Graph Theory – Lecture 8

In this lecture, we will investigate the concept of a planar graph. We start with its definition.

Definition: A graph G is planar if the graph can be drawn in the plane without edges intersecting, except
at their endpoints. In this case, we say that the graph is embedded in the plane.

Here are some examples.

Example 1: Is the following graph planar? This is K 4 .

Your first reaction might be, this graph is not planar. There are edges that cross each other. The point is,
and this can be difficult to figure out, is there some way of drawing the graph so that the edges are not
intersecting?

This graph can be drawn without edge crossings, as follows.

This shows that K 4 is planar.

Example 2: Is the following graph planar?

Looking carefully at the graph, we can see that the vertices F and G are the problem. We can move F to
the left of B and G to the right of E. That shows the original graph is planar.

Example 3: Is the following graph planar? This is K5 .

We saw in Example 1 that K 4 is planar. We can analyze K5 by adding a new vertex to K 4 and
connecting that vertex to the 4 existing vertices.

There are two cases.

Case 1: The new vertex is inside K 4 .

That would give us a picture like this:

Vertex E is the next vertex. To create K5 , we must connect E to each of the other vertices. Clearly, we
can connect E to B,C,D without any edge crossings. But there is no way to connect E and A without
intersecting edges.

Case 2: The new vertex is outside K 4 .

Vertex E is the new vertex. It can be connected to vertices A,B,C without edge crossings. But E and D
cannot be connected without intersecting edges.
Therefore, K5 is nonplanar.

Example 4: Is the following graph planar? This is K 2,3 .

We can simply redraw the graph so that A,B are on opposite sides of C,D,E

This shows that K 2,3 is planar.

Example 5: Is the following graph planar? This is K 3,3 .

We showed that K 2,3 is planar. We can analyze K 3,3 by adding a vertex to K 2,3 and connecting it to the
vertices on the other side.

Case 1: The new vertex is inside K 2,3 .

Vertex F is the new vertex. F can be connected to A,B,C,D without edge crossings. But F and E cannot be
connected without intersecting edges.
Case 2: The new vertex is outside K 2,3 .

Vertex F is the new vertex. You can connect F to vertices A,B,C,E without edge crossings. But you cannot
connect F and D without intersecting edges.

This proves that K 3,3 is not planar.

As we will see later in the lecture, K5 and K 3,3 are the simplest nonplanar graphs. And in fact, all
nonplanar graphs contain, in some sense, copies of K5 and K 3,3 .

The following definition is useful, but can be confusing.

Definition: A graph that is drawn without any edge crossings is called a plane graph.

This leads to the obvious question, what is the difference between a planar graph and a plane graph? A
planar graph is one that can be drawn without edge crossings. A plane graph is one that is drawn
without edge crossings. Take a look back at Example 1:
Both graphs are planar. Only the graph on the right is a plane graph (the graph on the left has edge
crossings).

We now consider an invariant of planar graphs, known as the Euler characteristic.

Definition: A planar graph divides the plane into distinct regions or faces. A face inside the graph is
called an interior face. There is one face outside the graph, called the exterior face.

Note: the textbook uses the word regions, while I prefer faces.

Example 6: Here is an example of a planar graph, with the faces labeled. There are 8 interior faces
labeled f1 ,..., f8 and the exterior face f 0 .

Here is a large category of planar graphs.

Theorem 8.1: A tree is a planar graph.

Proof: We will prove this by induction on n  the number of vertices in the tree.

Basis: n  1 . A graph with one vertex is clearly planar.

Induction hypothesis: Let n  1 . Assume that a tree with n vertices is planar.

Induction step: Prove that a tree with n  1 vertices is planar. Let T be a tree with n  1 vertices. By
Theorem 2.4, T will have n edges.
By Theorem 2.3, T has at least two leaves (i.e., vertices of degree 1). As shown in the proof of Theorem
2.3, the endpoints of a longest path will be leaves. Let a be such a leaf in T . Then T  a is a tree with
n vertices. Hence, T  a is planar by the induction hypothesis. Now add back the vertex a and its one
incident edge. It is possible to do this without any edges crossing. QED

Note, the concept of faces only makes sense for a plane graph. But this raises the question, for a planar
graph does the number of faces depend on how you draw the graph as a plane graph?

Interestingly enough, the answer is no. The number of faces does not depend on how you draw a planar
graph as a plane graph.

Let | V |  #vertices, | E |  #edges, | F |  #faces in a planar graph.

Definition: Let  | V |  | E |  | F | . This is called the Euler Characteristic.

The following theorem states that the Euler Characteristic is an invariant of all connected planar graphs.

Theorem 8.2: (Euler Polyhedral Formula)

Let G be a connected, planar graph. Then   | V |  | E |  | F |  2 .

Proof: First consider the case where G is a tree. Then | E | | V  1| and | F |  1 . The Euler characteristic
for a tree is   | V |  | E |  | F |  | V | (| V | 1)  1  2 .

Suppose that the theorem is not true for graphs that are not trees. Let G be a connected, plane graph
of smallest size for which the theorem does not hold. We then have | V |  | E |  | F |  2 . Since G is
not a tree, it has an edge e that is not a bridge (by Theorem 6.1). Consider the graph G  e . We know
that the size of G  e is less than the size of G .Therefore, the theorem holds for G  e .

G  e has order | V | , size | E |  1 , and number of faces | F | 1 . That last formula is not easy to prove,
but the basic idea is that, since e is not a bridge, it is on the boundary of two faces. So removing e
merges two faces. Hence G  e has one fewer face than G .

Using the Euler Polyhedral Formula on G  e , we get,

| V | (| E | 1)  (| F |  1)  2
|V |  | E |  | F | 2

The last formula contradicts our assumption that the theorem is not true for G . Therefore, the theorem
is true for all connected, planar graphs. QED

Let’s consider some planar graphs that we looked at previously.

Here | V | 4, | E | 6,| F | 4 , so | V |  | E |  | F |  2

Here | V | 7, | E | 10,| F | 5 , so | V |  | E |  | F |  2

Here | V | 11, | E | 18,| F | 9 , so | V |  | E |  | F |  2

Why is this called the Euler Polyhderal Formula? First of all, it is named after Leonhard Euler, whom we
encountered before. Second, Euler proved it for polyhedra.

What are polyhedra? These are three‐dimensional solids with faces bounded by straight edges and
vertices as the endpoints of the edges.

A convex polyhedron has the further property that if you connect any two points lying on different faces
of the surface by a straight line, the line (other than its endpoints) lies entirely in the interior of the
polyhedron.

The Platonic solids are examples of convex polyhedra (as shown below). Just as for planar graphs, we
have V  E  F  2 for all convex polyhedra.

Platonic Solids

Name Shape Vertices Edges Faces Euler

V E F Characteristic
Tetrahedron 4 6 4 2

Cube 8 12 6 2

Octahedron 6 12 8 2

Dodecahedron 20 30 12 2

Icosahedron 12 30 20 2

For a non‐convex polyhedron, you get a different value for the Euler characteristic. For example,

The tetrahemihexahedron which has 6 vertices, 12 edges, and 7 faces, for an Euler characteristic of 1.

Here is a 3‐d view of the tetrahemihexahedron and a 2‐d view of it unfolded.

In the unfolded view you can see there are 3 squares and 4 triangles, hence the 7 faces.
Let’s look at another non‐convex polyhedron, the octahemioctahedron. There are 4 hexagons (labeled
A,B,C,D) and 8 triangles (only 3 of which are visible). A flattened view is on the right.

Altogether, the octahemioctahedron has 12 vertices, 24 edges, and 12 faces. This gives an Euler
characteristic of 0.

There is an interesting relationship between the number of vertices and the number of edges in a planar
graph. There are several steps leading up the main theorem.

Definition: In a planar graph, the degree of a face is the length of the shortest closed walk around the
edges of the face.

This is a little trickier than it looks. Here are a couple of examples.

Example 7: Find the degree of each face in the graph below.

deg( f1 )  4 ( A, B, E , D, A)
deg( f 2 )  4 ( B, C , E , F , B )
deg( f 0 )  6 ( F , C , B, A, D, E , F )

Example 8: Find the degree of each face in the graph below.

deg( f1 )  8 ( A, B, C , F , E , G , E , D, A)
deg( f 0 )  6 ( F , C , B, A, D, E , F )

Theorem 8.3: Let G be a planar graph. Then,

2 | E |  deg( f )
f F

Here, the sum is taken over all the faces of the graph.

Proof: In walking around all the faces of the graph, each edge is encountered twice (once for each face
that shares that edge). Hence, the sum of the degrees of all the faces equals two times the number of
edges.
Consider this theorem applied to Example 7. There are 7 edges. We computed the sum of the degrees of
each face.

 deg( f )  4  4  6  14  2*7  2 | E | .
f F

Consider this theorem applied to Example 8. There are 7 edges. We computed the sum of the degrees of
each face.

 deg( f )  8  6  14  2*7  2 | E | .
f F

The following is known as the edge‐face inequality. It uses the girth of the graph, which is the length of
the shortest cycle in the graph.

Theorem 8.4: (Edge – Face Inequality) Let G be a connected, planar graph which is not a tree. Then,

2 | E |  girth(G ) | F | .

Proof: By Theorem 8.3, 2 | E |   deg( f ) .

f F

Since the boundary of each face contains a cycle, deg( f )  girth(G ) for each face f . There are | F |
number of faces. Therefore, 2 | E |   deg( f )  girth(G )  F .
f F
QED

Let’s apply this theorem to Example 7. There are 7 edges and 3 faces. Also, the shortest cycle is a 4‐
cycle.

2 | E |  2*7  14  4*3  12  girth(G ) | F | .

Here is the theorem we have been building up to.

Theorem 8.5: Let G be a connected, planar graph with at least 3 vertices. Then,

| E |  3 | V | 6 .

Proof:

Let’s first consider the case where G is a tree. Then | E |  | V | 1 . Then the inequality above becomes,

5
| V |  1 3 | V | 6  5  2 | V |   | V | . The last inequality is true, since G has at least 3 vertices.
2
Now assume that G is not a tree.

(1) |V |  | E |  | F | 2 (Euler Polyhedral Formula)

(2) girth(G ) | F |  2 | E | (Edge‐Face Inequality)

(3) 3 | F |  girth(G ) | F |  2 | E | (Since the shortest possible cycle is a 3‐cycle)

2
(4) | F | | E | (From (3))
3
2 1
(5) 2  | V |  | E |  | F |  | V |  | E |  | E | | V |  | E | (Substituting (4) into (1))
3 3
(6)

1
| E |  | V | 2
3 As was to be shown.
| E |3 |V | 6

Example 9: Show that K5 is non‐planar.

We have shown this before. But let’s prove this using Theorem 8.5

K5 has 5 vertices and 10 edges.

3 | V | 6  9  | E | .

Hence K5 does not satisfy the inequality in Theorem 8.5. That means K5 is non‐planar.

Example 10: Let G be a graph with 8 vertices and at least 19 edges. Is it possible for G to be planar?

3 | V | 6  18  | E | . Such a graph does not satisfy the inequality in Theorem 8.5. That means G is non‐
planar.

Example 11: Does Theorem 8.5 show that K 3,3 is non‐planar?

K3,3 has 6 vertices and 9 edges.

3 | V | 6  12  | E | .

That means K 3,3 satisfies the conclusion of Theorem 8.5. But this theorem gives a necessary condition
for a graph to be planar.

If the graph is planar then it satisfies the inequality | E |  3 | V | 6 .

If the graph does not satisfy the inequality then it is not planar.

If the graph satisfies the inequality then the graph could be planar or non‐planar. No conclusion.

Theorem 8.5 is a necessary condition for a graph to be planar. Is there a necessary and sufficient
condition? Yes, in fact, there are two related necessary and sufficient conditions, which we will discuss
now.

We have used the concept of isomorphism to compare graphs. We have to expand our methods of
comparison.

Definition: Given an edge e  uv in a graph G , an edge‐subdivision of e consists in adding a new

vertex w and replacing e by two edges e '  uw, e ''  wv .

Here is a picture illustrating the concept of edge‐subdivision.

Figure 1: Edge Subdivision

Definition: A graph G ' is said to be a subdivision of a graph G if G ' is formed by performing a

sequence of edge‐subdivisions on G .

Example 12: Here is an example of subdivision.

Start with a graph G . This graph has 5 vertices and 8 edges.

Here is a subdivision G1 of G . G1 has 10 vertices and 13 edges.
Here is a different subdivision G2 of G . G2 11 vertices and 14 edges.

Definition: We say that graphs G and H are homeomorphic if a subdivision of G is isomorphic to a

subdivision of H .

Being homeomorphic is a weaker condition than being isomorphic. That is if graphs G and H are
isomorphic then G and H are homeomorphic. But the converse is not true.

Example 13: The following graphs are homeomorphic but not isomorphic.

G and H are homeomorphic because they both have the following common subdivision G ' .
But G and H are not isomorphic. H has a 3‐cycle but G does not.

How does any of this relate to planar graphs? We have the following two theorems.

Theorem 8.6: Let the graph G ' be a subdivision of a graph G . Then G ' is planar if and only if G is
planar.

Proof: If G is planar and G ' is a subdivision of G then G is clearly planar (since subdividing an edge
cannot add an edge crossing).

If G is nonplanar, then any subdivision G ' of G is also nonplanar. For suppose that G ' were planar.
Then removing the subdivisions (and restoring the original edges) will not create any edge crossing.
Therefore G would be planar, contrary to hypothesis. Therefore any subdivision of a nonplanar graph is
nonplanar.

Theorem 8.7: Suppose that graphs G and H are homeomorphic. Then G is planar if and only if H is
planar.

Proof: Since G and H are homeomorphic then there is a subdivision G ' of G and a subdivision H ' of
H such that G ' and H ' are isomorphic. If G is planar then by Theorem 8.6, G ' is planar. H ' is
isomorphic to G ' , hence H ' is planar. Again using Theorem 8.6, H is planar.

Similarly, we can go in the other direction. QED

Here is one of the more famous theorems of graph theory. This theorem was proven in 1930 by the
Polish mathematician Kazimierz Kuratowski.

Theorem 8.8: Kuratowski’s Theorem. A graph is planar if and only if it contains no subgraph
homeomorphic to K5 or K 3,3 .

Another way of saying this, is that a planar graph has no subgraph which is a subdivision of K5 or K 3,3 .
Proof: () Assume that G is a planar graph. First of all, we noted earlier that K5 and K 3,3 are
nonplanar. Hence, by Theorem 8.7, any graph homeomorphic to K5 or K 3,3 is nonplanar.

Next , if a graph G is planar then any subgraph of G is also planar. In a drawing of G without edge
crossings, any subgraph is drawn without edge crossings.

We conclude that G must contain no subgraph homeomorphic to K5 or K 3,3 .

() This half of the proof is much more difficult and will not be done in class.

There is a generalization of Kuratowski’s Theorem. We have to define another type of operation on

graphs.

Definition: Let e  uv be an edge in graph G . We define a new graph G ' by the contraction of the
edge e as follows. The vertices u, v are identified. The edge e is removed. All other edges incident with
u or v remain.
In general, we say that G ' is obtained from G by edge contraction.

Let’s look at an example.

Example 14:

Let’s contract the edge BE. We will obtain the new graph,
Edge contraction can be combined with other operations.

Definition: Given a graph G , a graph G ' that is obtained from G by a sequence of edge contractions,
edge removals, and vertex removals is called a minor of G .

Here are some examples of graph minors.

Example 15: Start with the following graph.

Delete vertices A,B and contract edge FG. We get the following graph, which is a minor of the original
graph.
Example 16: This is the Petersen graph.

Let’s contract the following edges: AF, BG, CI, DJ, EH. We will obtain the following graph.
This graph is K5 . We just proved that K5 is a minor of the Petersen graph. We will see below why that
is significant.

Here is the main theorem relating planar graphs and graph minors. This was proven in 1937 by the
German mathematician Klaus Wagner.

Theorem 8.9: Wagner’s Theorem.

A graph G is planar if and only if neither K5 nor K 3,3 is a minor of G .

We will not prove this theorem. The proof is quite long, but somewhat similar to Kuratowski’s Theorem.

We have focused on planar graphs. We will now look at nonplanar graphs and a measure of how
nonplanar they are.

Definition: Given a graph G , the crossing number cr (G ) is the minimum number of edge crossings
that can occur in drawing the graph in the plane.

Obviously, a planar graph has the crossing number of 0. Here are some nonplanar examples.

Example 17: cr ( K5 )  1
Example 18: cr ( K 6 )  3

We will actually prove this is true. First of all, here is a drawing of K 6 with 3 edge crossings.

Is it possible to draw K 6 with fewer edge crossings?

Is it possible to draw the graph with one edge crossing? Take one of the edges that cross. Remove one
of the vertices on that edge. Then the graph that remains is K5 and we would have drawn it without
any edge crossings. Meaning that K5 is planar. Contradiction.
Is it possible to draw the graph with two edge crossings? Each edge in the crossing involves two vertices.
That means that 8 vertices are involved in at least one of the crossing edges. But there are only 6
vertices in the graph. Some vertex is part of two edge crossings. Remove that vertex and again we get a
drawing of K5 with no edge crossings. Contradiction.

This proves that cr ( K 6 )  3 .

Example 19: cr ( K 3,4 )  2

Below is a drawing of K3,4 with two edge crossings (CE and BD, CG and BH).

Is it possible to draw K3,4 with one edge crossing? If it were, then remove one of the vertices one of the
crossing edges. Then you have K 3,3 drawn with no edge crossings. Contradiction.

This proves that cr ( K 3,4 )  2 .

Let’s look at some of the theory involved with the crossing number. Since trees are planar and every
connected graph has a spanning tree (that uses all of the vertices of the graph), we know that every
graph has a subgraph that is spanning.

Theorem 8.10: Let G be a connected graph. Then

cr (G ) | EG | 3 | VG | 6 .
Proof:
Let H be a planar, spanning subgraph of G with a maximum number of edges. Applying Theorem 8.5 to
H , we have | EH | 3 | VH | 6 . Since H is spanning, we have | VH |  | VG | . Hence,

| EH |  3 | VG | 6 .

Consider the edges of G not in H . Each such edge, when it is added to a plane drawing of H , must
have at least one edge crossing. There are | EG |  | EH | edges in G but not in H . We have,

cr (G ) | EG |  | EH |  | EG | 3 | VG | 6 . QED

This theorem gives us a more direct way to prove that cr ( K 6 )  3 . We know that K 6 has 6 vertices and
15 edges. By Theorem 8.10, cr ( K 6 )  15  3  6  6  3 . Since we showed a drawing of K 6 with 3 edge
crossings, this immediately proves that cr ( K 6 )  3 .

Definition: The thickness of a graph G , denoted  (G ) , is the fewest number of planar, spanning
subgraphs of G whose edges partition the edge set of G .

Generally, we use the following thickness to help us compute the thickness of a graph.

Theorem 8.11: Let G be a connected graph. Then

 | EG | 
 (G )    (we round the right side up to the nearest integer).
 3 | VG | 6 

Proof: Let H1 , H 2 ,..., H (G ) be a collection of the fewest number of planar, spanning subgraphs whose
edges partition the edge set of G . For each subgraph H i , by Theorem 8.5,
| EHi |  3 | VHi | 6  3 | VG | 6 .
 (G )  (G )
| EG |
Hence, | EG |   | EH i | 
i 1
 (3 | V
i 1
G | 6)   (G )(3 | VG | 6)   (G ) 
3 | VG | 6
.

 | EG | 
Since  (G ) is an integer, we then have  (G )   .
 3 | VG | 6 
Let’s do an example.
Example 20: Let’s compute the thickness of K8 .

 28   28 
K8 has 8 vertices and 28 edges. By Theorem 8.11,  ( K8 )     1.55  2 .
 3  8  6   18 

Below we give a decomposition of K8 into two planar, spanning subgraphs. This proves that  ( K8 )  2 .

The planar, spanning subgraphs are H1 , H 2 .

If you look at the two subgraphs, you will see that for each vertex v in K8 , we have
degG (v)  deg H1 (v)  deg H 2 (v)  7 . Also, no edge appears in both subgraphs. You may notice that the
vertices in H 2 are labeled in a different way from H1 . That is just a matter of convenience. Labeling the
vertices in the same order makes it difficult to draw H 2 without using several edges that are not
straight lines.

That raises an interesting question. Is it always possible to draw a planar graph using straight lines for all
the edges?

This question is answered by the following theorem, proven by the Hungarian mathematician István
Fáry in 1948.

Theorem 8.12: Fáry’s Theorem. Every planar graph has a plane drawing in which all edges are straight
lines.

The proof of this theorem is fairly long, but not extremely difficult (it uses induction on the number of
vertices). We will skip it, in the interest of time.
Up to now, we’ve restricted our attention to planar graphs. Or as we said earlier, these are embeddings
of graphs in the plane. We will briefly discuss embeddings into other sets, where we don’t have edge
crossings.

First of all, embedding into three‐dimensional space 3 is not interesting. Every graph can be
embedded into 3 without any edge crossings.

A more interesting embedding is on a sphere. Using ideas of the German mathematician Bernhard
Riemann (1826 – 1866), we can easily relate embedding a graph in a plane and embedding in a sphere.
The process is called stereographic projection.

Points in the plane and on the sphere can be related as follows. Connect a point in the plane with the
North Pole of the sphere (called the Riemann sphere for this application). That line will intersect the
sphere at a point, which is called the point of projection. This process can be done in reverse (mapping
points on the sphere to points in the plane).

Figure 2: Riemann Sphere

Any geometric figure in the plan will be mapped to a geometric figure on the sphere (and vice versa).
This leads to a fairly obvious theorem.

Theorem 8.13: A graph is planar if and only if its projection on the sphere can be drawn without edge
crossings.

Proof: This follows immediately from the process of stereographic projection.

Next, we will look at embedding graphs on the following surface, a torus.

Figure 3: Torus

Note that the two circles shown in the figure. One circle goes around the hole of the torus. The other
circle goes around the tube of the torus.

We want to discuss embedding graphs on the surface of a torus. That is, drawing graphs on the surface,
without any edge crossings. That condition was sufficient to define an embedding on the plane or the
sphere.

For embedding a graph on a torus or other surfaces, we require an additional condition. This condition is
automatically true for graphs drawn on a plane or sphere.

Definition: A 2‐cell embedding of a graph on a surface S is an embedding in which every face has the
property that any closed curve can be contracted (or shrunk) to a point.

What does this mean?

Definition: A simply connected surface is one in which every can be contracted to a point without
leaving the surface.

Here is a simply connected region.

Example 21:
Figure 4: Not simply connected region

This figure is not simply connected because of the hole inside the outer circle. In the figure below, we
have drawn two closed curves inside the region.
We can shrink C1 to a single point without leaving the region. But we cannot shrink C2 to a point
without passing through the hole, which is not part of the region. That’s why this region is not simply
connected.

The torus itself is not simply connected. If you look back to Figure 3, neither circle shown in the drawing
can be contracted to a point. The circle that goes around the hole in the torus cannot be contracted to a
point without passing through the hole. The circle that goes around the tube cannot be contracted to a
point without passing inside the tube.

How does this relate to graphs drawn on a torus? We want to draw a graph on a torus so that each face
of the drawing is simply connected. Again, we call this a 2‐cell embedding. The real difficulty is, of
course, the fact that the torus itself is not simply connected, whereas the plane and the sphere are
simply connected.

Y1ou can create a torus by identifying sides of a square, as below.

Identifying the top and bottom sides of the square gives you a cylinder. Then identify the ends of the
cylinder. You obtain a torus, where the sides of the square have now become circles.

It is easier to draw a graph in the square and then see how it is transferred to the torus.

Let’s begin with drawing K 4 on a torus. There are different ways of doing this. Some are 2‐cell
embeddings and some are not, as we will see. But in all cases, K 4 has been drawn without an edge
crossings. On top is the drawing of K 4 on the torus and below is the corresponding drawing of K 4 in the
square.
Figure 5: Three drawings of K4 on the torus.

Note, this drawing comes from the textbook, which is why the faces are denoted as regions.

Version (a) is the standard plane drawing of K 4 . This is not a 2‐cell embedding on the torus. The
exterior face R4 is not simply connected. For example, R4 contains the circle that goes around the tube
of the torus, which cannot be contracted to a point.

Version (b) uses an arc going around the tube of the torus as one of the edges of the graph. This also not
a 2‐cell embedding on the torus. The exterior face R3 is not simply connected. If you take a circle
around the tube, but not intersecting the graph, then that circle cannot be contracted to a point.

Version (c) uses the circle going around the torus and going around the tube as edges of the graph. This
version is a 2‐cell embedding. Why is that? Take a circle that goes around the torus. It won’t actually get
all the way around, because it will run into an edge of the graph. Also, a circle going around the tube,
won’t actually go all the way around, because it will run into an edge of the graph.

This proves that there is a 2‐cell embedding of K 4 on the torus.

The following picture shows that there is a 2‐cell embedding of K5 on the torus.
Figure 6: Embedding K5 on the torus

This is basically the 2‐cell embedding of K 4 with an extra vertex in the middle.

There are some theoretical results on embedding that we will now discuss.

We have to introduce some terminology.

Here is a two hole torus.

Figure 7: Two hole torus

Here is a three hole torus.

Figure 8: Three hole torus

One can keep on going and define a 4‐hole torus, 5‐hole torus, etc. up to a general n‐hole torus. This
leads to the following definition.

Definition: A connected surface has genus g if the topologically equivalent to a torus with g holes.

A sphere has genus 0. A torus has genus 1.

Definition: A connected graph G has genus  (G ) if the smallest genus of a surface that G can be
embedded in is  (G ) .

A planar graph has genus  (G )  0 . We saw that K5 cannot be embedded in the plane, but can be
embedded on the torus. Therefore,  ( K5 )  1 .

The following theorem generalizes the Euler Polyhedral Formula.

Theorem 8.14: Let G be a connected graph that has a 2‐cell embedding on a surface of genus g . Then

| VG |  | EG |  | FG |  2  2 g .

Proof: Omitted, but can be found in the textbook.

We saw that K 4 is planar and can be embedded on the torus. We also know that K3 is planar.
Example 22: Can K3 be embedded on the torus?

Assume that K3 can be embedded on the torus, which has genus g  1 . Also, K3 has three vertices and
three edges. When K3 is embedded on the torus, it will have two faces. Applying this information to
Theorem 8.14 we get,

3  3  2  2  2 1  0  2  0 . This is a contradiction, which means that K3 cannot be embedded on

the torus.

Corollary 8.15: Let G be a connected graph that has genus  (G ) . Then, if G is embedded in a surface
of genus  (G ) ,

| VG |  | EG |  | FG |  2  2 (G ) .

Proof: Follows immediately from Theorem 8.14.

We can use this Corollary to derive a formula for the genus of a graph.

Corollary 8.16: If G is a connected graph with at least 3 vertices then,

| EG | | VG |
 (G )   1.
6 2
Proof: Let G be embedded in a surface of genus  (G ) . By Corollary 8.15,
| VG |  | EG |  | FG |  2  2 (G ) .

Let | FG | F and f1 , f 2 ,..., f F be the faces of G . Let mi  # edges on the boundary of face Fi (where
1  i  F ). That means mi  3 . Every edge is on the boundary of one or two faces. We get,
F
3F   mi  2 | EG |  3F  2 | EG | .
i 1

Therefore,

6  6 (G )  3(2  2 (G ))  3(| VG |  | EG |  | FG |)  3 | VG | 3 | EG | 3 | FG |
 3 | VG | 3 | EG | 2 | EG |  3 | VG |  | EG |

Solving this for  (G ) ,

| EG | | VG |
 (G )   1.
6 2
Let’s look at the above formula for some specific graphs. Note, the formula is an inequality. It doesn’t
given you the exact value of the genus.

Graph Formula Actual Genus

K4  ( K4 )  0  (G )  0
K5 1  (G )  1
 ( K5 ) 
6
K6 1  (G )  1
 ( K6 ) 
2
K7  ( K7 )  1  (G )  1

The formula values tells us that the graphs K5 , K 6 , K 7 are all nonplanar. We showed that K5 can be
embedded on the torus. One can also show, not as easily, that K 6 , K 7 can be embedded on the torus.

Let’s look at K8 in more detail. K8 has 8 vertices and 28 edges. From the formula in Corollary 8.16, we
28 8 5
get  ( K8 )    1  . This means that K8 cannot be embedded on the torus. In fact,  ( K8 )  2 .
6 2 3
We actually have formulas for the genus of complete graphs and complete bipartite graphs.

The genus of K n , proven by Gerhard Ringel and J.W.T. Youngs in 1968.

 (n  3)(n  4) 
 (Kn )    .
 12

The genus of K m , n , proven by Gerhard Ringel in 1965.

 (m  2)(n  2) 
 ( K m,n )    .
 4

For example,

 (5  3)(5  4)   1 
 ( K5 )      6   1
 12
 (6  3)(6  4)   1 
 ( K6 )      2   1
 12
 (7  3)(7  4) 
 ( K7 )     1  1
 12
 (8  3)(8  4)  10 
 ( K8 )      6   2
 12
And for complete bipartite graphs.

 (3  2)(3  2)   1 
 ( K 3,3 )      4   1
 4
 (3  2)(3  2)  1 
 ( K 4,4 )     1   1
 4
 (4  2)(5  2)   6 
 ( K 4,5 )      4   2
 4

This shows that K 4,5 cannot be embedded on the torus, but can be embedded on the two hole torus.

This brings us to the end of this lecture. In the next lecture, we will discuss graph coloring.