GRAPH THEORY (BCS405B) 2024

MODULE-4
Planar Graphs: Definition, Kuratowski’s theorem (proof not required), Different representations of planar
graphs, Euler's theorem, and Geometric dual.
Graph Representations: Matrix representation of graphs-Adjacency matrix, Incidence Matrix, Circuit Matrix,
Path Matrix.

Planar Graphs: A Graph G is said to be planar if there exists some geometric representations of G which can be
drawn on a plane such that no two of its edges intersect between the vertices. That mean edges intersects at
vertex only.

A drawing of a geometric representation of a graph on a plane such that no two of its edges intersect is called
embedding. If there is no embedding of a graph on a plane then the graph is non-planar.

An embedding of a planar graph 𝐺 on a plane is called plane representation of 𝐺.

Complete graph 𝐾𝑛 is planar for 𝑛 ≤ 4 and non-planar for 𝑛 ≥ 5.

Planar embedding of 𝐾𝑛 for 𝑛 ≤ 4 are given below.

𝐾1 𝐾2 𝐾3 𝐾4

Kuratowski’s two graphs:

1. Complete graph 𝐾5 is Kuratowski’s first graph.
2. Complete bipartite graph 𝐾3,3 is Kuratowski’s second graph.
Properties of Kuratowski’s two graphs:
1. Both are regular Graphs
2. Both are non-planar.
3. Removal of one edge or a vertex makes each planar graph.
4. Kuratowski’s first graph is the non-planar graph with minimum number of vertices.
Kuratowski’s second graph is the non-planar graph with minimum number of edges.
Note: Any simple planar graph can be embedded in a plane such that every edge is drawn as a straight line
segment.
Region: A plane representation of a graph divides the plane into regions (also called windows, faces, or
meshes). A region is characterized by the set of edges forming its boundary.
Degree of a region is defined by number of boundary edges of the region (an edge with one end vertex is inside
the region is considered as two boundary edges for that region.
Since each edge is boundary for two regions or two boundary for same region, 2𝑒 = ∑ deg(𝑓𝑖 )
Note: A region is not defined in a non-planar graph, or even in a planar graph not embedded in a plane. Thus a
region is property of the specific plane representation.
Infinite region: The portion of the plane lying outside an embedded graph is called infinite region.

DEPARTMENT OF CUMPUTER SCIENCE & ENGINEERING /C.E.C.

1
GRAPH THEORY (BCS405B) 2024

The following graph has 5 regions, the region 3 is infinite region.

3
2
1 4 5

Clearly 𝑑𝑒𝑔(1) = 2 , 𝑑𝑒𝑔(2) = 3, 𝑑𝑒𝑔(3) = 6, 𝑑𝑒𝑔(4) = 6, 𝑑𝑒𝑔(5) = 1, and ∑ deg(𝑓𝑖 ) = 18 = 2𝑒 .

That is, the region 1 is bounded by 2 boundary edges, the regions 2, 3, 4 and 5 are bounded by 3, 6, 6 and 1
boundary edges respectively.

A planar graph may be embedded in a plane such that any specified region can be made the infinite region.

Euler’s formula: A connected planar graph with 𝑛 vertices and 𝑒 edges has 𝑒 − 𝑛 + 2 regions.
Or, for any plane representation, number of regions, 𝑓 = 𝑒 − 𝑛 + 2.

Proof: To prove the Euler’s formula it is sufficient to consider simple planar graph, because adding a self-loop
or a parallel edge increases 𝑓 and 𝑒 by 1. We can also disregard the edges entirely inside a region.
(or edges not the boundary for two regions), because 𝑒 and 𝑛 decreases by one, and 𝑒 − 𝑛 unaltered.
Now consider the plane representation of graph such that each edge is a straight line and each region is a
polygon. Let the polygonal net representing the given planar graph consists of 𝑓 regions, and 𝑘𝑝 be the
number of 𝑝-sided regions ( or the number of regions with degree 𝑝).
There is no loop and parallel edges, each region is bounded by at least 3 edges.
Since the each edge is on the boundary of exactly two regions, and ∑ deg(𝑓𝑖 ) = 2𝑒,
3𝑘3 + 4𝑘4 + 5𝑘5 ⋯ ⋯ 𝑟𝑘𝑟 = 2𝑒 , where 𝑘𝑟 is the number of regions with maximum(𝑟) edges, and
𝑘3 + 𝑘4 + 𝑘5 ⋯ ⋯ 𝑘𝑟 = 𝑓 .
WKT, sum of all interior angles of 𝑝-sided polygon is (𝑝 − 2)𝜋 ,
and sum of exterior angle is (𝑝 + 2)𝜋 = (𝑝 − 2)𝜋 + 4𝜋.
Since sum of all angles subtended at each vertex in the polygonal net is 2𝑛𝜋, this sum is sum of all interior
angles of (𝑓 − 1) interior regions plus sum of exterior angle of infinite region. That is sum of all interior
angles of all regions +4𝜋 .
Therefore,
𝜋[(3 − 2)𝑘3 + (4 − 2)𝑘4 + (5 − 2)𝑘5 ⋯ ⋯ + (𝑟 − 2)𝑘𝑟 ] + 4𝜋 = 2𝑛𝜋
⟹ [(3𝑘3 + 4𝑘4 + 5𝑘5 ⋯ ⋯ 𝑟𝑘𝑟 ) − 2(𝑘3 + 𝑘4 + 𝑘5 ⋯ ⋯ 𝑘𝑟 )] + 4 = 2𝑛
⟹ [2𝑒 − 2𝑓] + 4 = 2𝑛 ⟹ 𝑒 − 𝑓 + 2 = 𝑛,
Or, 𝑓 = 𝑒 − 𝑛 + 2.
Exercise:
1. Prove that Kuratowski’s first graph is non-planar. (𝐾5 is non-planar).
Proof: For 𝐾5 , 𝑒 = 10, 𝑛 = 5.
Suppose Kuratowski’s first graph, 𝐾5 is planar.
Then there exists an embedding of 𝐾5 , and number of regions, 𝑓 = 𝑒 − 𝑛 + 2 = 7.
Since 𝐾5 is simple graph, each region is bounded by at least 3 edges. (Degree of each region is at least 3).
2𝑒 = ∑ 𝑑𝑒𝑔(𝑓𝑖 ) ⟹ 2𝑒 ≥ 3𝑓 ⟹ 20 ≥ 3 × 7 ⟹ 20 ≥ 21 .
DEPARTMENT OF CUMPUTER SCIENCE & ENGINEERING /C.E.C.
2
GRAPH THEORY (BCS405B) 2024

Which is impossible. Therefore 𝐾5 is non-planar.

1. Prove that Kuratowski’s second graph is non-planar. (𝐾3,3 is non-planar).
Proof: For 𝐾3,3 , 𝑒 = 9, 𝑛 = 6.
Suppose Kuratowski’s second graph, 𝐾3,3 is planar.
Then there exists an embedding of 𝐾3,3, and number of regions, 𝑓 = 𝑒 − 𝑛 + 2 = 5.
Since 𝐾3,3 is simple graph, and not containing triangle, each region is bounded by at least 4 edges.
⟹ 2𝑒 ≥ 4𝑓 ⟹ 18 ≥ 4 × 5 ⟹ 18 ≥ 20 .
Which is impossible. Therefore 𝐾3,3 is non-planar.
Elementary reduction of a graph for detection of planarity:
1. For a disconnected graph, consider only one component at a time.
2. For a separable graph, consider only one non-separable block at a time.
3. Remove all self-loops.
4. Remove all parallel edges but keep one edge between every pair of edges.
5. Merge the series edges. (Remove the vertex with degree two and replace two series edges by a single edge)
Repeat the step 4 and 5.
Then we get one of the following graph.
1. A single edge.
2. Complete graph, 𝐾4 or
3. A non-separable, simple graph with 𝑒 ≥ 7, 𝑛 ≥ 5.
Homeomorphic graphs: Two graphs are said to be homeomorphic if either two graphs are isomorphic or one
graph can be obtained from other by the creation of edges in series or by the merger of edges in series.
Kuratowski’s theorem: A necessary and sufficient condition for a graph G to be planar is that G does not
contain either of Kuratowski’s two graphs or any sub-graph homeomorphic to either of them.
Therefore, if G contains a sub-graph which is homeomorphic to either 𝐾5 or 𝐾3,3 then G is non-planar.
1. Prove that following graphs are planar by elementary reduction.

DEPARTMENT OF CUMPUTER SCIENCE & ENGINEERING /C.E.C.

3
GRAPH THEORY (BCS405B) 2024

2. Show that following graph is planar.

3. Using Kuratowski’s theorem prove that following graphs are not planar.

DEPARTMENT OF CUMPUTER SCIENCE & ENGINEERING /C.E.C.

4
GRAPH THEORY (BCS405B) 2024

Geometric Dual of a planar graph: Consider a plane representation of a planar graph. To construct a dual
relative to a particular embedding, place a vertex inside each region including infinite region (Consider 𝑓
vertices). For each edge shared by two regions, draw an edge connecting the vertices inside these regions. If an
edge is two boundary edge for one region, draw a self-loop at the vertex inside the region. By this procedure we
obtain a new graph 𝐺 ∗ called a dual of 𝐺.
Example: Draw the geometric dual of the graph 𝐺

Graph 𝐺

Graph 𝐺 ∗

Relationship between a planar graph 𝐺 and its dual 𝐺 ∗ .

Let 𝑛, 𝑒, 𝑓, 𝑟 and 𝜇 denote the number of vertices, edges, regions, rank and nullity of a planar graph 𝐺, and
𝑛∗ , 𝑒 ∗ , 𝑓 ∗ , 𝑟 ∗ and 𝜇 ∗ are the corresponding numbers in dual graph 𝐺 ∗ .
1. 𝑛∗ = 𝑓, 𝑒 ∗ = 𝑒, 𝑓 ∗ = 𝑛, 𝑟 ∗ = 𝜇 , 𝜇 ∗ = 𝑟.
2. An edge forming a self-loop in 𝐺 yields a pendent edge in 𝐺 ∗ .
3. A pendent edge in 𝐺 yields a self-loop in 𝐺 ∗ .
4. Series edges in 𝐺 produce parallel edges in 𝐺 ∗ .
5. Parallel edges in 𝐺 produce edges in series in 𝐺 ∗ .
6. Degree of a region in 𝐺 (number of boundary edges of a region) is equal to degree of the corresponding
vertex in 𝐺 ∗ and vice versa.
7. The graph 𝐺 ∗ is also planar.
8. A circuit with 𝑘 edges in 𝐺 corresponds to a cut-set with 𝑘 edges in 𝐺 ∗ vice versa.

Note: Duals of two isomorphic graphs (two embedding of a planar graph) need not be isomorphic.
Find the duals of following two isomorphic graphs and show that duals are not isomorphic.

DEPARTMENT OF CUMPUTER SCIENCE & ENGINEERING /C.E.C.

5
GRAPH THEORY (BCS405B) 2024

Matrix representation of graphs:

Adjacency matrix (or connection matrix): Adjacency matrix of a graph G with n vertices and no parallel
1, 𝑖𝑓 the vertices 𝑣𝑖 and 𝑣𝑗 are adjacent.
edges is 𝐴𝑛×𝑛 matrix defined by 𝐴 = 〈𝑎𝑖𝑗 〉, where 𝑎𝑖𝑗 = {
0, if 𝑣𝑖 and 𝑣𝑗 are not adjacent.

Example:
𝑣1 𝑣2

𝑣3

𝑣5 𝑣4

𝑣1 𝑣2 𝑣3 𝑣4 𝑣5
𝑣1 0 1 0 1 1
𝑣2 1 0 1 1 1
Adjacency matrix 𝐴 = 𝑣3 0 1 1 1 0
𝑣4 1 1 1 0 1
𝑣5 [1 0 0 1 0]
Note:

1. The principal diagonal entries of adjacency matrix are all zero if and only if the graph has no self-loop.

2. For the simple graph, degree of a vertex equals the number of 1’s in the corresponding row or column of
adjacency matrix.
Incidence Matrix: Incidence matrix of a graph 𝐺 with 𝑛 vertices, 𝑒 edges and no self-loop is 𝐵𝑛×𝑒 matrix
1, 𝑖𝑓 the vertex 𝑣𝑖 is incident from the edge 𝑒𝑗 .
defined by 𝐵 = 〈𝑏𝑖𝑗 〉, where 𝑏𝑖𝑗 = {
0, Otherwise.
𝑣1 𝑒4 𝑣5

𝑒5

𝑒1 𝑒6 𝑣4
𝑒3

𝑒7

𝑣2 𝑣3
𝑒2

DEPARTMENT OF CUMPUTER SCIENCE & ENGINEERING /C.E.C.

6
GRAPH THEORY (BCS405B) 2024

𝑒1 𝑒2 𝑒3 𝑒4 𝑒5 𝑒6 𝑒7
𝑣1 1 0 1 1 0 0 0
𝑣2 1 1 0 0 0 0 0
Incidence matrix 𝐵 = 𝑣3 0 1 1 0 0 1 1
𝑣4 0 0 0 0 1 0 1
𝑣5 [0 0 0 1 1 1 0]
Note:

1. Since every edge is incident on exactly two vertices, each column of 𝐵 has exactly two 1’s.

2. Degree of a vertex is equals the number of 1’s in the corresponding row.

3. A row with all 0’s represents an isolated vertex.

4. Parallel edges in a graph produce identical columns in its incidence matrix.

5. Two graphs are isomorphic if and only if their incidence matrices differ only by permutations of rows and
columns.

Fill up the missing entries in the given incidence matrix and draw the graph.

𝑒1 𝑒2 𝑒3 𝑒4 𝑒5 𝑒6 𝑒7
𝑣1 0 1 0 1 0 1 0
𝑣2 1 0 1 0 1 0 0
Incidence matrix 𝐵 = 𝑣3 0 1 1 0 0 0 1
𝑣4 . . .. .. .. .. .. ..
𝑣5 [0 0 0 1 0 1 0]
Circuit Matrix: Let the number of different circuits be 𝑞 and number of edges 𝑒.

1, if the 𝑗th edge is in 𝑖th circuit .

The circuit matrix 𝐵𝑞×𝑒 is defined by 𝐵 = 〈𝑏𝑖𝑗 〉 where 𝑎𝑖𝑗 = {
0, Otherwise.

In the previous example there are six circuits 𝑐1 = {𝑒1 , 𝑒2 , 𝑒3 } , 𝑐2 = {𝑒3 , 𝑒4 , 𝑒6 }, 𝑐3 = {𝑒5 , 𝑒6 , 𝑒7 },

𝑐4 = {𝑒1 , 𝑒2 , 𝑒6 , 𝑒4 }, 𝑐5 = {𝑒3 , 𝑒4 , 𝑒5 , 𝑒7 } and 𝑐6 = {𝑒1 , 𝑒4 , 𝑒5 , 𝑒7 , 𝑒2 } .

𝑒1 𝑒2 𝑒3 𝑒4 𝑒5 𝑒6 𝑒7
𝑐1 1 1 1 0 0 0 0
𝑐2 0 0 1 1 0 1 0
𝑐3
Circuit matrix 𝐵 = 𝑐 0 0 0 0 1 1 1
4 1 1 0 1 0 1 0
𝑐5 0 0 1 1 1 0 1
𝑐6 [1 1 0 1 1 0 1]

DEPARTMENT OF CUMPUTER SCIENCE & ENGINEERING /C.E.C.

7
GRAPH THEORY (BCS405B) 2024

Path Matrix: Path matrix is defined for a specific pair of vertices in a graph. Let 𝐺 be a connected graph with e
edges and suppose there are 𝑘 paths between the two vertices 𝑥 and 𝑦. Then the path matrix 𝑃(𝑥, 𝑦) is 𝑘 × 𝑒
1, if the 𝑗th edge lies in 𝑖th path
matrix defined by 𝑃(𝑥, 𝑦) = 〈𝑝𝑖𝑗 〉 where 𝑝𝑖𝑗 = { .
0, Otherwise

Example: There are 4 paths between the vertices 𝑣1 and 𝑣2 . They are {𝑒3 }, {𝑒1 , 𝑒2 }, {𝑒4 , 𝑒6 } and {𝑒4 , 𝑒5 , 𝑒7 }.

𝑣1 𝑒4 𝑣5

𝑒5

𝑒1 𝑒6 𝑣4
𝑒3

𝑒7

𝑣2 𝑣3
𝑒2

𝑒1 𝑒2 𝑒3 𝑒4 𝑒5 𝑒6 𝑒7
1 0 0 1 0 0 0 0
Path matrix 𝑃(𝑣1 , 𝑣2 ) = 2 [1 1 0 0 0 0 0]
3 0 0 0 1 0 1 0
4 0 0 0 1 1 0 1

DEPARTMENT OF CUMPUTER SCIENCE & ENGINEERING /C.E.C.

8