INTRODUCTION TO 

COMBINATORICS

K10310_FM.indd 1 7/22/10 12:45:31 PM

 DISCRETE
MATHEMATICS
ITS APPLICATIONS
Series Editor
Kenneth H. Rosen, Ph.D.
R. B. J. T. Allenby and Alan Slomson, How to Count: An Introduction to Combinatorics,
Third Edition
Juergen Bierbrauer, Introduction to Coding Theory
Francine Blanchet-Sadri, Algorithmic Combinatorics on Partial Words
Richard A. Brualdi and Dragos̆ Cvetković, A Combinatorial Approach to Matrix Theory and Its Applications
Kun-Mao Chao and Bang Ye Wu, Spanning Trees and Optimization Problems
Charalambos A. Charalambides, Enumerative Combinatorics
Gary Chartrand and Ping Zhang, Chromatic Graph Theory
Henri Cohen, Gerhard Frey, et al., Handbook of Elliptic and Hyperelliptic Curve Cryptography
Charles J. Colbourn and Jeffrey H. Dinitz, Handbook of Combinatorial Designs, Second Edition
Martin Erickson, Pearls of Discrete Mathematics
Martin Erickson and Anthony Vazzana, Introduction to Number Theory
Steven Furino, Ying Miao, and Jianxing Yin, Frames and Resolvable Designs: Uses, Constructions,
and Existence
Mark S. Gockenbach, Finite-Dimensional Linear Algebra
Randy Goldberg and Lance Riek, A Practical Handbook of Speech Coders
Jacob E. Goodman and Joseph O’Rourke, Handbook of Discrete and Computational Geometry,
Second Edition
Jonathan L. Gross, Combinatorial Methods with Computer Applications
Jonathan L. Gross and Jay Yellen, Graph Theory and Its Applications, Second Edition
Jonathan L. Gross and Jay Yellen, Handbook of Graph Theory
David S. Gunderson, Handbook of Mathematical Induction: Theory and Applications
Darrel R. Hankerson, Greg A. Harris, and Peter D. Johnson, Introduction to Information Theory and
Data Compression, Second Edition
Darel W. Hardy, Fred Richman, and Carol L. Walker, Applied Algebra: Codes, Ciphers, and
Discrete Algorithms, Second Edition
Daryl D. Harms, Miroslav Kraetzl, Charles J. Colbourn, and John S. Devitt, Network Reliability:
Experiments with a Symbolic Algebra Environment

K10310_FM.indd 2 7/22/10 12:45:31 PM

 Titles (continued)
Silvia Heubach and Toufik Mansour, Combinatorics of Compositions and Words
Leslie Hogben, Handbook of Linear Algebra
Derek F. Holt with Bettina Eick and Eamonn A. O’Brien, Handbook of Computational Group Theory
David M. Jackson and Terry I. Visentin, An Atlas of Smaller Maps in Orientable and Nonorientable
Surfaces
Richard E. Klima, Neil P. Sigmon, and Ernest L. Stitzinger, Applications of Abstract Algebra
with Maple™ and MATLAB®, Second Edition
Patrick Knupp and Kambiz Salari, Verification of Computer Codes in Computational Science
and Engineering
William Kocay and Donald L. Kreher, Graphs, Algorithms, and Optimization
Donald L. Kreher and Douglas R. Stinson, Combinatorial Algorithms: Generation Enumeration and Search
C. C. Lindner and C. A. Rodger, Design Theory, Second Edition
Hang T. Lau, A Java Library of Graph Algorithms and Optimization
Elliott Mendelson, Introduction to Mathematical Logic, Fifth Edition
Alfred J. Menezes, Paul C. van Oorschot, and Scott A. Vanstone, Handbook of Applied Cryptography
Richard A. Mollin, Advanced Number Theory with Applications
Richard A. Mollin, Algebraic Number Theory
Richard A. Mollin, Codes: The Guide to Secrecy from Ancient to Modern Times
Richard A. Mollin, Fundamental Number Theory with Applications, Second Edition
Richard A. Mollin, An Introduction to Cryptography, Second Edition
Richard A. Mollin, Quadratics
Richard A. Mollin, RSA and Public-Key Cryptography
Carlos J. Moreno and Samuel S. Wagstaff, Jr., Sums of Squares of Integers
Dingyi Pei, Authentication Codes and Combinatorial Designs
Kenneth H. Rosen, Handbook of Discrete and Combinatorial Mathematics
Douglas R. Shier and K.T. Wallenius, Applied Mathematical Modeling: A Multidisciplinary Approach
Alexander Stanoyevitch, Introduction to Cryptography with Mathematical Foundations and
Computer Implementations
Jörn Steuding, Diophantine Analysis
Douglas R. Stinson, Cryptography: Theory and Practice, Third Edition
Roberto Togneri and Christopher J. deSilva, Fundamentals of Information Theory and Coding Design
W. D. Wallis, Introduction to Combinatorial Designs, Second Edition
W. D. Wallis and J. C. George, Introduction to Combinatorics
Lawrence C. Washington, Elliptic Curves: Number Theory and Cryptography, Second Edition

K10310_FM.indd 3 7/22/10 12:45:31 PM

K10310_FM.indd 4 7/22/10 12:45:31 PM
 DISCRETE MATHEMATICS AND ITS APPLICATIONS
Series Editor KENNETH H. ROSEN

INTRODUCTION TO 
COMBINATORICS

W. D. Wallis
Southern Illinois University
Carbondale, Illinois, USA

J. C. George
Gordon College
Barnesville, Georgia, USA 

K10310_FM.indd 5 7/22/10 12:45:31 PM

Cover images portray permutations, random experiments, sampling, Latin squares, and Kirkman’s Schoolgirl
Problem.

Chapman & Hall/CRC

Taylor & Francis Group
6000 Broken Sound Parkway NW, Suite 300
Boca Raton, FL 33487-2742

© 2011 by Taylor and Francis Group, LLC

Chapman & Hall/CRC is an imprint of Taylor & Francis Group, an Informa business

No claim to original U.S. Government works

Printed in the United States of America on acid-free paper

10 9 8 7 6 5 4 3 2 1

International Standard Book Number-13: 978-1-4398-0623-4 (Ebook-PDF)

This book contains information obtained from authentic and highly regarded sources. Reasonable efforts
have been made to publish reliable data and information, but the author and publisher cannot assume
responsibility for the validity of all materials or the consequences of their use. The authors and publishers
have attempted to trace the copyright holders of all material reproduced in this publication and apologize to
copyright holders if permission to publish in this form has not been obtained. If any copyright material has
not been acknowledged please write and let us know so we may rectify in any future reprint.

Except as permitted under U.S. Copyright Law, no part of this book may be reprinted, reproduced, transmit-
ted, or utilized in any form by any electronic, mechanical, or other means, now known or hereafter invented,
including photocopying, microfilming, and recording, or in any information storage or retrieval system,
without written permission from the publishers.

For permission to photocopy or use material electronically from this work, please access www.copyright.
com (http://www.copyright.com/) or contact the Copyright Clearance Center, Inc. (CCC), 222 Rosewood
Drive, Danvers, MA 01923, 978-750-8400. CCC is a not-for-profit organization that provides licenses and
registration for a variety of users. For organizations that have been granted a photocopy license by the CCC,
a separate system of payment has been arranged.

Trademark Notice: Product or corporate names may be trademarks or registered trademarks, and are used
only for identification and explanation without intent to infringe.
Visit the Taylor & Francis Web site at
http://www.taylorandfrancis.com

and the CRC Press Web site at

http://www.crcpress.com
For Our Families
Contents

List of Figures xiii

Preface xv

1 Introduction 1
1.1 Some Combinatorial Examples . . . . . . . . . . . . . . . . . 1
1.2 Sets, Relations and Proof Techniques . . . . . . . . . . . . . 13
1.3 Two Principles of Enumeration . . . . . . . . . . . . . . . . . 15
1.4 Graphs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
1.5 Systems of Distinct Representatives . . . . . . . . . . . . . . 20
Exercises 1A . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
Exercises 1B . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25

2 Fundamentals of Enumeration 27
2.1 Permutations and Combinations! " . . . . . . . . . . . . . . . . 27
n
2.2 Applications of P (n, k) and . . . . . . . . . . . . . . . . 29
k
2.3 Permutations and Combinations of Multisets . . . . . . . . . 32
2.4 Applications and Subtle Errors . . . . . . . . . . . . . . . . . 36
2.5 Algorithms . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39
Exercises 2A . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
Exercises 2B . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44

3 The Pigeonhole Principle and Ramsey’s Theorem 47

3.1 The Pigeonhole Principle . . . . . . . . . . . . . . . . . . . . 47
3.2 Applications of the Pigeonhole Principle . . . . . . . . . . . 48
3.3 Ramsey’s Theorem — the Graphical Case . . . . . . . . . . . 51
3.4 Ramsey Multiplicity . . . . . . . . . . . . . . . . . . . . . . . 54
3.5 Sum-Free Sets . . . . . . . . . . . . . . . . . . . . . . . . . . 56
3.6 Bounds on Ramsey Numbers . . . . . . . . . . . . . . . . . . 59
3.7 The General Form of Ramsey’s Theorem . . . . . . . . . . . 63
Exercises 3A . . . . . . . . . . . . . . . . . . . . . . . . . . . 63
Exercises 3B . . . . . . . . . . . . . . . . . . . . . . . . . . . 64
Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66

ix
x

4 The Principle of Inclusion and Exclusion 69

4.1 Unions of Events . . . . . . . . . . . . . . . . . . . . . . . . . 69
4.2 The Principle . . . . . . . . . . . . . . . . . . . . . . . . . . . 71
4.3 Combinations with Limited Repetitions . . . . . . . . . . . . 75
4.4 Derangements . . . . . . . . . . . . . . . . . . . . . . . . . . 77
Exercises 4A . . . . . . . . . . . . . . . . . . . . . . . . . . . 80
Exercises 4B . . . . . . . . . . . . . . . . . . . . . . . . . . . 81
Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 83

5 Generating Functions and Recurrence Relations 85

5.1 Generating Functions . . . . . . . . . . . . . . . . . . . . . . 85
5.2 Recurrence Relations . . . . . . . . . . . . . . . . . . . . . . 89
5.3 From Generating Function to Recurrence . . . . . . . . . . . 94
5.4 Exponential Generating Functions . . . . . . . . . . . . . . . 95
Exercises 5A . . . . . . . . . . . . . . . . . . . . . . . . . . . 97
Exercises 5B . . . . . . . . . . . . . . . . . . . . . . . . . . . 99
Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 100

6 Catalan, Bell and Stirling Numbers 103

6.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . 103
6.2 Catalan Numbers . . . . . . . . . . . . . . . . . . . . . . . . 104
6.3 Stirling Numbers of the Second Kind . . . . . . . . . . . . . 108
6.4 Bell Numbers . . . . . . . . . . . . . . . . . . . . . . . . . . . 113
6.5 Stirling Numbers of the First Kind . . . . . . . . . . . . . . . 114
6.6 Computer Algebra and Other Electronic Systems . . . . . . 117
Exercises 6A . . . . . . . . . . . . . . . . . . . . . . . . . . . 119
Exercises 6B . . . . . . . . . . . . . . . . . . . . . . . . . . . 121
Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 122

7 Symmetries and the Pólya-Redfield Method 123

7.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . 123
7.2 Basics of Groups . . . . . . . . . . . . . . . . . . . . . . . . . 124
7.3 Permutations and Colorings . . . . . . . . . . . . . . . . . . 130
7.4 An Important Counting Theorem . . . . . . . . . . . . . . . 131
7.5 Pólya and Redfield’s Theorem . . . . . . . . . . . . . . . . . 134
Exercises 7A . . . . . . . . . . . . . . . . . . . . . . . . . . . 138
Exercises 7B . . . . . . . . . . . . . . . . . . . . . . . . . . . 139
Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 140

8 Introduction to Graph Theory 143

8.1 Degrees . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 143
8.2 Paths and Cycles in Graphs . . . . . . . . . . . . . . . . . . 146
8.3 Maps and Graph Coloring . . . . . . . . . . . . . . . . . . . 149
Exercises 8A . . . . . . . . . . . . . . . . . . . . . . . . . . . 153
Exercises 8B . . . . . . . . . . . . . . . . . . . . . . . . . . . 155
Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 156
 xi

9 Further Graph Theory 159

9.1 Euler Walks and Circuits . . . . . . . . . . . . . . . . . . . . 159
9.2 Application of Euler Circuits to Mazes . . . . . . . . . . . . 164
9.3 Hamilton Cycles . . . . . . . . . . . . . . . . . . . . . . . . . 166
9.4 Trees . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 170
9.5 Spanning Trees . . . . . . . . . . . . . . . . . . . . . . . . . . 174
Exercises 9A . . . . . . . . . . . . . . . . . . . . . . . . . . . 180
Exercises 9B . . . . . . . . . . . . . . . . . . . . . . . . . . . 184
Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 187

10 Coding Theory 189

10.1 Errors; Noise . . . . . . . . . . . . . . . . . . . . . . . . . . . 189
10.2 The Venn Diagram Code . . . . . . . . . . . . . . . . . . . . 190
10.3 Binary Codes; Weight; Distance . . . . . . . . . . . . . . . . 192
10.4 Linear Codes . . . . . . . . . . . . . . . . . . . . . . . . . . . 195
10.5 Hamming Codes . . . . . . . . . . . . . . . . . . . . . . . . . 198
10.6 Codes and the Hat Problem . . . . . . . . . . . . . . . . . . 200
10.7 Variable-Length Codes and Data Compression . . . . . . . . 201
Exercises 10A . . . . . . . . . . . . . . . . . . . . . . . . . . 203
Exercises 10B . . . . . . . . . . . . . . . . . . . . . . . . . . 204
Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 205

11 Latin Squares 207

11.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . 207
11.2 Orthogonality . . . . . . . . . . . . . . . . . . . . . . . . . . 211
11.3 Idempotent Latin Squares . . . . . . . . . . . . . . . . . . . . 217
11.4 Partial Latin Squares and Subsquares . . . . . . . . . . . . . 219
11.5 Applications . . . . . . . . . . . . . . . . . . . . . . . . . . . 221
Exercises 11A . . . . . . . . . . . . . . . . . . . . . . . . . . 225
Exercises 11B . . . . . . . . . . . . . . . . . . . . . . . . . . 228
Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 229

12 Balanced Incomplete Block Designs 231

12.1 Design Parameters . . . . . . . . . . . . . . . . . . . . . . . . 232
12.2 Fisher’s Inequality . . . . . . . . . . . . . . . . . . . . . . . . 236
12.3 Symmetric Balanced Incomplete Block Designs . . . . . . . . 238
12.4 New Designs from Old . . . . . . . . . . . . . . . . . . . . . 240
12.5 Difference Methods . . . . . . . . . . . . . . . . . . . . . . . 242
Exercises 12A . . . . . . . . . . . . . . . . . . . . . . . . . . 245
Exercises 12B . . . . . . . . . . . . . . . . . . . . . . . . . . 247
Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 248
xii

13 Linear Algebra Methods in Combinatorics 251

13.1 Recurrences Revisited . . . . . . . . . . . . . . . . . . . . . . 251
13.2 State Graphs and the Transfer Matrix Method . . . . . . . . 253
13.3 Kasteleyn’s Permanent Method . . . . . . . . . . . . . . . . 260
Exercises 13A . . . . . . . . . . . . . . . . . . . . . . . . . . 265
Exercises 13B . . . . . . . . . . . . . . . . . . . . . . . . . . 267
Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 268

Appendix 1: Sets; Proof Techniques 271

A1.1 Sets and Basic Set Operations . . . . . . . . . . . . . . . . . 271
A1.2 The Principle of Mathematical Induction . . . . . . . . . . . 279
A1.3 Some Applications of Induction . . . . . . . . . . . . . . . . 281
A1.4 Binary Relations on Sets . . . . . . . . . . . . . . . . . . . . 283
Exercises A . . . . . . . . . . . . . . . . . . . . . . . . . . . 285
Exercises B . . . . . . . . . . . . . . . . . . . . . . . . . . . 286

Appendix 2: Matrices and Vectors 291

A2.1 Definitions . . . . . . . . . . . . . . . . . . . . . . . . . . . . 291
A2.2 Vector and Matrix Products . . . . . . . . . . . . . . . . . . 294
A2.3 Inverses . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 296
A2.4 Determinants . . . . . . . . . . . . . . . . . . . . . . . . . . 298
Exercises A . . . . . . . . . . . . . . . . . . . . . . . . . . . 299
Exercises B . . . . . . . . . . . . . . . . . . . . . . . . . . . 301

Appendix 3: Some Combinatorial People 305

Solutions to Set A Exercises 313

Hints for Problems 341

Solutions to Problems 347

References 367

Index 375
List of Figures

1.1 Pancake with three slices . . . . . . . . . . . . . . . . . . . . 3

1.2 A sudoku puzzle . . . . . . . . . . . . . . . . . . . . . . . . . 5
1.3 The Tower of Hanoi puzzle . . . . . . . . . . . . . . . . . . . 6
1.4 Königsberg and its bridges . . . . . . . . . . . . . . . . . . . . 11
1.5 Some Venn diagrams . . . . . . . . . . . . . . . . . . . . . . . 14
1.6 K6 and K3,4 . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
1.7 A graph and its complement . . . . . . . . . . . . . . . . . . . 19

2.1 How many ways to walk? . . . . . . . . . . . . . . . . . . . . 31

3.1 K5 in two colors . . . . . . . . . . . . . . . . . . . . . . . . . 51

3.2 Decomposition of K8 proving R(3, 4) ≥ 9 . . . . . . . . . . . 53

4.1 Enumerating X ∪ Y using (4.1) . . . . . . . . . . . . . . . . . 69

4.2 Illustration of (4.2) (left) and (4.3) (right) . . . . . . . . . . . 71

6.1 Triangulating the hexagon . . . . . . . . . . . . . . . . . . . . 104

6.2 Rooted binary trees with n internal points . . . . . . . . . . . 107

7.1 Square with numbered corners . . . . . . . . . . . . . . . . . 125

7.2 Three hexagons with labeled vertices and sides . . . . . . . . 129
7.3 Two figures for coloring . . . . . . . . . . . . . . . . . . . . . 133
7.4 Coloring corners or edges . . . . . . . . . . . . . . . . . . . . 135
7.5 Square with eight points to color . . . . . . . . . . . . . . . . 142

8.1 Testing degree sequences . . . . . . . . . . . . . . . . . . . . . 144

8.2 Tabular form of Dijkstra’s Algorithm . . . . . . . . . . . . . . 150
8.3 Two representations of K4 . . . . . . . . . . . . . . . . . . . . 151

9.1 The original Königsberg bridges . . . . . . . . . . . . . . . . . 159

9.2 The Königsberg bridges today . . . . . . . . . . . . . . . . . . 163
9.3 Find an efficient eulerization . . . . . . . . . . . . . . . . . . . 164
9.4 Graphical representation of a maze . . . . . . . . . . . . . . . 165
9.5 Two graphs from the map . . . . . . . . . . . . . . . . . . . . 166
9.6 A labeling of K4 . . . . . . . . . . . . . . . . . . . . . . . . . 169
9.7 All Hamilton cycles of the K4 . . . . . . . . . . . . . . . . . . 170
9.8 Some trees . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 170

xiii
xiv

9.9 Illustration of Lemma 9.3 . . . . . . . . . . . . . . . . . . . . 171

9.10 Graphs for counting trees . . . . . . . . . . . . . . . . . . . . 176
9.11 Counting trees . . . . . . . . . . . . . . . . . . . . . . . . . . 178

10.1 Venn diagram encoding . . . . . . . . . . . . . . . . . . . . . 191

10.2 Venn diagram decoding . . . . . . . . . . . . . . . . . . . . . 192
10.3 The codewords in Venn diagram encoding . . . . . . . . . . . 193
10.4 The Morse code . . . . . . . . . . . . . . . . . . . . . . . . . . 201

11.1 A golf design for seven clubs . . . . . . . . . . . . . . . . . . . 224

11.2 Array to generate a golf design for 17 teams . . . . . . . . . . 225

12.1 Residual and derived designs . . . . . . . . . . . . . . . . . . 242

12.2 Some difference sets . . . . . . . . . . . . . . . . . . . . . . . 244
12.3 Some supplementary difference sets . . . . . . . . . . . . . . . 245

13.1 Gear shift graph . . . . . . . . . . . . . . . . . . . . . . . . . 253

13.2 State graph and adjacency matrix . . . . . . . . . . . . . . . 254
13.3 A diagram and two adjacency matrices . . . . . . . . . . . . . 255
13.4 Digraph and adjacency matrix for the string problem . . . . . 257
13.5 Digraph and adjacency matrix . . . . . . . . . . . . . . . . . 258
13.6 Digraph and adjacency matrix for domino tiling . . . . . . . . 260

A.1 R ∩ (S ∪ T ) = (R ∩ S) ∪ (R ∩ T ) . . . . . . . . . . . . . . . . 276
Preface

This book is a text for introductory courses on combinatorics, usually offered

around the junior year. The audience for such a course consists primarily of
mathematics majors, although the course is often taken by computer science
students and occasionally by electrical engineering students.
The preparation of the students for these courses is very mixed, and one
cannot assume a strong background. In some cases this will be the first
course where the students see several real proofs, while others will have a
good background in linear algebra, will have completed the calculus stream,
and will have started abstract algebra. For this reason we have included
two chapters on background material — one on sets, induction and proof
techniques, and one on vectors and matrices. We have made them appendices
so that students can get straight to the meat of the course; the better-prepared
students can ignore the appendices entirely, while others can refer to them as
needed.
Some textbooks contain biographical material on the mathematicians who
invented the concepts that are studied, but students are often irritated by
the interruption. Others omit this; the reader misses out on some interesting
information, and a different set of students is unhappy. In an attempt to get
around this problem, we have included some brief sketches, but have made
them into a separate section of the book — a third appendix. We refer readers
to the biographies when relevant to the topic being studied.
Not all books discuss the increasing role played by technology (computer
algebra systems, among many other examples); we have added a discussion of
Maple and Mathematica where appropriate and commented on other techno-
logical tools (e.g., spreadsheets and calculators). These are not emphasized,
so that students or instructors who prefer not to have them do not need access
to any of these tools. Similarly we have used more modern examples for some
of the traditional problems and exercises (e.g., digital music tracks in MP3
players, rather than books on a shelf).

What Is Combinatorics Anyway?

Broadly speaking, combinatorics is the branch of mathematics that deals
with different ways of selecting objects from a set or arranging objects. It
tries to answer two major kinds of questions, namely counting questions (how
many ways can a selection or arrangement be chosen with a particular set of
properties?) and structural questions (does there exist a selection or arrange-

xv
xvi

ment of objects with a particular set of properties?).

The Book: Problems and Exercises

Exercises are collected at the end of the chapters. They are divided into
three parts. Exercise sets A and B are fairly straightforward, and the only
difference is that answers and partial or complete solutions are provided for
the A exercises. Many textbooks provide answers for every second exercise or
for what appears to be a random selection, but we have found that the “two
sets” model works well in the classroom. There is also a set of “Problems.”
These contain some more difficult or more sophisticated questions, and also a
number of exercises where the student is asked to provide a formal proof when
a result has been treated informally in the text. We have provided solutions
for some Problems, and hints for some others. There are also Exercises, but
not Problems, in the Appendices on background material.

The Book: Outline of the Chapters

We start by briefly discussing several examples of typical combinatorial
problems, to give the reader a better idea of what the subject covers. Chapter
1 also contains some basic information on sets, proof techniques, enumeration
and graph theory, topics that will reappear frequently in the book. The next
few chapters explore enumerative ideas, including the pigeonhole principle
and inclusion/exclusion.
In Chapters 5 through 7 we explore enumerative functions and the relations
between them. There are chapters on generating functions and recurrences,
on some important families of functions, or “numbers” (Catalan, Bell and
Stirling numbers), and the theorems of Pólya and Redfield. These chapters
also contain brief introductions to computer algebra and group theory.
The next five chapters study structures of particular interest in combi-
natorics: graphs, codes, Latin squares and experimental designs. The last
chapter contains further discussion of the interaction between linear algebra
and combinatorics.
We conclude with the Appendices mentioned earlier, Solutions and Hints,
and the Bibliography.

Acknowledgments
This book has benefited significantly from the comments of Ken Rosen, and
those of anonymous reviewers. We would also like to thank David Grubbs and
Bob Stern of Taylor & Francis for their assistance.
The first edition of any textbook will, unavoidably, contain errors. Each of
us wishes to state categorically that all such errors are the fault of the other
author.
Chapter 1
Introduction

Broadly speaking, combinatorics is the branch of mathematics that deals with

different ways of selecting objects from a set or arranging objects. It tries to
answer two major kinds of questions, namely the existence question (Does
there exist a selection or arrangement of objects with a particular set of prop-
erties?) and the enumerative question (How many ways can a selection or
arrangement be chosen with a particular set of properties?). But you may be
surprised by the depth of problems that arise in combinatorics.
The main point to remember is that it really doesn’t matter what sort
of objects are being discussed. For example, we shall often assume that we
are talking about sets of numbers, and sometimes use their arithmetical or
algebraic properties. But these methods are used to prove results that we
then apply to all sorts of objects.
In the first section we shall show you a few examples of combinatorial
problems. In Section 1.2 we briefly summarize some ideas of and notations
of set theory (if you need to review this material, see Appendix 1). The
remainder of the chapter introduces some basic combinatorial ideas.

1.1 Some Combinatorial Examples

Some of these problems have a recreational flavor, puzzles and so on, be-
cause they will be more familiar, but all these ideas have very serious appli-
cations. We address many of them in more detail in subsequent chapters.

Passwords
We start with an enumerative problem. Enumeration (the theory of count-
ing) is an important part of combinatorics.
Most computer systems require passwords, in order to protect your informa-
tion and your privacy. For example, the social networking program YoufaceTM
requires everybody to have an eight-character password made up of letters and
digits. The passwords are case-sensitive, but the letter O is not allowed (to
avoid confusion with zero), so there are 60 symbols available. How many
different passwords could you choose?

1
2 Introduction to Combinatorics

There are 60 choices for the first character. For each of those there are 60
possible second characters, for 3600 possible 2-character starts. And so on.
In all, there are 608 , or about 168 trillion, passwords. This calculation is an
example of the Multiplication Principle, which we’ll discuss further in Section
1.3, later in this chapter.
Suppose a hacker wants to break into your YoufaceTM account. She has a
program that can generate and test a thousand passwords per second. Is she
dangerous?
Well, if she tests every possibility, it will take her over 5,000 years. So no,
your password is pretty secure.
However, you need to be careful about passwords. Dr John Walters, a
computer science professor that we shall meet again in this volume, always
uses the login jwalters. Having a poor memory, he always starts his password
with his initials, jw, and ends with the last two digits of the year. The hacker
found this out, so in 2009 she worked out that his password had the form
jw****09. There are 604 = 12, 360, 000 passwords of this form. That sounds
like a lot, but in 3.6 hours she could run every possible password. Even if
she is very unlucky, and her computer does not find the actual password until
very late in the run, she could still hack into his account in an afternoon.
In order to protect people like Dr Walters, YoufaceTM introduced some
further requirements. Your password cannot begin with the same two symbols
as your username, and the last two digits cannot be the last two digits of the
year. In 2010, when he rebuilt his account, Dr Walters could not choose any
password of the form jw****** or ******10.
How many possible passwords remained? We start with 608 possibilities,
and subtract the number that were banned. There were 608 passwords origi-
nally. There are 606 passwords of type jw******, and 606 of type ******10.
Subtracting these numbers leaves 608 − 2 × 606 . However, we have taken away
some passwords twice: those of form jw****10. There are 604 passwords like
this, so we have “oversubtracted” 604 from the total. So the final result is
608 − 2 × 606 + 604 .
This method of counting — subtract all objects in one class, do the same
with those in another class, then add back those that were common to both
classes — is the first case of the Principle of Inclusion and Exclusion. We
shall deal with this further in Chapter 4.

The Pancake Problem

Another good example of a combinatorial problem is this: Into how many
regions can the plane be divided by n straight lines, given that no lines are
parallel and at most two lines intersect at any point? These conditions ensure
the maximum number of regions. This is sometimes called the Pancake Prob-
lem because we may draw a large enough circle (the “pancake”) surrounding
all points of intersection of the lines, so that the problem may be described
 Introduction 3

as: What is the maximum number of pieces remaining after cutting a pancake
with n cuts (none parallel and no three concurrent)?

2
1
3
5
4
6
7

FIGURE 1.1: Pancake with three slices

The first thing we do is define a notation and try a few examples. This
is almost always a good start to any kind of problem. We let Pn be the
maximum number of pieces. Then it is easy to see that P0 = 1 (with no cuts,
the pancake is still one piece) and P1 = 2. When we notice that P2 = 4,
we see the progression 1, 2, 4 and wonder whether we might have powers of
two; perhaps, we think, Pn = 2n ? This would make sense if each cut passes
through every region, so that each region is cut into two. However, this does
not happen, and we find P3 = 7, as we can see from Figure 1.1. A few more
tries will give us P4 = 11 and P5 = 16. Clearly, another approach is needed.
How many new regions do we get by adding a new cut? If we have n − 1
cuts in the pancake, and add another cut, the rules require that the new cut
must intersect each of the n − 1 old cuts. If we start the cut on the edge of
the pancake in the middle of a region, we cut along the pancake and cut off a
new region when we reach the first of the n − 1 old cuts. Then as we continue
cutting in a straight line, we intersect each of the other old cuts, cutting off a
new region when we do. Then at the end, we reach the edge of the pancake,
and we cut off a final piece. This means that we have added n pieces; one
for each of the original n − 1 cuts, and one when we reach the edge of the
pancake. This tells us that Pn = Pn−1 + n. This is a formula for Pn , but not
a nice one because we cannot easily calculate, for example, P100 without a
great deal of effort. We will work a little harder and see whether we can find
a more efficient formula.
We start by repeating the formula for larger values of n. We get P1 = P0 +1,
and P2 = P1 +2 = P0 +1+2, then P3 = P0 +1+2+3, P4 = P0 +1+2+3+4, and
so on. The pattern becomes clear, and we write Pn = P0 +(1+2+· · ·+n). Now
we nearly have our answer. The reader may be familiar with the formula for
1 + 2 + · · · + n or of the first n terms of any arithmetic progression; the story
goes that the eminent mathematician Carl Friedrich Gauss (see Appendix
C) found the sum of the first one hundred integers almost immediately by
realizing how the formula worked (for more details, see [15], p. 509). The
4 Introduction to Combinatorics

“two-rows proof” is as follows:

1 2 3 ... n−1 n
n n−1 n−2 ... 2 1

Each of these rows clearly sums to the same value, S. If we add the columns,
we find that each column adds to n+1, and there are n columns; thus the sum
of all elements in both rows is 2S = n(n+1). Thus 1+2+· · ·+n = n(n+1)/2.
This gives us, finally, our formula for Pn , the maximum number of pieces
after n cuts.
n(n + 1)
Pn = 1 +
2

Sudoku
Over the last few years we have seen a great surge of interest in the sudoku
puzzle. For those few readers who have been vacationing on another planet,
the puzzle consists of a 9 × 9 array, partitioned into 3 × 3 subarrays (sub-
squares), in which some cells contain entries chosen from {1, 2, 3, 4, 5, 6, 7, 8, 9}.
The object is to complete the array so that every row, every column and every
subarray contains each symbol precisely once.
Sudoku was apparently invented by an American architect, Howard Garns
(see Appendix 3), in 1979, but did not achieve great popularity until 1986,
when it started appearing in Japanese publications. Worldwide circulation
dates to about 2004. A nice history can be found in [132].
We’ll use the phrase sudoku square for a completed 9 × 9 square that could
be the solution to a sudoku puzzle. In a well-posed puzzle, there is one and
only one sudoku square that completes the original partial array.
Figure 1.2 shows a typical sudoku puzzle in the top left-hand position. It
needs a 1 in column 3. There are already 1’s in the top left and bottom left
subsquares, so there is only one possible space in the column; the top right
array shows the 1 in place.
Now look at the top right subsquare. It needs a 2, 4, 5 and 7. The only
possible place for a 4 in this subsquare is the middle top. At first it looks
as though the 5 can go in any of the other three vacant cells. However, the
second row is not available (the 2 must go there), and the 7 must go in the
last column (there is already a 7 in the seventh column, and the 4 filled up
the eight column). So the 5 only has one possible home. The completion of
the top right subsquare is shown in the lower left copy.
Continuing in this way, we obtain the sudoku square shown at the lower
right of the figure.
The methods used in solving sudoku are combinatorial in nature. However,
not many of the millions who play sudoku realize that a sudoku square is
an example of a combinatorial object, called a Latin square, that is widely
used in statistical and geometric applications. We shall study Latin squares
in Chapter 11.
 Introduction 5

9 6 2 8 1 9 6 2 8
1
1 8 3 1 8 3
2 1 6 9 2 1 6 9
7 8 4 3 7 8 4 3
5 9 4 6 5 9 4 6
6 3 7 8 6 1 3 7 8
5 4 7 8 5 4 7 8
1 9 5 1 9 5
3 2 5 9 4 3 2 5 9 4

9 6 2 8 5 4 1 9 7 3 6 2 8 5 4 1
1 2 8 3 6 1 5 9 4 7 2 8 3
2 1 6 9 7 8 4 2 5 3 1 6 9 7
7 8 4 3 7 2 8 1 6 4 9 3 5
5 9 4 6 5 3 9 8 7 2 4 1 6
6 1 3 7 8 4 6 1 3 9 5 7 2 8
5 4 7 8 2 5 4 7 1 3 8 6 9
1 9 5 1 9 7 4 8 6 3 5 2
3 2 5 9 4 3 8 6 2 5 9 1 7 4
FIGURE 1.2: A sudoku puzzle

The Tower of Hanoi

Virtually every student in combinatorics and computer science runs into
the Tower of Hanoi problem. Because it is such a basic example of combina-
torial reasoning and so quickly yields to straightforward methods of attack,
and because it may be used to illustrate such ideas as recurrence relations
and recursive programming and has ties to such topics as Gray codes, it finds
its way into textbooks in programming, algorithms, and discrete and combi-
natorial mathematics. The puzzle itself predates the modern study of these
topics; it was first described in 1883 by Edouard Lucas (see Appendix 3), who
also publicized the Fibonacci numbers (as discussed in Chapter 5). Lucas
introduced the puzzle under the name M. Claus. One year later, Henri de
Parville published the following fanciful legend concerning the puzzle [25]:
In the great temple at Benares, beneath the dome which marks
the center of the world, rests a brass plate in which are fixed three
diamond needles, each a cubit high and as thick as the body of a
bee. On one of these needles, at the creation, God placed sixty-
four discs of pure gold, the largest disk resting on the brass plate,
and the others getting smaller and smaller up to the top one. This
is the Tower of Bramah. Day and night unceasingly the priests
transfer the discs from one diamond needle to another according
to the fixed and immutable laws of Bramah, which require that
the priest on duty must not move more than one disc at a time
6 Introduction to Combinatorics

and that he must place this disc on a needle so that there is no

smaller disc below it. When the sixty-four discs shall have been
transferred from the needle on which at the creation God placed
them to one of the other needles, tower, temple, and Brahmins
alike will crumble into dust, and with a thuderclap the world will
vanish.
(translation by W. W. Rouse Ball [88]). The legend, oddly, places the towers
in a temple to the god Brahma in Benares (modern Vanarasi, India), which
renders the reference to Hanoi (in Vietnam, over 1000 miles away) puzzling.
In essence, the puzzle is this; we have three “needles” or “spindles” and a
set of n differently-sized disks with holes in the middle to allow the disks to
be placed over the needles. The disks are arranged on the leftmost needle in
order of diminishing size (smallest on the top). We are to move the entire
column to a different needle (say, the rightmost) according to the rules:
• At each play, precisely one disk is moved.
• No disk may be placed over a spindle that already has a smaller disk.
An illustration for the case n = 3 is shown below. Commercial versions of the
puzzle typically include from five to eight disks.
The problem is twofold; first, what is the most efficient way of moving the
tower? Secondly, how many plays does this most efficient way take for a tower
of n disks? One might add a third (or more properly, first) question: Can it
be done at all? A careful analysis will answer all three questions.
First of all, observe that the problem is trivial in the cases of n = 1 (simply
move the disk from the starting needle to the ending needle) and n = 2 (move
the smallest disk to the middle needle; then the larger disk to the ending
needle; then place the smallest over the largest). Either of these may form the
basis case for a proof by mathematical induction. (If you are unfamiliar with
induction methods, or need to revise this material, see Section 1.2, below.)
Suppose that it is known that we can solve the puzzle for n disks, and we
have n+1 disks before us. We may use the solution for n disks to move the top
n disks to the middle needle, ignoring for the moment the largest disk. Now
we may freely move the largest disk from the leftmost needle to the rightmost
needle. At this point, we use the solution for n disks to move the n disks

FIGURE 1.3: The Tower of Hanoi puzzle

 Introduction 7

from the middle needle to the rightmost needle. This will cause no problems
since the disk on the rightmost needle is the largest, so that any disk may be
placed on it without violating a rule.
It is at least intuitively clear that this technique gives us the most efficient
way to move a stack of n disks. Clearly the basis case uses the fewest possible
moves (one move for one disk, three moves for two disks). Also, it seems clear
that there is no shorter way to move n + 1 disks than by moving the stack
of n disks first. Without going through a rigorous proof, it is at least very
plausible that our approach uses the fewest moves possible.
Now the hard part: How many plays does this method take for n disks?
We denote this (still unknown) quantity by Hn . Clearly, we have H1 = 1 and
H2 = 3. For any n > 2, we can compute Hn = 2 · Hn−1 + 1. This follows
because we use Hn−1 plays to move the n − 1 disks to the middle needle, one
play to move the largest disk to the rightmost needle, and another Hn−1 plays
to move the n − 1 disks from the middle to the rightmost needle.
This kind of formula for Hn , like the formula for Pn of the previous example,
is called a “recurrence” because the expression we are solving for “recurs” in
the equation. This does not give us a simple formula for Hn , but it gives us
a way to calculate it. We will run through a few values.

n 1 2 3 4 5
Hn 1 3 7 15 31

At this point, we might be tempted to conjecture, based on the numbers

shown, that Hn is always one less than a power of two, or more precisely,
Hn = 2n − 1. How might we prove this? Again, proof by induction is the
easiest approach with which we are familiar (another method will be presented
in Chapter 5). The basis is n = 1; suppose that Hn = 2n − 1 and use the
recurrence to find Hn+1 , which our guess says should be 2n+1 − 1. The
recurrence tells us that Hn+1 = 2 · Hn + 1. The induction hypothesis allows
us to replace Hn by 2n − 1, yielding Hn+1 = 2 · (2n − 1) + 1. We simplify to
obtain Hn+1 = 2n+1 − 1, as required.

The Seven Wargamers

A group of friends (Adam, Beth, . . . ) meet frequently to play the wargame
Vampire Cat Wars, a game for four individual players (no partnerships) in
which the players are vampires who have to beat off attacks by the others.
They decide to conduct a tournament over several sessions. In each session
four of the players will take part in one game. They would like it if each
competitor played against each of the others equally — in this case let’s call
the competition equitable. Various questions come up: for example, what is
the minimum number of games in such a tournament?
When they first decided to organize a tournament, there were seven players,
initials A, B, C, D, E, F, G. At first, they thought maybe everyone could play
8 Introduction to Combinatorics

everyone else once. If Adam is going to play every other competitor exactly
once, he would play in two matchups; writing the initials of each player to
represent a game, the schedule could include ABCD and AEFG. But who will
Beth play? One matchup is already given, ABCD. She cannot meet A, C or
D again, so the only possibility is BEFG. But a schedule cannot include both
AEFG and BEFG: pairs like EF would meet twice.
So they decided to try a schedule where everyone plays everyone else twice.
Everybody will have four games (each person must meet six opponents twice
each — 12 appearances — and each game accounts for three). We may as well
assume Adam plays in ABCD. He must play against Beth again, and the two
matches AB... could have a total of four, three or two players in common. If
we try four players — that is, use the matchup ABCD again — Adam’s other
two games must be AEFG twice. But Beth has been scheduled against A, C
and D twice, so her other two matchups must each be BEFG, and EF (as well
as other combinations) has occurred too many times. Similarly, if there are
two games with three players in common, say ABCD and ABCE, we strike
problems: Adam plays twice more, and must meet both F and G twice. The
only possibility is ADFG, AEFG. But the same argument shows that we must
have BDFG and BEFG. Now F and G have played each other too often.
Okay, this argument shows that the two AB... matchups have no other
member in common, and the same must apply to all other pairs. So no
two games can have three players in common. Try to start with ABCD and
ABEF. Adam’s other two matchups have the form A...G, in order for him to
meet Greg twice. They can’t use B again, and ACDG would have three in
common with ABCD, so they try ACEG and ADFG (or ACFG and ADEG,
which would be equivalent, just exchange two names). The same argument,
avoiding three in common with the first two games, leads to either BCEG or
BCFG, and BCEG clashes with ACEG, so they chose BCFG and BDEG. The
final matchup is CDEF, and the schedule is

ABCD, ABEF, ACEG, ADFG, BCFG, BDEG, CDEF.

The players are very happy — and then Helga joins their group. What
now? And what if other players join up?
Let’s write v for the number of players, and say each pair play together λ
times. Then each player has λ(v −1) instances of an opponent. Each matchup
accounts for three of these. So each player takes part in λ(v − 1)/3 games.
You can’t schedule partial games, so λ(v − 1) must be a multiple of 3. So, for
v = 8, λ must be 3 or 6 or . . . . For the 8-player version, the smallest possible
case is λ = 3, in which case there will be 14 games and each player is in three
of them.
The wargamers didn’t feel up to calculating a schedule for 14 games, so
they started playing the 7-player version again. Each week they scheduled
one game. Each week the other three players came to watch, and played a
game with Helga. And, at the end of seven weeks, they realized they had
 Introduction 9

played an equitable 8-player tournament

ABCD, ABEF, ACEG, ADFG, BCFG, BDEG, CDEF,

EFGH, CDGH, BDFH, BCEH, ADEH, ACFH, ABGH.

The problem here is, given a set with v elements (the players), to choose a
collection of subsets of equal size (4 in the example, more generally written
k) such that every pair of elements occurs in the same number of subsets (our
λ). This is called a balanced incomplete block design, and we’ll study them
in Chapter 12. The word “balanced” refers to fact that pairs occur together
equally often, “incomplete” to the fact that k is smaller than v (or else there
would be no problem), and the equal-size subsets are called “blocks.” It is
called a design because the main application is in the design and statistical
analysis of experiments. The numbers v, k and λ are the parameters of the
design (together with b, the number of blocks, and r, the number of blocks
in which a given object occurs; but b and r can be calculated from the other
parameters). It is common to refer to these designs by specifying their pa-
rameters, for example one says “a (v, b, r, k, λ)-design”; the wargamers’ first
example is a (7, 7, 4, 4, 2)-design.
We have already seen that some sets of parameters are impossible, because
r and b have to be integers. But there are other cases, for example v = 15, k =
5, λ = 2 (for which r = 7 and b = 21, both whole numbers) for which there
is no design; and there is an infinitude of parameter-sets for which we do not
know whether a design exists.
The second example — the 8-player schedule — is an (8, 14, 7, 4, 3)-design
in which the blocks can be partitioned into sets, where every object belongs
to exactly one member of each set. This special case is called a resolvable
balanced incomplete block design. These designs are of special interest for
statisticians and also arise in finite models of Euclidean geometry.

The Hat Game

Here is the outline of a new television game show. There is a team of three
contestants. Just before the show begins, the players are taken into separate
rooms. Each is blindfolded and a hat — either red (R) or black (B) — is put
onto his or her head. The blindfolds are removed, but the player cannot see
which color hat they have been given. The players are then taken into the
studio and line up behind three desks; they can see each others’ hats, but still
not their own. No communication between them is allowed.
The M.C. says, “Players, do you think your hat is red or black? Please
write down your answer. If you don’t know, you can pass — don’t write
anything.”
Then the players’ “votes” are revealed. If any of them has made the wrong
guess, the team loses. If they all pass, they lose. If at least one player gets the
right answer, and there are no wrong answers, they win three milion dollars.
10 Introduction to Combinatorics

As you would expect, the players meet beforehand to discuss their options
and choose a strategy. The organizers say they decide on the hat colors by
tossing a coin for each player, so that the allocations are independent and
each player has a 50-50 chance of red or black. We’ll assume they are telling
the truth, and we’ll also assume the players don’t try to concoct any type of
cheating scheme. What is the team’s best plan?
One obvious strategy would be for the team to nominate one of their number
who will always say “red” while the other two players always pass. This
would give them a 50% chance of winning the prize. Can they do better?
Most people, and in fact most mathematicians, initially think not. Since
each person’s hat color is independent of the other players’ colors and no
communication is allowed, it seems impossible for them to learn anything just
by looking at one another; all the players can do, it seems, is guess.
But in three out of four cases, we expect two of the players will have hats of
the same color and the third player’s hat will be the opposite color. The group
can win every time this happens if the players follow the following strategy:
Once the game starts, look at the other two players’ hats. If you see two hats
of different colors, pass. If they are the same color, guess that your hat is
the opposite color. Every time the hat colors are distributed two and one,
one player will guess correctly and the others will pass, and the team will win
the game. When all the hats are the same color, all three players will guess
incorrectly and the group will lose; but on average this will happen in only a
quarter of cases.
There are eight ways the hats can be distributed: RRR, RRB, RBR, RBB,
BRR, BRB, BBR and BBB (from left to right). Suppose the game was played
eight times, and suppose it just so happened that every possible combination
occurred once. In the first and last cases, all three players make the wrong
guess; in the other six there is just one correct guess. So, of the twelve guesses,
half are right and half are wrong, as you would expect.
How about more players? With four team members, the easiest strategy is
to tell one player to stay silent throughout. The other three don’t even look
at that player! They play as they would if Number Four was not there: each
looks at the remaining two players and proceeds as if it was a team of size
three. They still have a 75% chance of winning.
The Hat Game problem was introduced by Todd Ebert in his dissertation
[27], and is in fact related to error-correcting codes. It has been discussed in
the press [86], and a recent survey paper is [16]. We’ll examine the problem
further, and explain the connection to error-correcting codes, in Chapter 10.

The Bridges of Königsberg

Leonhard Euler (see Appendix 3), a pioneering Swiss mathematician and
physicist, is seen by most as the leading mathematician of the Eighteenth
Century. Euler (pronounced OY-ler) founded the area of combinatorics called
Graph Theory (see Section 1.1.4, Chapter 8 and Chapter 9) and invented
 Introduction 11

the concepts of Latin Squares and orthogonality (see Chapter 11.1). He also
made equally significant contributions to analytic geometry, trigonometry,
geometry, calculus and number theory.
In 1735, Euler spoke to the St. Petersburg Academy on a problem in recre-
ational mathematics. The Prussian city of Königsberg was set on both sides
of the Pregel River. It included a large island called The Kneiphof, and the
river branched to the east of it. So there were four main land masses — let’s
call them the four parts of the city, and label them A, B, C, D — connected
to each other by seven bridges, as shown in the rough map in Figure 1.4:

B
D

FIGURE 1.4: Königsberg and its bridges

The problem was to find a walk through the city that would cross each
bridge once and only once, and visit all four parts. The only way to go
between the different pieces of land was to cross the bridges.
So we have a combinatorial problem. We have a set with seven elements —
the seven bridges — and we need to find out whether they can be arranged in
a certain way, specifically so that the corresponding bridges could be traversed
in the given order.
The solution is very easy. Look for example at the north bank of the river,
part A. Whenever you walk over a bridge that leads to A, your next bridge
must take you back out of A. So walking into A and out again uses up two
bridges (we say “uses them up,” because you can’t cross a bridge twice).
There are three bridges; the walk must take you onto A and off again once,
and there is one bridge left over. The only possibility is that the walk must
start or finish in A (but not both).
The same argument can be applied to C and D, (with three bridges each),
and to B (five bridges, so you walk through B twice). Each part of town is
either the start or finish. But a walk can have only one start, and only one
finish, and there are four parts of town! This is clearly impossible, so there is
no such walk.
Of course, Euler proved that no walk is possible. But he did much more.
As we said, he essentially invented graph theory and showed how it could be
used to represent any set of islands and bridges, or indeed any set of towns and
roads joining them, and concocted an algorithm for traversability problems.
12 Introduction to Combinatorics

We shall look at his methods in Section 9.9.1.

Six Degrees of Separation

We have all experienced the following phenomenon: meeting a perfect
stranger, and subsequently finding out that we have a common acquaintance.
While this is very rare, a little further investigation may well show that the
new person is a “friend of a friend” or a “friend of a friend of a friend.”
Guglilmo Marconi, the pioneer of radio, thought that these phenomena were
becoming far more common because of the growth of communication. In his
1909 speech accepting the Nobel Prize, he suggested that the average two
people can be connected through a chain of at most five acquaintances (that
is, six links).
This idea has been explored by many people since. In particular, Harvard
psychologist Stanley Milgram [72, 113] studied the problem experimentally.
The following description of his experiment is based on [131].
1. Milgram typically chose individuals in Omaha, Nebraska and
Wichita, Kansas to be the starting points and Boston, Massa-
chusetts to be the end point of a chain of correspondence.
2. Information packets were sent to randomly selected individuals
in Omaha or Wichita. They included letters, which detailed the
study’s purpose, and basic information about a target contact per-
son in Boston, a roster on which they could write their own name,
and business reply cards that were pre-addressed to Harvard.
3. Upon receiving the invitation to participate, the recipient was
asked whether he or she personally knew the contact person in
Boston. If so, the person was to forward the letter directly to that
person. (Knowing someone “personally” was defined as knowing
them on a first-name basis.)
4. If the person did not personally know the target, they were to
think of a friend or relative they knew personally who they thought
more likely to know the target. They then signed the roster and
forwarded the packet to the friend or relative. A postcard was also
sent to Harvard.
5. When and if the package eventually reached the contact person
in Boston, the researchers could examine the roster to count the
number of times it had been forwarded from person to person.
Additionally, for packages that never reached the destination, the
incoming postcards helped identify the break point in the chain.
This experiment has a number of obvious flaws. Many people refused to
participate. In one case, 232 of the 296 letters never reached the destination.
However, completed chains seemed to run to six or fewer links. Subsequent
experiments have shown that this number is about right, and the phrase “six
degrees of separation” has become fashionable.
 Introduction 13

We can look at the world as a huge communication network. Smaller com-

munication networks include social networks like facebook and myspace, the
internal communications (or intranet) of a large company, the physical net-
work of wires connecting components of a computer system, and so on. The
average length of communication links could be very small (for example, if
all pairs are directly linked) or very large (some computer networks consist
of a ring in which each component is linked only to the pair of components
adjacent in the ring). Networks with a small average length are called small
world networks (as in the phrase, “it’s a small world”). The analysis of these
networks — for example, finding out whether a particular type of network ex-
hibits “small world” properties — is basically a combinatorial problem, and
again graph theory comes into play.

1.2 Sets, Relations and Proof Techniques

A set is a collection of objects with a well-defined rule, called the mem-
bership law, for determining whether a given object belongs to the set. The
individual objects in the set are called its elements or members and are said
to belong to the set. If S is a set and s is one of its elements, we denote this
fact by writing
s ∈ S.
While nearly all readers will be familiar with basic set theory, you may wish
to review the standard set ideas and notations in Appendix 1. In particular,
you need to know about the standard operations on sets: union, intersection,
complementation; you need to know de Morgan’s Laws; and you should be
familiar with the standard number sets, the set of integers, denoted Z, the
rational numbers Q, the real numbers R and the complex numbers C.
If n is a positive integer, the residue classes modulo n are the n sets
[0], [1], . . . , [n − 1], where [i] is the set of all integers that leave remainder
i on division by n. Members of the same residue class are called congruent
modulo n and we write y ≡ x(mod n). When no confusion arises we simply
write x instead of [x].
If you have done a little abstract algebra, you will know that the set Zn of
residue classes modulo n is an example of a finite group. We discuss groups
further in Chapter 7.
We often illustrate sets and operations on sets by diagrams. A set R is
represented by a circle, and the elements of R correspond to points inside
the circle. If we need to show a universal set (for example, if complements
are involved) then this universal set is shown as a rectangle enclosing all
the other sets. These illustrations are called Venn diagrams, because they
were popularized by John Venn (see Appendix 3) in 1880 [114, 115], although
14 Introduction to Combinatorics

similar diagrams were studied earlier by Leibniz and Euler, and related ideas
were studied by Lull as far back as the 13th Century (see, for example, [5]). It
has recently become fashionable to refer to “Euler diagrams,” but Venn was
primarily responsible for their modern usage.
For example, the Venn diagrams or R ∪ S, R ∩ S, R, R\S and R ∩ S ∩ T
are shown in Figure 1.5.

R S R S

R !S R"S
R R S

_
R R\S
R

S T

R"S"T

FIGURE 1.5: Some Venn diagrams

We define the cartesian product (or cross product) S × T of sets S and T

to be the set of all ordered pairs (s, t) where s ∈ S and t ∈ T :

S × T = {(s, t) : s ∈ S, t ∈ T }.

There is no requirement that S and T be disjoint; in fact, it is often useful to

consider S × S.
A (binary) relation ρ from a set S to a set T is a rule that stipulates,
given any element s of S and any element t of T , whether s bears a certain
relationship to t (written s ρ t) or not (written s \ρ t). Alternatively, we can
define a binary relation ρ from the set S to the set T as a set ρ of ordered
pairs (s, t), where s belongs to S and t belongs to T , with the notation that
s ρ t when (s, t) ∈ ρ and s \ρ t otherwise. This means that, formally, a binary
relation from S to T can be defined as a subset of the Cartesian product S ×T .
A relation α on a set A is called an equivalence relation on A if and only if
it is reflexive, symmetric, and transitive.
 Introduction 15

The obvious equivalence relation is equality, on any set. In sets other than
number sets, equal objects are often called “equal in all respects.” More
generally, an equivalence relation can be considered as a statement that two
objects are “equal in some (specified) respects.” One example, on the integers,
is the relation α, where a α b is true if and only if a = ±b — a and b have the
same absolute value. Another, is congruence, on the set of all plane triangles.
Equivalence relations give us an alternative way to discuss multiple ele-
ments. You can view a multiset as an object based on a set of distinct ele-
ments on which an equivalence relation has been defined, so that the elements
of a multiset are the equivalence classes of the underlying set.
One of the most important proof techniques for working with finite sets or
with sets of integers is the method of mathematical induction, the application
of the principle of mathematical induction:
Suppose the proposition P (n) satisfies
(i) P (1) is true; and
(ii) for every positive integer n, whenever P (n) is true, then P (n + 1) is
true.
Then P (n) is true for all positive integers n.
If you are not familiar with induction, or need to review the topic, we treat
it in detail in Appendix 1.

1.3 Two Principles of Enumeration

In this section we introduce two easy rules for enumerating the numbers of
elements of certain sets or certain types of arrangements.
The Addition Principle states that if we are building an arrangement of
objects starting with a set of mutually-exclusive beginning states, then the
number of arrangements is the sum of the number of arrangements starting
from each beginning state. This rather complicated-sounding sentence is far
more trivial than it might appear. For instance, suppose that a young woman
wants to go on a job interview, and has only one clean dress skirt and one clean
pair of dress slacks. With the skirt, she can create 3 presentable outfits; with
the pants, five. It follows that she has 3 + 5 = 8 ways to dress appropriately.
The Multiplication Principle states that if we are building an arrangement
of objects in stages, and the number of choices at each stage does not depend
on the choice made at any earlier stage, then the number of arrangements is
the product of the number of choices at each stage. Again, this principle is
simpler than it appears; if a restaurant, for example, offers two kinds of salad,
16 Introduction to Combinatorics

nine entrees, and seven desserts, the number of meals consisting of one salad
and one entree and one dessert is 2 × 9 × 7 = 126.
Although these principles seem painfully obvious, each may be proved by
results from set theory; for instance, the Multiplication Principle follows from
the fact that the cardinality of a Cartesian product of sets is the product of
the cardinalities of the underlying sets. However, we prefer to emphasize the
uses rather than the proofs of these principles.
Example 1.1: We wish to have lunch at a fast-food place, and there are
three within walking distance. One place offers 8 soft drinks, 7 sandwiches,
and 2 choices of a “side dish” (French fries or onion rings); another offers only
6 drinks, but 10 sandwiches and 2 sides; and the last (a coffee shop) offers
16 drinks, 5 sandwiches, and only one side. How many different lunches are
possible?
The Addition Principle says that (because we can eat lunch at only one
place) the answer is the sum of the number of ways to eat lunch at each place.
The Multiplication Principle says that each place has a number of options
equal to the product of the options at each stage: drink, sandwich, and side.
So the first restaurant offers us 8 × 7 × 2 = 112 possibilities; the second,
6 × 10 × 2 = 120; and the third, 16 × 5 × 1 = 80. Altogether, then, there are
112 + 120 + 80 = 312 possibilities for lunch. !
The next few examples will introduce some mathematical concepts that
may already be familiar to the reader. These concepts arise frequently in
enumeration problems. The reader should already be familiar with n! to
denote the product 1 · 2 · · · · · n. By convention, we assume 0! = 1! = 1.
Example 1.2: A set has n distinct elements. How many subsets does it
have?
We line the elements of the set up in some order; beneath each, we will
place a mark if the element is in our subset, and no mark if the element is
not. We will make n decisions as we go through the list of elements, and each
decision has two alternatives, mark or no mark. Since no decision will affect
any other decision, we see that there are 2n subsets of a set of n elements. !

Example 1.3: Suppose we have five tunes stored on our portable digital
music player, and we wish to listen to each of them one after another. How
many ways are there to arrange the playlist?
We have five tunes, so there are five possibilities for the first tune. Once
we have chosen this tune, then regardless of which tune we choose to play
first there are four tunes left (because we don’t wish to repeat a tune). After
we choose the second tune, there are three possibilities left, and so on. The
result is that we may choose a playlist in 5! = 5 × 4 × 3 × 2 × 1 ways, by
the Multiplication Principle. In more general terms, we find that there are n!
 Introduction 17

ways to order n distinct objects; so we can arrange 7 books on a shelf in 7!

ways, place 12 people into a line in 12! ways, and so forth. !

Example 1.4: A license plate consists of three letters of the alphabet and
three digits from 0 to 9, with the letters preceding the digits. How many
license plates are possible? How many are possible if we do not use any letter
twice? How many are possible if we may use letters twice but not digits? How
many if we do not use either twice?
We imagine six boxes in a row as shown.

For the first box, we may choose any of the 26 letters, so we have 26 choices.
For the first question, where we may repeat letters and digits, we have 26
possible choices for each of the next two boxes as well. The last three boxes
each have 10 possibilities. It follows from the Multiplication Principle that
there are 263 · 103 possible license plates. If we may not repeat a letter, then
we must work a little harder; the second box has only 25 possible letters, and
the third has only 24. We then get 26 · 25 · 24 · 103 possible plates. In the
same way, there are 263 · 10 · 9 · 8 plates where letters may be repeated but
digits may not; and 26 · 25 · 24 · 10 · 9 · 8 possible plates where neither may be
repeated. !

Example 1.5: How many n-digit numbers in base k are there if we do not
allow a leading digit of 0? How many numbers without leading zeros in base
k are there with at most n digits?
We recall that a number in base k has possible digits 0, 1, . . . k − 1, so that
each of the n digits may be assigned any of the k symbols, except for the
first. We apply the Multiplication Principle to see that there are (k − 1) · kn−1
possible numbers. To find how many numbers of at most n digits in base k
there are, we may simply permit leading zeros; so the number 1 (in any base)
might be considered a “five-digit number” 00001. We find that there are kn
such numbers. !

Example 1.6: A large corporation gives employees ID codes consisting of

two or three letters and two digits. How many codes are possible if we may
use any of the 26 upper-case letters and any of the digits from 0 to 9, and
letters precede digits?
The Multiplication Principle tells us that there are 262 · 102 possibilities for
the two-letter codes, and 263 · 102 for the three-letter codes. The Addition
Principle tells us that there are
262 · 102 + 263 · 102 = 67, 600 + 1, 757, 600 = 1, 825, 200
codes. !
18 Introduction to Combinatorics

1.4 Graphs
A great deal has been done in the special case where all sets have precisely
two elements. In that case the members of the universal set are often repre-
sented graphically, as points in a diagram, and the set {x, y} is represented
by a line drawn joining x and y. Provided the universal set is not empty, such
a structure is called a graph. The sets are called edges (or lines), and the
elements of the universal set are vertices (or points). The universal set is the
vertex-set of the graph. For any graph G, we write V (G) and E(G) for the
sets of vertices and of edges of G.
The edge {x, y} is simply written xy, when no confusion can arise; x and
y are called its endpoints. When x and y are endpoints of some edge, we say
they are adjacent and write x ∼ y for short; the vertices adjacent to x are
called its neighbors. The set of all neighbors of x is its (open) neighborhood,
N (x). If x and y are not adjacent we write x (∼ y.
Two vertices either constitute a set or not, so a graph can never contain two
edges with the same pair of vertices. However, there are some applications
where two edges joining the same vertices might make sense. For this reason
we sometimes talk about networks or multigraphs in which there can be several
edges joining the same pair of vertices; those edges are called multiple edges.
Another generalization is to allow loops, edges of the form xx. There is
no very good term for a graph-type structure in which loops are allowed,
and we will usually call one of these a “looped graph” or “looped network”
or“network with loops” although strictly speaking it is not a graph or network
at all. Moreover, when no confusion arises, the word “graph” can be used for
any generalized graph.
Any binary relation can be represented by a diagram. If ρ is a binary
relation on the set S, the elements of S are shown as vertices, and if xρy is
true, then an edge is shown from x to y, with its direction indicated by an
arrow. Provided the set S is finite, all information about any binary relation
on S can be shown in this way. Such a diagram is called a directed graph
or digraph, and the edge together with its arrow is called an arc. If ρ is
symmetric, the arrows may be dropped and the result is a graph (possibly
with loops).
Several families of graphs have been studied. Given a set S of v vertices,
the graph formed by joining each pair of vertices in S is called the complete
graph on S and denoted KS . Kv denotes any complete graph with v vertices.
As you would expect, we often call K3 a triangle. The complete bipartite
graph on V1 and V2 has two disjoint sets of vertices, V1 and V2 ; two vertices
are adjacent if and only if they lie in different sets. We write Km,n to mean a
complete bipartite graph with m vertices in one set and n in the other. K1,n
in particular is called an n-star . Figure 1.6 shows copies of K6 and K3,4 .
 Introduction 19

FIGURE 1.6: K6 and K3,4

Suppose H is a graph all of whose vertices and edges are vertices and edges
of some graph G — that is, V (H) ⊆ V (G) and E(H) ⊆ E(G). Then H is
a subgraph of G; we write H ≤ G. Every graph G has itself as a subgraph;
if H is a subgraph of G but H (= G, H is a proper subgraph of G, and we
write H < G. In particular, if S is some set of vertices of G then +S,G is
the subgraph consisting of all edges of G with both endpoints in S. If G
is a complete graph whose vertex-set contains S then the subscript “G ” is
dropped, and +S, is the complete subgraph based on S. Any subgraph of a
complete bipartite graph is itself called bipartite.
Instead of saying Figure 1.6 shows two graphs, we could say it is a single
graph that consists of two separate subgraphs, with no edges joining one part
to the other. We call such a graph disconnected ; a graph that is all in one piece
is called connected. The separate connected parts of a disconnected graph are
called its components.
The graph G is trivially a subgraph of the complete graph KV (G) . The set
of all edges of KV (G) that are not edges of G will form a graph with V (G) as
its vertex set; this new graph is called the complement of G, and written G.
More generally, if G is a subgraph of H, then the graph formed by deleting
all edges of G from H is called the complement of G in H, denoted H − G.
The complement K S of the complete graph KS on vertex set S is called a null
graph; we also write K v as a general notation for a null graph with v vertices.
Figure 1.7 shows a graph and its complement.

A A

B C B C

D E D E
G G
FIGURE 1.7: A graph and its complement
20 Introduction to Combinatorics

We define the degree or valency d(x) of the vertex x to be the number of

edges that have x as an endpoint. If d(x) = 0, then x is called an isolated
vertex. A graph is called regular if all its vertices have the same degree. If
the common degree is r, it is called r-regular. In particular, a 3-regular graph
is called cubic.

THEOREM 1.1 (Sum of Degrees)

In any graph or multigraph, the sum of the degrees of the vertices equals
twice the number of edges (and consequently the sum of the degrees is an even
integer).

Proof. Suppose the graph or multigraph has e edges; label the edges, say
y1 , y2 , . . . , ye . Consider a list in which each edge appears twice, once for each
endpoint. For example, if y1 has endpoints x4 and x7 , you might make entries
y1 : x4 and y1 : x7 . Vertex x will appear in precisely d(x) entries, so the total
number of entries equals the sum of the degrees of the vertices. On the other
hand, each edge appears twice, so the total number of entries equals twice the
number of edges. !

COROLLARY 1.1.1
In any graph or multigraph, the number of vertices of odd degree is even. In
particular, a regular graph of odd degree has an even number of vertices.
Suppose you will encounter two graphs with exactly the same structure but
different vertex-sets. For example, look at Figure 1.7 again; the two subgraphs
with vertex-sets {B, D, E} and {C, D, E} are both triangles. There is no
real difference between them until you need to consider the meaning of the
vertices. Formally, we say two graphs G and H are isomorphic if there is
a one-to-one correspondence between their vertex-sets V (G) and V (H) such
that two vertices of G are adjacent if and only if the corresponding vertices
of H are adjacent. For example, any two complete graphs with v vertices are
isomorphic.
Not all graphs with the same number of vertices are isomorphic; for ex-
ample, the graph G of Figure 1.7 and its complement are quite different, but
both have five vertices.

1.5 Systems of Distinct Representatives

An important topic when discussing collections of subsets of size greater
than 2 is the existence of systems of distinct representatives. If D = {B1 , B2 ,
. . . , Bk } are any k sets, we define a system of distinct representatives, or
 Introduction 21

SDR for D to be a way of selecting a member xi from each set Bi such that
x1 , x2 , . . . are all different.
As an example, consider the sets

{124, 124, 134, 235, 246, 1256}.

One system of distinct representatives for them is

1, 2, 3, 5, 4, 6

(where the representatives are listed in the same order as the sets). There are
several others. On the other hand, the sets

{124, 124, 134, 23, 24, 1256}

have no SDR.
If the collection of sets D is to have an SDR, it is clearly necessary that
∪B∈D B have at least as many elements as there are sets in D. The example
above shows that this is not sufficient.

THEOREM 1.2 (Philip Hall’s Theorem) [49]

A collection D of sets has a system of distinct representatives if and only if
it never occurs that some n sets contain between them fewer than n elements.

Proof. We proceed by induction on the number of sets. If D consists of one

set, the result is obvious. Assume the theorem to be true for all collections
of fewer than b sets. Suppose D has k sets B1 , B2 , . . . , Bk and between them
they have v elements {1, 2, . . . , v}; and suppose D satisfies the hypothesis that
the union of any n sets has size at least n, for 1 ≤ n ≤ k. By induction, any n
of the sets will have an SDR, provided that n < k. We distinguish two cases.
(i) Suppose no n sets contain between them fewer than n + 1 elements,
for n < k. Select any element x1 ∈ B1 , and write Bi∗ = Bi \{xi }, for i ∈
{1, 2, . . . , k}. Then the union of any n of the Bi∗ has at least n elements, for
n = 1, 2, . . . , k−1. By the induction hypothesis, there is an SDR x2 , x3 , . . . , xk
for B2 , B3 , . . . , Bk , so x1 , x2 , x3 , . . . , xk is an SDR for the original design D.
(ii) Suppose there is a collection of n of the sets whose union has precisely
n elements, for some n less than k. Without loss of generality, take these
sets as B1 , B2 , . . . , Bn . For i > n, write Bi∗ to mean Bi with all members
of B1 ∪ B2 ∪ . . . ∪ Bn deleted. It is easy to see that the design with sets
Bn+1∗ ∗
, Bn+2 , . . . , Bk∗ satisfies the conditions of the theorem (if Bn+i ∗
1
∗
, Bn+i2
,
. . . , Bn+it were t sets whose union has fewer than t elements, then B1 , B2 ,
∗
∗
. . . , Bn , Bn+i 1
∗
, Bn+i 2
∗
, . . . , Bn+it
would be n + t sets of D whose union has
fewer than n+t elements, which is impossible). From the induction hypothesis
both sets have SDRs, and clearly they are disjoint, so together they comprise
an SDR for D. !
22 Introduction to Combinatorics

Exercises 1A
1. The seven wargamers decide to play a new game, where players compete
in groups of three. Show that they can find an equitable schedule of
seven games.
2. In each case, represent the set in a Venn diagram.

(i) R ∪ S ∪ T (ii) R ∪ S ∪ T

3. Show that the Hat Game team has at least a 75% chance of winning,
for any number of players greater than 3.
4. Find a formula for the sum of the first n odd positive integers,

1 + 3 + . . . + (2n − 3) + (2n − 1)

(i) using Gauss’s formula for the sum of the first n positive integers;
(ii) using induction.
5. Prove by induction that
n
!
i(i + 1) = 13 n(n + 1)(n + 2).
i=1

6. A computer store has four refurbished Gateway computer systems, three

refurbished Dell systems, and seven refurbished Acer systems. In how
many ways can we purchase a system from this store?
7. A computer store has four kinds of towers, three kinds of monitor, and
seven models of printer. In how many ways can we take home a system
consisting of one tower, one monitor, and one printer?
8. In order to gain a high-paying career in the exciting field of basketweav-
ing, we wish to take a class in Underwater Basketweaving and a class
in Mountaintop Basketweaving. The local junior college offers three
sections of each course. The nearby university offers six sections of Un-
derwater and four of Mountaintop. A somewhat reputable online school
has ten sections of Underwater and two of Mountaintop. Assuming that
we will take both courses from the same school and there are no other
conflicts, how many ways can we do this?
9. We have five movies on DVD and we want to watch three of them
tonight. How many ordered sequences of three of the five movies are
possible?
 Introduction 23

10. A state has license plates consisting of two letters and four digits. How
many license plates are possible?
11. Suppose a set X has 2n + 1 elements. How many subsets of X have n
or fewer elements?
12. What are the degrees of the vertices in these graphs?

(i) (ii) (iii)

13. A graph has 48 edges. There are no isolated vertices. What are the
minimum and maximum numbers of vertices the graph can have?
14. How many edges does the star K1,n have? What are the degrees of its
vertices?
15. Eight people attend a meeting. At the beginning various people shake
hands. In total there are eighteen handshakes.
(i) Show that there is at least one person who shook hands at least
five times.
(ii) Is it possible there were exactly three people who shook hands an
odd number of times?
16. Do there exist SDRs for the following sets?
(i) 12, 145, 12, 123
(ii) 123, 145, 12, 13, 23.

Exercises 1B
1. The seven wargamers have become more ambitious. They wish to play
a tournament as follows: each week two groups of three (with no one in
common) play; the seventh member acts as host. Find a schedule for
them to do this in seven weeks.
2. In each case, represent the set in a Venn diagram.