1462360172E textofChapter7Module3
1462360172E textofChapter7Module3
Chapter 7
Graph properties
By
A. Satyanarayana Reddy
Department of Mathematics
Shiv Nadar University
Uttar Pradesh, India
E-mail: [email protected]
2
Objectives
Definition 1. A simple graph is said to be planar if it can be drawn in the plane in such a way that
edges do not cross.
2. Since K4 is isomorphic to W3 , all graphs with four or less vertices are planar.
Verify that the two graphs K5 and K3,3 have the following properties.
3. The graph obtained be removing exactly one edge or a vertex from K5 or K3,3 is a planar
graph.
• A planar graph divides the plane into various connected regions, one of which is called the
exterior region. Every region, including the exterior, is bounded by edges.
3
R1 R2
4
R4 , the exterior region
R3
1 2
• The degree of a region is defined as the number of edges which forms its boundary.
• The sum of the degrees of all regions is equal to twice the number of edges.
Since each edge is on the boundary of exactly two regions.
• A tree and a null graph determines just one region in plane, the exterior region, and every
edge of the tree is a part of the boundary of this region.
Theorem 3 ( Euler’s theorem). Let X be a connected simple planar graph with |V | vertices, |E|
edges, and R regions. Then, |V | − |E| + R = 2.
Proof. We use induction on |E|, the number of edges. If |E| = 0, then X is K1 , a graph with 1 vertex
and 1 region. So, |V | − |E| + R = 1 − 0 + 1 = 2, and the result is true. If |E| = 1, then the number of
vertices in X is two and the number of regions is 1. Therefore, |V | − |E| + R = 2 − 1 + 1 = 2 and
again the result is true. Assume that the result is true for all connected simple planar graphs with
fewer than |E| edges and let et X be a simple connected planar graph with |E| edges.
Case I: Suppose X is a tree. Then |E| = |V | − 1 and R = 1, as a planar representation of a tree has
only one region. Thus, |V | − |E| + R = |V | − (|V | − 1) + 1 = 2, and the result holds.
Case II: Suppose X is not a tree. Then, X has at least one cycle. Let C be a cycle in X and let e
be an edge of C. The graph X \ e has one edge less than the number of edges in the graph X.
Also, the number of vertices in X \ e and X are same. But, the removal of the edge e merges
4
the two regions in X into one in X \ e. Therefore, X \ e has one region less than that in X.
Thus, by the induction hypothesis, in X \ e, we have |V | − (|E| − 1) + (R − 1) = 2, so that
|V | − |E| + R = 2. Hence the result.
Corollary 4. Let X = (V, E) be a connected simple planar graph with |V | ≥ 3 vertices and |E|
edges. Then, |E| ≤ 3|V | − 6.
Proof. Since in a simple planar graph every region is bounded by at least three edges, we have
3R ≤ 2|E|. Thus, using Euler’s theorem, we see that 3(2 − |V | + |E|) ≤ 2|E| ⇒ |E| ≤ 3|V | − 6.
Proceeding on the lines of the proof of Corollary 4, we obtain the following result.
Corollary 5. Let X be a simple connected planar graph and let the length of the smallest cycle
r
(girth) in X be r. Then, |E| ≤ (r−2) (|V | − 2).
Corollary 6. Every simple planar graph contains at least one vertex of degree less than or equal to
5.
Proof. Suppose that the degree of every vertex is grater than or equal to six (that is, δ (X) ≥ 6).
Then, using Corollary 4, we get 6|V | ≤ 2|E| ≤ 6|V | − 12, a contradiction.
The degree of each vertex of the complete graph Kn is n − 1. Hence, Kn , for n ≥ 7, is not a
planar graph.
Proof. If K5 is planar, then it has to satisfy |E| ≤ 3|V | − 6. But, in K5 , note that 10 = |E| ≤
3|V | − 6 = 15 − 6 = 9, a contradiction.
Thus, we note that the graph K6 is non-planar as well. Now, we use Corollary 5 to prove the
following result.
5
Example 8. Show that K3,3 and the Petersen graphs are non-planar.
Proof. Since K3,3 is a bipartite graph, it has no triangles. So, its girth is ≥ 4 (check that it is exactly
4
4). Hence, if K3,3 is planar, then it has to have at most (4−2) (6 − 2) = 8 edges. But K3,3 has 9 edges.
Thus, K3,3 is non-planar.
We know that the girth of the Petersen graph is 5. Thus, a planar drawing of the Petersen graph
5 40
can have at most 5−2 (10 − 2) = 3 edges, a contradiction, as the Petersen graph has 15 edges.
c1
c2 c6
c7 c10 c5
c8 c9
c3
c4
Definition 9. 1. Let X be graph and v be a vertex of degree 2. Suppose N(v) = {u, w} and
u 6= w. Then, the edges {u, v}, {v, w} are said to be in series. Deleting the vertex v and
replacing the edges {u, v}, {v, w} by {u, w} is called a series reduction.
3. Two graphs are said to be homeomorphic if each graph can be obtained from the same
graph by a sequence of series reductions or equivalently obtained from the same graph by a
sequence of elementary subdivisions.
For example, any two cycles are homeomorphic. Since K2,2 is isomorphic to C4 , it is homeo-
morphic to K3 or C3 .
6
Theorem 10 (Kuratowski’s theorem). A simple graph is planar if and only if it has no sub graph
homeomorphic to either K5 or K3,3 .
Example 11. Show that the graph on the left (see the figure below) is non-planar.
c6 c6
c7 c5 c7
c4 c4
c3 c3
c8 c8
c1 c2 c1
Solution:It is easy to see that given graph is homeomorphic to the graph that appears on the right
and it contains K5 as a subgraph.
Exercises 12. Show that the Petersen graph contains a subgraph which is a subdivision of K3,3 .
Dual of given planar Graph: Given a planar graph X (need not be a simple graph), we can define
another multi graph X* as follows:
• Corresponding to each edge e of X there is an edge e* of X* i.e., two vertices r* and s* are
joined by the edge e* if and only if their corresponding regions r and s are separated by the
edge e in X.
• A loop is added at a vertex r* of X* for each cut edge of X that belongs to the boundary of
the region r.
7
Let |E|*, |R|*, |V |* and |E|, |R|, |V | denote the number of edges, regions, and vertices of X* and X,
respectively. Then,
• |E|*=|E|, |V |*=|R|.
• X* is planar
Theorem 13. Let X be a connected planar graph. Then, X is isomorphic to its double dual, (X ∗ )∗ .
Proof. Let X be a plane connected graph and X* be the dual of X. Now, any region r of the dual
X* contains exactly one vertex of X, namely its corresponding vertex v in X. This is because the
number of the regions of X* is same as the number of vertices of X. Thus, in the construction of
X**, we take the vertex v to be the vertex in X** corresponding to the region r of X*. This choice
gives us the required isomorphism.
Definition 14. A connected graph X is called self-dual if it is isomorphic to its dual X*.
For example, K4 is self-dual. The next result is a characterization of bipartite graphs in terms of
duality.
Theorem 15. A simple connected planar graph X is bipartite if and only if its dual graph X* is
Eulerian.
8
Proof. Let X be a simple connected bipartite graph. Then, X does not contain odd cycles. Since
X is a planar graph, all the regions of X are of even length. So, the degree of every region in
X is even. Let X* be the dual graph of X. Since the vertex degrees of X* are the same as the
corresponding degrees of the regions of X, the degree of every vertex in X* is even. Hence X* is
Eulerian. Conversely, assume X* is Eulerian. Then, degree of every vertex in X* is even. Since the
degree of a vertex in X* corresponds to the degree of the region in X, the degree of every region in
X is even. This shows that X contains only even cycles. Hence, X is bipartite.
Coloring of a Graph
A k-coloring (or a k-vertex coloring) of a graph X is a mapping α : V (X) → [k], where [k] =
{1, 2, . . . , k}. The coloring is said to be proper, if the adjacent vertices obtain distinct colors. That
is, for all {u, v} ∈ E(X), α(u) 6= α(v). A color i ∈ [k] said to be available for a vertex v, if no
neighbor of v is colored by i. A graph X is said to be k-colorable, if there is a proper k-coloring for
X. The (vertex) chromatic number, denoted by χ(X), is defined as
1. χ(X) ≤ |V |
2. χ(Kn ) = n
4. If deg(v) = d, then at most d colors are required to color the vertices adjacent to v.
9
Proof. Let X be any graph with n vertices. To prove the result, we use induction on n. For
n = 1, X = K1 and χ(X) = 1 and ∆(X) = 0. Therefore, the result is true for n = 1.
Assume that the result is true for all graphs with n−1 vertices and let X be a graph on n vertices.
Therefore, by induction hypothesis, for a fixed vertex v of X, χ(X \ {v}) ≤ ∆(X \ {v}) + 1.
This shows that X \ {v} can be colored by using ∆(X \ {v}) + 1 colors. Since, ∆(X) is the
maximum degree of a vertex in X, vertex v has at most ∆(X) neighbors in X. Thus, these
neighbors use up at most ∆(X) colors in the coloring of X \ {v}.
If ∆(X) = ∆(X \ {v}), then there is at least one color not used by v’s neighbors and that can
be used to color the vertex v giving a ∆(X) + 1 coloring for X.
In case ∆(X) 6= ∆(X \ {v}), then ∆(X \ {v}) < ∆(X). Therefore, using a new color for v, we
have a ∆(X \ {v}) + 2 coloring of X and clearly, ∆(X \ {v}) + 2 ≤ ∆(X) + 1. Hence, in both
cases, it follows that χ(X) ≤ ∆(X) + 1.
• X is 2-colorable.
• X is a bipartite graph.
c1
c2 c6
c10 c5
c7
c8 c9
c3
c4
Welsh-Powell algorithm:
• Use the 1st color to color the first vertex and to color, in sequential order, each vertex which
is not adjacent to the previously colored vertex.
• Repeat the process using the second color and the remaining uncolored vertices. Continue the
process with third color and so on.
The following result establishes every planar graph is five colorable. This statement means that
if X is a planar graph, then χ(X) ≤ 5. In fact the famous four color theorem states that every planar
graph is four colorable.
Proof. We prove the result by induction on the number of vertices. Since every graph with five
or less vertices need at most five colors, hence the result is true for the base case. Let X be a
planar graph with k ≥ 5 vertices and assume that all planar graphs with less than k vertices are five
colorable. Since, X is a planar graph, by Corollary 6, it contains a vertex of degree less than or
equal to 5. Let this vertex be v. Then, by the induction hypothesis X \ {v} is five colorable. Now,
11
we need to color v. If deg(v) ≤ 4, we are done, as the neighbors of v are colored with at most four
colors. If deg(v) = 5 and its neighbors need only four colors, then also we are done. The only case
left is: deg(v) = 5 and the neighbors of v need five colors.
We will show that this case is not possible, by showing that this condition will lead to a non-
planar graph. Suppose N(v) = {a, b, c, d, e}, the neighbors of v. Also, suppose that we have used
the colors 1, 2, 3, 4 and 5 to color the vertices a, b, c, d and e, respectively. Now, in the graph X \ {v},
there is a path from a to c whose vertices are alternately colored with 3 and 1 (see the figure given
below). Now, without loss of generality, we assume that the cycle obtained by the union of the
alternating path from a to c and the edges {a, v} and {v, c} contains the vertex b. Then, we cannot
have a similar path, in the graph X \ {v}, in between the vertices b and d as X is a planar graph.
Hence, b and d can be colored with the same color, say 2. Thus, we are left with the color 4 which
can be used to color the vertex v. Hence, the required result follows.
3
color-1 a 1
a
3
color-5 color-2 e
e b b 1
v v
3
c
d d 1
color-4 color-3 c