs exist bipartite graphs only.
Example 2: In the graph of Fig-(a) we mark a vertex covering of size 2 and show a
matching of size 2 in heavy lines. The optimal values are same since the graph is
bipartite. But in Fig-(b) the optimal values differ by 1.
Properties:
1. For any graph G the set of its vertices is trivially a vertex covering of G.
2. Every vertex covering contains a minimal covering.
3. For any graph G(V, E), V(G) + maximum independent set = number of
vertices in V.
4. For a complete bipartite graph Gm,n has V(Gm,n) = min{m, n}.
Independent sets and Covering: Independent number of a graph is the maximum size of
an independent set of vertices. No vertex covers two edges of a matching. Similarly, no
edge contains two vertices of an independent set. This yields another dual covering
Example 3: In fig.- a graph and two of its minimal covering are shown in heavy lines.
Properties:
1. For any graph G the set of its vertices is trivially a vertex covering of G.
2. An edge covering exist if and only if the graph has no isolated vertex.
3. An edge covering of an n-vertex graph will have at least n/2 edges. ( x
denotes the smallest integer not less than x).
4. Every pendant edge in a graph is included in every edge covering of the
graph.
5. Every edge covering contains a minimal covering.
6. The set of edges E(G) – L(G) is an edge covering if and only if for every
vertex v, the degree of vertex in G(V, E-L) (degree of vertex v in G(V, E)) –
1.
7. A minimal covering of an n-vertex graph can contain no more than n-1 edges.
8. For a complete bipartite graph Gm,n has E(Gm,n) = max{m, n}.
Objective Questions:
1. How many vertices does a regular graph of degree 4 with 10 edges have? Ans 5
Permutations - The different arrangements that can be made with a given number of
things taking some or all of them at a time are called permutations. The symbol n P r or
P(n,r) is used to denote the number of permutations of n things taken r at a time.
permutation is the arrangement of objects with ordering of same and different objects in
different ways. Permutation is always greater than combination as in combination we
neglect the ordering of the same objects in different ways, so as peter said abve ordering
is very important in permutation.
When things are arranged in places along a line with first and last place, they form a
linear permutation. So far we have dealt only with linear permutations. When things are
arranged in places along a closed curve or a circle, in which any place may be regarded as
the first or last place, they form a circular permutation.
The permutation in a row or along a line has a beginning and an end, but there is nothing
like beginning or end or first and last in a circular permutation. In circular permutations,
we consider one of the objects as fixed and the remaining objects are arranged as in linear
permutation.
Thus, the number of permutations of 4 objects in a row = 4!, where as the number of
circular permutations of 4 objects is (4-1)! = 3!.
The following arrangements of 4 objects O1, O2, O3, O4 in a circle will be considered as
one or same arrangement.
Observe carefully that when arranged in a row, O1 O2 O3 O4, O2O3O4 O1, O3O4O1O2,
O4O1O2O3 are different permutations. When arranged in a circle, these 4 permutations
are considered as one permutation.
Theorem:
The number of circular permutations of n different objects is (n-1)!.
Proof:
Each circular permutation corresponds to n linear permutations depending on where we
start. Since there are exactly n! linear permutations, there are exactly permutations.
Hence, the number of circular permutations is the same as (n-1)!.
Suggested answer:
The arrangement is as shown in the figure, the boy X will have B2, B3 as neighbours. The
girl Y will have G2, G3 as neighbours. The two boys B2, B3 can be arranged in two ways.
The two girls G2, G3 can be arranged in two ways.
Hence, the total number of arrangements = 2 x 2 = 4.
Example 2:
In how many different arrangements can 6 gentlemen and 6 ladies sit around a table if
Suggested answer:
i) Here, the total number = 6 + 6 = 12.
Then, corresponding to each seating arrangement for the gentlemen, the 6 ladies can be
seated in 6! ways.
The required number of arrangements = (5!)(6!)
= (120)(720) = 86400
1.1.7 Experiment:
What is the probability that all three children in a family have different birthdays?
(Assume, 1 year = 365 days).
Answer: The 1st child may be born on any of 365 days of the year, the 2nd also on any of
the 365 days and similarly the 3rd child. Hence, the total number of possible ways in
which the 3 children have birth days 365 365 365. These cases are mutually
exclusive, exhaustive and equally likely.
1.1.11 Problem:
Among twenty-five articles, nine are defective, six having only minor defects and three
having major defects. Determine the probability that an article selected at random has
major defects given that it has defects.
Solutions: Hint. Let MD-major defects, and D-defects
P( MD D)
Then we have to find P[MD/D] = =(3/25)/(9/25)
P( D)
= 3/9 = 1/3.
1.2 Total probability and example:
1.2.1 Theorem:
(i) If two events A and B are mutually exclusive, then the probability of occurrence of
either A or B is given by the sum of their probabilities, i.e.
Probability of (A or B ) = Probability of A + Probability of B .
In symbols,
P(A + B) = P(A) + P(B).
Proof : Let us suppose that a random experiment has n possible outcomes, which are
mutually exclusive ,exhaustive and equally likely .If m1 of these cases are favorable to
1.2.2 Experiment:
A card is drawn from a well-shuffled pack of playing cards. What is the probability that it
is either a spade or an ace?
Solution: The sample space S of drawing a card from a well-shuffled pack of playing
cards consists of 52 sample points.
If A and B denote the events of drawing a „spade card‟ and „an ace‟ respectively
then A consists 13 sample points and B consists of 4 sample points. So that,
13 4
P( A) and P(B) .
52 52
The compound event A∩B consists of only one sample point, i.e. ace of spade,
so that,
1
P( A B)
52
The probability that the card drawn is either a spade or a ace is given by
P(AUB) = P(A) + P(B) – P(A∩B)
13 4 1 4
=
52 52 52 13