Foreword

Graphs are an important and burgeoning topic in Combinatorics, not only

in research, but also in mathematical olympiads. They have a strong presence
in the International Mathematical Olympiad, National Team Selection Tests
and sometimes National Olympiads as well.
Yet, when asked by an aspiring problem solver where to learn graphs from,
we inevitably hesitate to respond. There are very good books on graphs, such
as Diestel’s excellent Graph Theory, but, being written as advanced textbooks,
they are not really suitable for our problem solver. They quickly dive into
complex topics, assume a fair amount of pre-requisite knowledge, and do not
contain many problems. There are also some good combinatorics books for
olympiads, but they do not to present the basics of graphs in the structured
way in which they should be presented.
So we thought of writing a book to bridge this gap between the enthusi-
astic problem solver and the beautiful field of graphs. We are going carefully
through the basics, but without throwing too much theory at the reader. The
guiding principle for writing this book was not to show more than needed: for
a method or idea, we tried to present a reasonably easy example, leaving the
harder ones for the reader to solve.
We also aimed at a very friendly and informal book, with a lot of comments,
drawings, and most of all, intuitions from someone who has been, not too long
ago, in the position of the reader. We often gave more than one solution to
a problem, so it is worth checking after having solved a problem. The source
and author of a problem, if known, are also mentioned at the end.
The book has an organic structure, with chapters in their natural order.
We kept the theory part of the chapters within reasonable limits in order
to encourage the reader to read through carefully before proceeding to the
problems. Our recommendation would then be to try to go through the book
in order, reading the theory with patience. It is also worth trying to prove
vi Foreword

a theorem by oneself before reading the proof. Alternatively, if the reader

wants, she could also read each chapter, work on most of the problems, and
then go to the next chapter, leaving the hardest problems for after she finished
all chapters.
In graph theory, unlike, say, in geometry, there isn’t a clear distinction
between theorems and problems. There is hardly a theorem that can be ap-
plied over and over again, while many of the problems could be regarded as
mini-theorems. For this reason, many of the results proven in this book are
noncommittally labelled ‘Proposition’, a term which captures precisely this
ambiguity. Anyway, we would like to encourage the reader to join into this
way of thinking, and not to look for theorems and applications of theorems.
We tried to assume as little as possible, the only things being made use of
in the ten chapters being some basic principles in combinatorics, such as the
pigeonhole principle, and some elementary inequalities, such as AM-GM. We
did add two appendices though, one on the probabilistic method in graphs, and
the other on linear algebra in graph theory. Both assume a good knowledge of
probabilities and linear algebra, so they should not be used to actually learn
about these topics. Also, we encourage the reader to resist the temptation,
common nowadays, to try to learn more without having mastered the basics.
The appendices are for those who have already mastered the ten chapters.
There is a common practice in mathematical competitions to avoid graphs
terminology, in particular the very word ‘graph’. To achieve this, graphs
problems, that presumably came to their authors as graphs problems, are
often painted over with a story involving airlines, roads, and what not. A
reason for this is that it is not a pre-requisite for olympiads to know graphs.
To us, this sounds a bit flimsy: it is unlikely that someone who has never
heard of graphs would solve a hard graphs problem, even if it is formulated
in terms of airlines. So, in this book, we have reformulated the problems in
graph theoretical terms, to make everything neater and easier to read. The
only exceptions are those in which the interpretations actually motivate the
problem and make it more intelligible.
We wrote this book, our first one, as we would have liked to read when in
high school. Hopefully, things haven’t changed too much.
Contents

Foreword v

1 Introduction 1

2 Eulerian Circuits and Hamiltonian Cycles 29

3 Trees 37

4 Chromatic Numbers 49

5 Planar Graphs 55

6 Matchings in Bipartite Graphs 65

7 Extremal Graph Theory 77

8 Ramsey Theory 89

9 Directed Graphs 99

10 Infinite Graphs 117

Appendix A: Probabilities in Graph Theory 129

Appendix B: Linear Algebra in Graph Theory 143

viii Contents

Solutions 153
1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 153
2 Eulerian Circuits and Hamiltonian Cycles . . . . . . . . . . . . 191
3 Trees . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 203
4 Chromatic Numbers . . . . . . . . . . . . . . . . . . . . . . . . 227
5 Planar Graphs . . . . . . . . . . . . . . . . . . . . . . . . . . . 241
6 Matching in Bipartite Graphs . . . . . . . . . . . . . . . . . . . 252
7 Extremal Graph Theory . . . . . . . . . . . . . . . . . . . . . . 262
8 Ramsey Theory . . . . . . . . . . . . . . . . . . . . . . . . . . . 290
9 Directed Graphs . . . . . . . . . . . . . . . . . . . . . . . . . . 307
10 Infinite Graphs . . . . . . . . . . . . . . . . . . . . . . . . . . . 333
Appendix A: Probabilities in Graph Theory . . . . . . . . . . . . . . 354
Appendix B: Linear Algebra in Graph Theory . . . . . . . . . . . . . 373

Bibliography and Further Reading 381

Glossary 383

Other Books from XYZ Press 391

Chapter 1

Introduction

Graphs can be seen as representing relations in the world. Suppose we

have a group of people, some pairs of them knowing each other, some not. If
‘knowing each other’ is, as the English language seems to have it, a mutual
relation, we can try to represent these relations by some drawing points on
the paper to represent the people, and connecting pairs that know each other.
It is irrelevant whether we connect two points with a straight line or not. In
this way, we might be able to visualise better how many friends each person
has, or perhaps how many common friends two certain people have.

Or suppose that we have a number of cities, and two-way airline services

between them. Again, we might try to represent the pairs that have a service
between them, perhaps to see how to get from one city to another with as few
changes as possible. We get the same kind of drawing.
2 Chapter 1

Figure 1.1: The airline services between New York, Washington,

Paris, Warsaw and Bucharest

These, and many other, types of relations can be represented by a graph.

Yet once we define what a graph is, and start to ask questions about graphs,
they will hopefully prove so fascinating in their own right as to make the reader
forget all the real-world interpretations from which we started.

Definition 1.1. An (undirected) graph G consists of a set V , called the vertex

set, and a set E of edges between the vertices (formally, pairs {u, v}, where u
and v are elements of V ). We write G(V, E).
Graphs can be finite or infinite, depending on whether V is finite or infinite.
Unless otherwise stated, we assume graphs to be finite.
A graph is simple if between every two vertices there is at most one edge
and there are no loops (edges between v and itself). The formal definition
we gave implies that graphs are simple, but more complex definitions could
render them otherwise.

Note that, even if we often represent the graph on paper, as described

before, the graph is not a geometrical object. There are many ways in which
we can ‘draw’ a graph.
Introduction 3

Figure 1.2: Two drawings of the same graph

There are many definitions in graph theory, many of them natural. We

will give most of them now, to be then able to speak graphese at will.
Definition 1.2. In a graph G(V, E):
• For an edge uv, u and v are called its endpoints
Two vertices are called adjacent if there is an edge between them. We
also say that they are connected or that they are neighbours.
Two edges are called adjacent if they share a common endpoint.
An edge is called incident on a vertex if it has that vertex as an endpoint.

• The degree of a vertex v ∈ V is the number of edges incident on v (i.e.

having v as an endpoint). It is typically denoted by d(v).

• The maximal degree of a graph G is typically denoted by ∆(G).

• The minimal degree of a graph G is typically denoted by δ(G).

We will now define some important objects that have a more global im-
portance in the graph. Suppose one wants to plan a trip, going from city to
city by airplane. This is what we will call a trail. Additional conditions, such
as not going through the same city twice, give rise to new notions.
Definition 1.3. In a graph G:
4 Chapter 1

• A trail from v to w is a sequence of vertices v = v1 , v2 , . . . , vn = w such

that vi vi+1 ∈ E for i = 1, 2, . . . , n − 1.
A walk from v to w is a trail in which all edges are distinct.
A path from v to w is a trail in which all vertices are distinct.

• The length of a trail/walk/path is the number of (not necessarily dis-

tinct) edges in it.

• The distance between two vertices u and v, usually denoted by d(u, v),
is the length of the shortest path between u and v.

• A closed trail is a sequence of vertices v1 , v2 , . . . , vn such that vi vi+1 ∈ E

for i = 1, 2, . . . , n, with indices taken cyclically. To emphasise that
a closed trail returns to v1 we will often denote a closed trail by
v1 , v2 , . . . , vn , v1 .
A circuit is a closed trail in which the edges vi vi+1 are pairwise distinct.
A cycle is a circuit in which the vertices are pairwise distinct.

Figure 1.3: Walk, paths, cycles, circuits

• The length of a circuit/cycle is the number of edges in it.

A cycle of length 3 is often called a triangle.

• The girth of a graph is the length of the minimal cycle (if one exists,
otherwise it is ∞).
Introduction 5

Definition 1.4. A graph is connected if for any two vertices v and w, there
is a path from v to w.
Remark. (Connected components) We observe that graphs that are not
connected are essentially ‘a collection of connected graphs’.
We will call the maximal connected subgraphs of a graph G the connected
components of G.

Figure 1.4: A graph with 3 connected components

There are some special kinds of graphs that we will use:

Definition 1.5. We have the following types of graphs:
• A tree is a connected graph with no circuits.

Figure 1.5: A tree

6 Chapter 1

• Given a graph G, its complement G consists of the same vertices and

exactly the edges that are not edges in G.

Figure 1.6: A graph G and its complement G

• A graph is called regular if all vertices have the same degree (k-regular
if all degrees are k).

• Kn is the complete graph on n vertices, i.e. the graph with all possible
edges between n vertices.

• A graph is called bipartite if its vertices can be partitioned into sets A

and B such that all edges are between a vertex in A and one in B.

• Km,n is the complete bipartite graph, i.e. the bipartite graph with sets
A and B such that |A| = m, |B| = n and all the possible edges between
A and B are drawn.
Introduction 7

Figure 1.7: K5 and K3,2

And now some definitions about the relations between graphs:

Definition 1.6. We have the following definitions:

• Two graphs are called isomorphic if ‘they are the same graph’.

(Formally, they are isomorphic if there is a bijection f between the ver-

tices of the two graphs such that uv is an edge if and only if f (u)f (v) is
an edge.)

• A subgraph of a graph G is a subset of the vertices of G together with

some the edges between them that were in G.

A subgraph is called induced if it contains all the edges between its

vertices that were edges in G.
8 Chapter 1

Figure 1.8: An induced and a non-induced subgraph on 4 vertices

A subgraph is called spanning if its set of vertices is the set of vertices

of G.

As an introduction to graphs, we shall ask and answer some basic questions

about graphs, yet chosen in such a way as to illustrate the basic methods of
graph theory. These are the building blocks that we will use in later chapters.

The problems in this introduction are meant not so much to develop theory
(this we will do later), as to get us used to graphs and to train our intuitions.

Pigeonhole Principle
The pigeonhole principle is the basic observation that if we have nk + 1
objects in n sets, then there is a set with at least k + 1 objects.

A classical problem in graphs is that if we have six people, either three of

them are pairwise friends, or three of them are pairwise strangers. Rephrased
in graph theoretical terms, we have the following:

Proposition 1.7. In any graph on 6 vertices, there exist three vertices that
are pairwise connected, or three vertices that are pairwise not connected.
Introduction 9

Proof. Pick a vertex, say v. By the pigeonhole principle, there are three other
vertices that are either all connected or all not connected to v. Assume the
former.
If two of these are connected between themselves, they form a triangle
with v. Otherwise, the three of them are pairwise not connected.
Similarly for the case in which the three vertices are not connected to v.

In general, the pigeonhole principle can come in handy when we have a lot
of edges and wanting to find something like a complete subgraph.

A more general version of this problem, and other similar problems, are to
be found in the Ramsey theory chapter.

Double Counting
The technique of double counting consists of expressing something in two
different ways and then deducing that the results are equal.
Simple as it might sound, the technique of double counting comes in very
handy in many areas of combinatorics, including graph theory. This is perhaps
to be expected, given the fact that we are dealing with two objects, vertices
and edges. The most basic observation is:

Theorem 1.8. (Handshaking lemma) In every graph G = (V, E),

X
d(v) = 2|E|.
v∈V

Proof. The left-hand side counts edges: d(v) counts those edges that are inci-
dent on v. But we can easily observe that each edge uv is counted twice, once
for d(u), and once for d(v). The conclusion follows.
(Formally, both sides count the number of pairs (v, e), where v is a vertex,
and e an edge incident on v. For each v, there are d(v) pairs, while for each
e, there are exactly two pairs).