A Walk Through Combinatorics

An Introduction to Enumeration and Graph Theory

Miklós Bóna

• Publisher: WSPC
• Year: 2006
• Print ISBN: 9789812568861
• eBook ISBN: 9789814365475
• Edition: 2
• Topic: Mathematics
• Subtopic: Counting & Numeration
• Index: Mathematics

This is a textbook for an introductory combinatorics course that can take up one or two semesters. An extensive list of problems, ranging from routine exercises to research questions, is included. In each section, there are also exercises that contain material not explicitly discussed in the preceding text, so as to provide instructors with extra choices if they want to shift the emphasis of their course.

Just as with the first edition, the new edition walks the reader through the classic parts of combinatorial enumeration and graph theory, while also discussing some recent progress in the area: on the one hand, providing material that will help students learn the basic techniques, and on the other hand, showing that some questions at the forefront of research are comprehensible and accessible for the talented and hard-working undergraduate. The basic topics discussed are: the twelvefold way, cycles in permutations, the formula of inclusion and exclusion, the notion of graphs and trees, matchings and Eulerian and Hamiltonian cycles. The selected advanced topics are: Ramsey theory, pattern avoidance, the probabilistic method, partially ordered sets, and algorithms and complexity.

As the goal of the book is to encourage students to learn more combinatorics, every effort has been made to provide them with a not only useful, but also enjoyable and engaging reading.

Contents:

• Basic Methods:
  • Seven Is More Than Six. The Pigeon-Hole Principle
  • One Step at a Time. The Method of Mathematical Induction
• Enumerative Combinatorics:
  • There Are a Lot of Them. Elementary Counting Problems
  • No Matter How You Slice It. The Binomial Theorem and Related Identities
  • Divide and Conquer. Partitions
  • Not So Vicious Cycles. Cycles in Permutations
  • You Shall Not Overcount. The Sieve
  • A Function is Worth Many Numbers. Generating Functions
• Graph Theory:
  • Dots and Lines. The Origins of Graph Theory
  • Staying Connected. Trees
  • Finding a Good Match. Coloring and Matching
  • Do Not Cross. Planar Graphs
• Horizons:
  • Does It Clique? Ramsey Theory
  • So Hard to Avoid. Subsequence Conditions on Permutations
  • Who Knows What It Looks Like, but It Exists. The Probabilistic Method
  • At Least Some Order. Partial Orders and Lattices
  • The Sooner The Better. Combinatorial Algorithms
  • Does Many Mean More Than One? Computational Complexity

Readership: Upper level undergraduates and graduate students in the field of combinatorics and graph theory.

Chapter 13

Does It Clique? Ramsey Theory

Instead of coloring the vertices of our graphs, in this chapter we will color their edges. We will see that this leads to a completely different set of problems. Our first excursion into the land of infinite graphs is also part of this chapter.

13.1 Ramsey Theory for Finite Graphs

Example 13.1. Six people are waiting in the lobby of a hotel. Prove that there are either three of them who know each other, or three of them who do not know each other.

This statement is far from being obvious. We could think that maybe there is some case in which everyone knows roughly half of the other people, and in the company of any three people there will be two people who know each other, and two people who do not. We will prove, however, that this can never happen.

Solution. (of Example 13.1) Take a K₆ so that each person corresponds to a vertex. Color the edge joining A and B red if A and B know each other, and blue if they do not. Do this for all 15 edges of the graph. The claim of the example will be proved if we can show that there will always be a triangle with monochromatic edges in our graph.

Take any vertex V of our bicolored graph. As V is of degree five, it must have at least three edges adjacent to it that have the same color. Assume without loss of generality that this color is red. Let X, Y and Z be the endpoints of three red edges adjacent to V. (The reader can follow our argument in Figure 13.1, where we denoted red edges by solid lines.)

Now if any edge of the triangle XYZ is red, then that edge, and the two edges joining (the endpoints of) that edge to V are red, so we have a triangle with three red edges. If the triangle XYZ does not have a red edge, then it has three blue edges.

Fig. 13.1 The colors of the edges of the triangle XYZ are crucial.

This beautiful proof is our first example in Ramsey theory. This field is named after Frank Plumpton Ramsey, who was the first one to study this area at the beginning of the twentieth century.

We point out that the result is tight, that is, if there were only five people in the lobby of the hotel, then the same statement would be false. Indeed, take a K₅, and draw it as a regular pentagon and its diagonals. Color all five sides red, and all five diagonals blue. As any triangle in this graph contains at least one side and at least one diagonal, there can be no triangles with monochromatic edges.

Instead of taking a K₆, and coloring its edges red and blue, we could have just taken a graph H on six vertices in which the edges correspond to people who know each other. In this setup, the edges of H correspond to the former red edges, and the edges of the complement of H correspond to the former blue edges. As a complete subgraph is often called a clique, the statement of Example 13.1 can be reformulated as follows. If H is a simple graph on six vertices, then at least one of H and the complement of H contains a clique of size three.

The arguments used in the proof of Example 13.1 strongly depended on the parameter three, the number of people we wanted to know or not to know each other. What happens if we replace this number three by a larger number? Is it true that if there are sufficiently many people in the lobby, there will always be at least k of them who know each other, or k of them who do not know each other? The following theorem answers this question (in fact, a more general one), in the affirmative.

Theorem 13.2. [Ramsey theorem for graphs] Let k and I be two positive integers, both of which is at least two. Then there exists a (minimal) positive integer R(k, l) so that if we color the edges of a complete graph with R(k, l) vertices red and blue, then this graph will either have a K_k subgraph with only red edges, or a K_l subgraph with only blue edges.

Note that any nonempty set of positive integers has a minimal element. Therefore, if we can show that there exists at least one positive integer with the desired property, then we will have shown that a minimal such integer exists.

Example 13.3. Example 13.1, and the discussion after it shows that R(3,3) = 6. We also have trivial fact R(2,2) = 2 relating to the graph with one edge.

Proof. (Of Theorem 13.2) We prove the statement by a new version of mathematical induction on k and l. This induction will run as follows. First we prove the initial conditions that R(k, 2) and R(2, l) exist ...

Table of contents

• Cover page
• Title page
• Copyright page
• Dedication
• Foreword
• Preface
• Acknowledgments
• I. basic methods
• II. enumerative combinatorics
• III. graph theory
• IV. horizons
• Bibliography
• Index
• Back Cover