See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/355658928

An Introduction to Chaotic Dynamical Systems

Book · October 2021

DOI: 10.1201/9780429280801

CITATIONS READS

366 3,519

1 author:

Robert L. Devaney
Boston University
154 PUBLICATIONS   3,898 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Complex Dynamics of Perturbed Polynomials View project

All content following this page was uploaded by Robert L. Devaney on 07 February 2022.

The user has requested enhancement of the downloaded file.

An Introduction to
Chaotic Dynamical
Systems
An Introduction to
Chaotic Dynamical
Systems
3rd Edition

Robert L. Devaney
Third edition published 2022
by CRC Press
6000 Broken Sound Parkway NW, Suite 300, Boca Raton, FL 33487-2742

and by CRC Press

2 Park Square, Milton Park, Abingdon, Oxon, OX14 4RN

© 2022 Taylor & Francis Group, LLC

First edition published by Benjamin-Cummings Publishing Co. 1986

Second edition published by CRC Press 2003

CRC Press is an imprint of Taylor & Francis Group, LLC

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 US Copyright Law, no part of this book may be reprinted, reproduced,
transmitted, 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, access www.copyright.com
or contact the Copyright Clearance Center, Inc. (CCC), 222 Rosewood Drive, Danvers, MA 01923,
978-750-8400. For works that are not available on CCC please contact mpkbookspermissions@tandf.
co.uk

Trademark notice: Product or corporate names may be trademarks or registered trademarks and are
used only for identification and explanation without intent to infringe.

ISBN: 978-1-032-15046-8 (hbk)

ISBN: 978-0-367-23615-1 (pbk)
ISBN: 978-0-429-28080-1 (ebk)

DOI: 10.1201/9780429280801

Publisher’s note: This book has been prepared from camera-ready copy provided by the authors.
Contents

Preface to the Third Edition ix

Author xiii

I One Dimensional Dynamics 1

1 A Visual and Historical Tour 3
1.1 Images from Dynamical Systems . . . . . . . . . . . . . . . . 3
1.2 A Brief History of Dynamics . . . . . . . . . . . . . . . . . . 6

2 Examples of Dynamical Systems 19

2.1 Population Models . . . . . . . . . . . . . . . . . . . . . . . . 20
2.2 Newton’s Method . . . . . . . . . . . . . . . . . . . . . . . . 23

3 Elementary Definitions 25
3.1 Orbits . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
3.2 Geometric Views of Orbits . . . . . . . . . . . . . . . . . . . 27

4 Hyperbolicity 33
4.1 Types of Periodic Points . . . . . . . . . . . . . . . . . . . . 33
4.2 A Glimpse of Bifurcations . . . . . . . . . . . . . . . . . . . 37

5 An Example: The Logistic Family 41

5.1 The Simplest Case . . . . . . . . . . . . . . . . . . . . . . . . 41
5.2 The Cantor Set Case . . . . . . . . . . . . . . . . . . . . . . 44

6 Symbolic Dynamics 49
6.1 The Sequence Space . . . . . . . . . . . . . . . . . . . . . . . 49
6.2 The Shift Map . . . . . . . . . . . . . . . . . . . . . . . . . . 50

7 Topological Conjugacy 53
7.1 The Itinerary Map . . . . . . . . . . . . . . . . . . . . . . . . 53
7.2 Conjugacy . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55

8 Chaos 61

9 Structural Stability 67

v
vi Contents

10 Sharkovsky’s Theorem 75

11 The Schwarzian Derivative 85

12 Bifurcations 97
12.1 Examples of Bifurcations . . . . . . . . . . . . . . . . . . . . 97
12.2 General Bifurcation Theorems . . . . . . . . . . . . . . . . . 102

13 Another View of Period Three 111

13.1 Subshifts of Finite Type . . . . . . . . . . . . . . . . . . . . . 111
13.2 The Period 3 Case . . . . . . . . . . . . . . . . . . . . . . . . 113

14 The Period-Doubling Route to Chaos 121

14.1 Renormalization . . . . . . . . . . . . . . . . . . . . . . . . . 121
14.2 The Orbit Diagram . . . . . . . . . . . . . . . . . . . . . . . 125

15 Homoclinic Points and Bifurcations 135

15.1 Homoclinic Points . . . . . . . . . . . . . . . . . . . . . . . . 135
15.2 Homoclinic Bifurcations . . . . . . . . . . . . . . . . . . . . . 138

16 Maps of the Circle 143

16.1 Rotation Numbers . . . . . . . . . . . . . . . . . . . . . . . . 144
16.2 The Standard Family . . . . . . . . . . . . . . . . . . . . . . 149

17 Morse-Smale Diffeomorphisms 155

II Complex Dynamics 163

18 Quadratic Maps Revisited 165
18.1 The Case c = 0 . . . . . . . . . . . . . . . . . . . . . . . . . 165
18.2 The Case |c| > 2 . . . . . . . . . . . . . . . . . . . . . . . . . 167
18.3 The Case c = –2 . . . . . . . . . . . . . . . . . . . . . . . . . 169

19 Normal Families and Exceptional Points 173

20 Periodic Points 177

20.1 Linearization . . . . . . . . . . . . . . . . . . . . . . . . . . . 177
20.2 Critical Values in the Basins of Attraction . . . . . . . . . . 181

21 Properties of the Julia Set 185

22 The Geometry of the Julia Sets 191

22.1 Quadratic Julia Sets . . . . . . . . . . . . . . . . . . . . . . . 191
22.2 A Julia Set for a Rational Map . . . . . . . . . . . . . . . . . 199
22.3 Fractals . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 202
Contents vii

23 Neutral Periodic Points 211

23.1 Rationally Indifferent Periodic Points . . . . . . . . . . . . . 211
23.2 Irrationally Indifferent Periodic Points . . . . . . . . . . . . . 216

24 The Mandelbrot Set 223

24.1 Connectivity of the Julia Set . . . . . . . . . . . . . . . . . . 223
24.2 The Mandelbrot Set . . . . . . . . . . . . . . . . . . . . . . . 226
24.3 Complex Bifurcations . . . . . . . . . . . . . . . . . . . . . . 231
24.4 Geometry of the Principal Bulbs . . . . . . . . . . . . . . . . 236
24.5 External Rays in the Dynamical Plane . . . . . . . . . . . . 241
24.6 External Rays in the Parameter Plane . . . . . . . . . . . . . 245

25 Rational Maps 253

25.1 Singular Perturbations . . . . . . . . . . . . . . . . . . . . . 253
25.2 Basic Properties . . . . . . . . . . . . . . . . . . . . . . . . . 254
25.3 The Escape Trichotomy . . . . . . . . . . . . . . . . . . . . . 256
25.4 The Special Case n = 2 . . . . . . . . . . . . . . . . . . . . . 262
25.5 Sierpinski Holes . . . . . . . . . . . . . . . . . . . . . . . . . 267

26 The Exponential Family 275

26.1 The Cantor Bouquet Case . . . . . . . . . . . . . . . . . . . 276
26.2 The Julia Set of ez . . . . . . . . . . . . . . . . . . . . . . . . 280
26.3 Indecomposable Continua . . . . . . . . . . . . . . . . . . . . 283

III Higher Dimensional Dynamics 291

27 Dynamics of Linear Maps 293
27.1 Behavior of Linear Maps . . . . . . . . . . . . . . . . . . . . 293
27.2 Stable and Unstable Subspaces . . . . . . . . . . . . . . . . . 297

28 The Smale Horseshoe Map 303

28.1 Symbolic Dynamics . . . . . . . . . . . . . . . . . . . . . . . 306

29 Hyperbolic Toral Automorphisms 313

29.1 Hyperbolic Toral Automorphisms . . . . . . . . . . . . . . . 314
29.2 Markov Partitions . . . . . . . . . . . . . . . . . . . . . . . . 318

30 Attractors 325
30.1 The Solenoid . . . . . . . . . . . . . . . . . . . . . . . . . . . 325
30.2 The Plykin Attractor . . . . . . . . . . . . . . . . . . . . . . 332

31 The Stable and Unstable Manifold Theorem 339

32 Global Results and Hyperbolic Maps 357

viii Contents

33 The Hopf Bifurcation 365

33.1 Planar Bifurcations . . . . . . . . . . . . . . . . . . . . . . . 365
33.2 Normal Forms . . . . . . . . . . . . . . . . . . . . . . . . . . 368
33.3 The Hopf Bifurcation Theorem . . . . . . . . . . . . . . . . . 373

34 The Hénon Map 377

A Mathematical Preliminaries 385

A.1 Preliminaries from Calculus . . . . . . . . . . . . . . . . . . . 385
A.2 Preliminaries from Geometry and Topology . . . . . . . . . . 392
A.3 Preliminaries from Complex Analysis . . . . . . . . . . . . . 394
A.4 Preliminaries from Linear Algebra . . . . . . . . . . . . . . . 400

Bibliography 413

Index 417
Preface to the Third Edition

In the thirty years since the previous edition of this text appeared, many
amazing things have occurred in the field of mathematics known as dynami-
cal systems. Indeed, there has been an explosion of interest in this field in the
mathematical community as well as in many areas of science. Scientists and
engineers have come to realize the power and the beauty of the geometric and
qualitative techniques developed during this period. More importantly, they
have been able to apply these techniques to a number of important nonlin-
ear problems ranging from physics and chemistry to ecology and economics.
The results have been truly exciting: systems that once seemed completely
intractable from an analytic point of view can now be understood in a ge-
ometric or qualitative sense rather easily. Chaotic and random behavior of
solutions of deterministic systems is now understood to be an inherent fea-
ture of many nonlinear systems, and the geometric theory developed over the
past few decades handles this situation quite nicely.
Perhaps the most important development in the past thirty years has been
the widespread availability of computers for use in the study of dynamical
systems. Computer graphics have played an extremely important role here.
Now, rather than finding specific solutions to dynamical systems (which is
almost always impossible), computer graphics have allowed us to view the
dynamical behavior geometrically. This has led to a major new approach to
the study of dynamical systems. In addition, the appearance of incredibly
beautiful and intricate objects such as the Mandelbrot set, the Julia set, and
other fractals have really piqued interest in the field.
There are many different types of mathematical dynamical systems, in-
cluding ordinary differential equations, partial differential equations, ergodic
systems, and discrete dynamical systems. In this book, we will concentrate
only on discrete dynamical systems (basically, iteration of mathematical func-
tions). This approach has the advantage of making many of the ideas from
other types of dynamical systems much more accessible. Indeed, the aim of
this text is to make the techniques in dynamical systems available to ad-
vanced mathematics majors as well as to graduate students and scientists in
other disciplines.
The field of dynamical systems and especially the study of chaotic systems
has been hailed as one of the important breakthroughs in science in this cen-
tury. While the field is still relatively young, there is no question that the field
is becoming more and more important in a variety of scientific disciplines.

ix
x Preface to the Third Edition

We hope that this text serves to excite and to lure many others into this
“dynamic” field.

A NOTE TO THE READER:

This is, first of all, a Mathematics text that is aimed primarily at advanced
undergraduate and beginning graduate students in mathematics. Throughout,
we emphasize the mathematical aspects of the theory of discrete dynamical
systems, not the many and diverse applications of this theory. The text be-
gins at a relatively unsophisticated level in Chapter 1, where we deal with
one-dimensional dynamics (primarily iteration of quadratic functions on the
real line). This part of the text is accessible to students with only a solid
background in freshman calculus. As we proceed, more advanced topics such
as topology, complex analysis, and linear algebra become important. Many of
these prerequisites are presented in the appendix.
The first chapter, one-dimensional dynamics, is by far the longest. It is
the author’s belief that virtually all of the important ideas and techniques of
nonlinear dynamics can be introduced in the setting of the real line or the
circle. This has the obvious advantage of minimizing the topological compli-
cations of the system and the algebraic machinery necessary to handle them.
In particular, the only real prerequisite for this chapter is a good calculus
course. (Oh well, we do multiply a 2 x 2 matrix once or twice in Section 13,
and we use the Implicit Function Theorem in two variables in Section 12, but
these are exceptions.) With only these tools, we manage to introduce such
important topics as chaos, structural stability, topological conjugacy, the shift
map, homoclinic points, and bifurcation theory. To emphasize the point that
chaotic dynamics occur in the simplest of systems, we carry out most of our
analysis in this section on two basic models, the quadratic mappings given
by F µ(x) = µx(l − x) or Qc (x) = x2 + c. These maps have the advantage of
being perhaps the simplest nonlinear maps, yet they illustrate virtually every
concept we wish to introduce. A few topological ideas, such as the notion of
a dense set or a Cantor set, are introduced in detail when needed.
Just to emphasize the contemporary nature of the field of discrete dy-
namical systems, because of the chaotic behavior of these quadratic maps, we
finally understood the total behavior of these maps in the 1990s. Think about
this: mathematicians finally understood x2 + c around thirty years ago! And,
to this day, we still don’t completely understand complex z 2 + c !
In previous editions of this book, the topic of complex dynamics was cov-
ered in the third and final chapters. Given the amazing resurgence of inter-
est in this field due to the spectacular computer graphics images, we have
greatly expanded coverage of this topic and moved it to Chapter 2. We again
Preface to the Third Edition xi

concentrate primarily on quadratic functions in the complex plane, specifically

Qc (z) = z 2 + c. This is the function whose overall behavior is summarized in
the beautiful and intricate image known as the Mandelbrot set. Moving to
the complex plane allows us to view the real quadratic functions in a very
different and much more helpful way. In this chapter, we also introduce some
very different dynamical behavior that arises when complex functions such as
rational maps and the exponential are itereated. This chapter assumes that
students are familiar with the complex plane as well as complex arithmetic
and geometry. Some topics from the elementary complex analysis are used,
and these are presented in the appendix.
The third chapter is devoted to higher dimensional dynamical systems.
With many of the prerequisites already introduced in the first chapter, the
discussion of such higher dimensional maps as Smale’s horseshoe, hyperbolic
toral automorphisms, and the solenoid become especially simple. This chapter
assumes that the reader is familiar with some multi-dimensional calculus as
well as linear algebra, including the notion of eigenvalues and eigenvectors
for 3 x 3 matrices and some additional concepts from topology. All of these
prerequisites are described in the appendix. One of the major differences be-
tween one dimensional and higher dimensional dynamics, the possibility of
both contraction and expansion at the same time, is treated at length in a
section devoted to the proof of the Stable and Unstable Manifold Theorem.
We end the chapter with a lengthy set of exercises all centered on the im-
portant Henon map of the plane. This section serves as a summary of many
of the previous topics in the book as well as a good “final” project for the
reader. Chapters 2 and 3 are independent of one another; they both only as-
sume familiarity with the basic concepts of dynamics as presented in the first
chapter. We also have provided some background mathematical material from
calculus, linear algebra, geometry and topology, and complex analysis for use
in these chapters in the appendix.
There are many themes developed in this book. We have tried to present
several different dynamical concepts in their most elementary formulation in
Chapter 1 and to return to these subjects for further refinement at later stages
in the book. One such topic is bifurcation theory. We introduce the most ele-
mentary bifurcations, the saddle-node, and the period-doubling bifurcations,
early in Chapter l. Later in the same chapter, we treat the accumulation
points of such bifurcations which occur when a homoclinic point develops. In
Chapter 2, we revisit these topics, but from a very different perspective of
the complex plane. We explore the very different bifurcations that occur in
complex dynamics and include a discussion of the global aspects of the saddle-
node bifurcation, and the exploding Julia sets that occur for the exponential
map. In Chapter 3, we return to bifurcation theory to discuss the Hopf bifur-
cation. Another recurrent theme throughout the book is symbolic dynamics.
We think of symbolic dynamics as a tool whereby complicated dynamical sys-
tems are equated with seemingly quite different systems, but these systems
have the advantage that they can be analyzed much more easily. Symbolic
xii Preface to the Third Edition

dynamics appears quite early in Chapter 1 when we first describe the chaotic
behavior of the quadratic map. It is clear that the most elementary setting
for the phenomena associated with the Smale horseshoe mapping occurs in
one dimension, and we fully exploit this idea. Later, symbolic dynamics is
extended to the case of subshifts of finite type via another quadratic example.
And finally, the related concepts of Markov partitions and inverse limits are
introduced in the third chapter.
Examples abound in this text. We often motivate new concepts by working
through them in the setting of a specific dynamical system. In fact, we have
often sacrificed generality in order to concentrate on a specific system or class
of systems. Many of the results throughout the text are stated in a form that
is nowhere near full generality. We feel that the general theory is best left to
more advanced texts, which presuppose much more advanced mathematics.
Much of what many researchers consider dynamical systems has been de-
liberately left out of this text. For example, we do not treat continuous systems
such as differential equations at all. There are several reasons for this. First,
as is well known, computations with specific nonlinear ordinary differential
equations are next to impossible. Secondly, the study of differential equations
necessitates a much higher level of sophistication on the part of the student,
certainly more than that necessary for Chapter 1 of this text. We adopt in-
stead the attitude that any dynamical phenomena that occur in a continuous
system also occurs in a discrete system, and so we might as well make life easy
and study iterated maps first. There are many texts currently available that
treat continuous systems almost exclusively. We hope that this book presents
a solid introduction to the topics treated in these more advanced texts.
Another topic that has been excluded is ergodic theory. It is our feeling
that measure theory would take us too far afield in this book. Of course, it can
be argued that measure theory is no more advanced than the complex anal-
ysis necessary for Chapter 2. However, we feel that the topological approach
adopted throughout this text is inherently easier to understand, at least for
undergraduates in Mathematics. There is no question, however, that ergodic
theory would provide an ideal sequel to the material presented here, as would
a course in nonlinear differential equations.
Finally, I am indebted to Bob Ross at CRC Press for arranging the pub-
lication of the long-overdue third edition of this text.

Robert L. Devaney
Boston University
Boston, MA
April, 2021
Author

Robert L. Devaney is currently a Professor of Mathematics at Boston Uni-

versity. He received his PhD from the University of California at Berkeley un-
der the direction of Stephen Smale. He taught at Northwestern University and
Tufts University before coming to Boston University in 1980. His main area
of research is dynamical systems, primarily complex analytic dynamics, but
also including more general ideas about chaotic dynamical systems. Lately, he
has become intrigued with the incredibly rich topological aspects of dynamics,
including such things as indecomposable continua, Sierpinski curves, and Can-
tor bouquets. He is also the author of A First Course in Chaotic Dynamical
Systems, Second Edition, published by CRC Press.

xiii
 Part I

One Dimensional Dynamics

1
A Visual and Historical Tour

Rather than jump immediately into the mathematics of dynamical systems,

we will begin with a brief tour of some of the amazing computer graphics
images that arise in this field. One of our goals in this book is to explain what
these images mean, how they are generated on the computer, and why they
are important in mathematics. We will do none of the mathematics in this
chapter. For now, you should simply enjoy the images. We hope to convince
you, in the succeeding chapters, that the mathematics behind these images
is even prettier than the pictures. In the second part of this chapter, we will
present a brief history of some of the developments in dynamical systems over
the past century. You will see that many of the ideas in dynamics arose fairly
recently. Indeed, none of the computer graphics images from the tour had
been seen before 1980!

1.1 Images from Dynamical Systems

This book deals with some very interesting, exciting, and beautiful topics
in mathematics—topics which, in many cases, have been discovered only in
the last fifty years. The main subject of the book is dynamical systems, the
branch of mathematics that attempts to understand processes in motion. Such
processes occur in all branches of science. For example, the motion of the
planets and the galaxies in the heavens is a dynamical system, one that has
been studied for centuries by thousands of mathematicians and scientists.
The stock market is another system that changes in time, as does the earth’s
weather. The changes chemicals undergo, the rise and fall of populations, and
the motion of a simple pendulum are classical examples of dynamical systems
in chemistry, biology, and physics. Clearly, dynamical systems abound.
What does a scientist wish to do with a dynamical system? Well, since the
system is moving or changing in time, the scientist would like to predict where
the system is heading, where it will ultimately go. Will the stock market go
up or down? Will it be rainy or sunny tomorrow? Will these two chemicals
explode if they are mixed in a test tube?
Clearly, some dynamical systems are predictable, whereas others are not.
You know that the sun will rise tomorrow and that when you add cream to

DOI: 10.1201/9780429280801-1 3
4 A Visual and Historical Tour

a cup of coffee, the resulting “chemical” reaction will not be an explosion.

On the other hand, predicting the weather a month from now or the Dow
Jones average a week from now seems impossible. You might argue that the
reason for this unpredictability is that there are simply too many variables
present in meteorological or economic systems. That is indeed true in these
cases, but this is by no means the complete answer. One of the remarkable
discoveries of twentieth-century mathematics is that very simple systems, even
systems depending on only one variable, may behave just as unpredictably as
the stock market, just as wildly as a turbulent waterfall, and just as violently
as a hurricane. The culprit, the reason for this unpredictable behavior, has
been called “chaos” by mathematicians.
Because chaos has been found to occur in the simplest of systems, scientists
may now begin to study unpredictability in its most basic form. It is to be
hoped that the study of these simpler systems will eventually allow scientists
to find the key to understanding the turbulent behavior of systems involving
many variables such as weather or economic systems.
In this book we discuss the chaos in these simple settings. We will see
that chaos occurs in elementary mathematical objects—objects as familiar
as quadratic functions—when they are regarded as dynamical systems. You
may feel at this point that you know all there is to know about quadratic
functions—after all, they are easy to evaluate and to graph. You can differ-
entiate and integrate them. But the keywords here are “dynamical systems.”
We will treat simple mathematical operations like taking the square root,
squaring, or cubing as dynamical systems by repeating the procedure over
and over, using the output of the previous operation as the input for the next.
This process is called iteration. This procedure generates a list of real or com-
plex numbers or points in a higher dimensional space that are changing as we
proceed—this is our dynamical system. Sometimes we will find that, when we
input certain numbers into the process, the resulting behavior is completely
predictable, while other inputs yield results that are often bizarre and totally
unpredictable.
For the types of complex functions that we will consider later, the set of
numbers that yield chaotic or unpredictable behavior in the plane is called
the Julia set after the French mathematician Gaston Julia, who first formu-
lated many of the properties of these sets in the 1920s. These Julia sets are
spectacularly complicated, even for quadratic functions. They are examples
of fractals. These are sets which, when magnified over and over again, always
resemble the original image. The closer you look at a fractal, the more you
see exactly the same object. Moreover, fractals naturally have a dimension
that is not an integer, not 1, not 2, but often somewhere in between, such
as dimension 1.4176, whatever that means! We will discuss these concepts in
more detail in Chapter 2.
Here are some examples of the types of images that we will study. In Plate
1 we show the Julia set of the simple mathematical expression z 2 + c, where
both z and c are complex numbers. In this particular case, c = −.122 + .745i.
Images from Dynamical Systems 5

This image is called Douady’s rabbit, after the French mathematician Adrien
Douady whose work we will discuss in Chapter 2. The black region in this
image resembles a “fractal rabbit.” Everywhere you look, you see a pair of
ears. In the accompanying figures, we have magnified portions of the rabbit,
revealing more and more pairs of ears.
As we will describe later, the black points you see in these pictures are
the “non-chaotic” points. They are points representing values of z that, under
iteration of this quadratic function, eventually tend to cycle between three
different points in the plane. As a consequence, the dynamical behavior in the
black regions is quite predictable. All of this is by no means apparent right
now, but by the time you have read Chapter 2, you will consider this example
a good friend. Points that are colored in this picture also behave predictably:
they are points that “escape”; that is, they tend to infinity under iteration.
The colors here simply tell us how quickly a point escapes, i.e., how many
iterations it takes to go beyond a pre-determined bound. Red points escape
fastest, followed in order by orange, yellow, green, blue, and violet points.
The boundary between these two types of behavior—the interface between
the escaping and the cycling points—is the Julia set. This is where we will
encounter all of the chaotic behavior for this dynamical system.
In Plates 2–5, we have displayed Julia sets for other quadratic functions
of the form z 2 + c. Each picture corresponds to a different value of c. For
example, Plate 5a is a picture of the Julia set for z 2 + i. As we see, these
Julia sets may assume a remarkable variety of shapes. Sometimes the images
contain large black regions as in the case of Douady’s rabbit. Other times the
Julia set looks like an isolated scatter of points, as in Plate 5b. Many of these
Julia sets are Cantor sets. These are very complicated sets that arise often in
the study of dynamics. We will begin our study of Cantor sets in Section 5
when we introduce the most basic fractal of all, the Cantor middle-thirds set.
All of the Julia sets in Plates 1–5 correspond to mathematical expressions
that are of the form z 2 + c. As we see, when c varies, these Julia sets change
considerably in shape. How do we understand the totality of all of these shapes,
the collection of all possible Julia sets for quadratic functions? The answer is
called the Mandelbrot set. The Mandelbrot set, as we will see in Section 24, is
a dictionary, or picture book, of all possible quadratic Julia sets. It is a picture
in the c-plane that provides us with a road map of the quadratic Julia sets.
This image, first viewed in 1980 by Benoit Mandelbrot, is quite important in
dynamics. It completely characterizes the Julia sets of quadratic functions.
It has been called one of the most intricate and beautiful objects in all of
mathematics. Amazingly, we still do not completely understand the structure
of this set. That is, we still do not fully understand the dynamics of the simple
quadratic function z 2 + c !
Plate 6 shows the full Mandelbrot set. Note that it consists of a basic
central cardioid shape, with smaller bulbs or decorations attached. Plates 7–
11 are magnifications of some of these decorations. Note how each decoration
differs from the others. Buried deep in various regions of the Mandelbrot set,
6 A Visual and Historical Tour

we also see small black regions which are actually small copies of the entire
set. Look at the “tail” of the Mandelbrot set in Plate 7. The Mandelbrot
set possesses an amazing amount of complexity, as illustrated in Plate 11
and its magnifications. Nonetheless, each of these small regions has a distinct
dynamical meaning, as we will discuss in Section 24.
In this book we will also investigate the chaotic behavior of many other
functions. For example, in Plates 12 and 13 we have displayed the Julia set for
several functions of the form c sin(z) and c cos(z). Plate 15 displays the Julia
sets for certain rational functions of the form z n +c/z n . These sets are what are
known as “Sierpinski curves,” probably the most interesting planar fractals,
as we shall discuss in Section 25. If we investigate exponential functions of
the form c exp(z), we find Julia sets that look quite different, for example,
in Plate 16. And we can also look at the parameter planes (the c-planes) for
these maps, a portion of which is shown in Plate 17.
In addition, we have referenced a number of online videos that are
posted on my website that will allow you to see the dramatic changes these
systems undergo as parameters vary. All of these videos are available at
math.bu.edu/DYSYS/animations.html.
The images in this mathematical tour show quite clearly the great beauty
of mathematical dynamical systems theory. But what do these pictures mean
and why are they important? These are questions that we will address in the
remainder of this book.

1.2 A Brief History of Dynamics

Dynamical systems has a long and distinguished history as a branch of math-
ematics. Beginning with the fundamental work of Isaac Newton, differential
equations became the principal mathematical technique for describing pro-
cesses that evolve continuously in time. In the eighteenth and nineteenth cen-
turies, mathematicians devised numerous techniques for solving differential
equations explicitly. These methods included Laplace transforms, power se-
ries solutions, variation of parameters, linear algebraic methods, and many
other techniques familiar from the basic undergraduate course in ordinary
differential equations.
There was one major flaw in this development. Virtually all of the analytic
techniques for solving differential equations worked mainly for linear differen-
tial equations. Nonlinear differential equations proved much more difficult to
solve. Unfortunately, almost all of the most important processes in nature are
inherently nonlinear.
An example of this is provided by Newton’s original motivation for develop-
ing calculus and differential equations. Newton’s laws enable us to write down
the equations that describe the motion of the planets in the solar system,
A Brief History of Dynamics 7

among many other important physical phenomena. Known as the n-body

problem, these laws give us a system of differential equations whose solutions
describe the motion of n “point masses” moving in space subject only to their
own mutual gravitational attraction. If we know the initial positions and ve-
locities of these masses, then all we have to do is solve Newton’s differential
equation to be able to predict where and how these masses will move in the
future.
This turns out to be a formidable task. If there are only one or two point
masses, then these equations may be solved explicitly, as is often done in a
freshman or sophomore calculus or physics class. For three or more masses, the
problem today remains completely unsolved, despite the efforts of countless
mathematicians over the past three centuries. It is true that numerical solu-
tions of differential equations by computers have allowed us to approximate
the behavior of the actual solutions in many cases, but there are still regimes
in the n-body problem where the solutions are so complicated or chaotic that
they defy even numerical computation.
Although the explicit solution of nonlinear ordinary differential equations
has proved elusive, there have been four landmark events over the past 130
years that have revolutionized the way we study dynamical systems. Per-
haps the most important event occurred in 1890. King Oscar II of Sweden
announced a prize for the first scientist who could solve the n-body prob-
lem and thereby prove the stability of the solar system. Needless to say, no-
body solved the original problem, but the great French mathematician Henri
Poincaré came closest. In a beautiful and far-reaching paper, Poincaré totally
revamped the way we tackle nonlinear ordinary differential equations. Instead
of searching for explicit solutions to these equations, Poincaré advocated work-
ing qualitatively, using topological and geometric techniques, to uncover the
global structure of all solutions. To him, a knowledge of all possible behaviors
of the system under investigation was much more important than the rather
specialized study of individual solutions.
Poincaré’s prize-winning paper contained a major new insight into the
behavior of solutions of differential equations. In describing these solutions,
mathematicians had previously made the tacit assumption that what we now
know as stable and unstable manifolds always match up. Poincaré questioned
this assumption. He worked long and hard to show that this was always the
case, but he could not produce a proof. He eventually concluded that the
stable and unstable manifolds might not match up and could actually cross
at an angle. When he finally admitted this possibility, Poincaré saw that this
would cause solutions to behave in a much more complicated fashion than
anyone had previously imagined. Poincaré had discovered what we now call
chaos. Years later, after many attempts to understand the chaotic behavior
of differential equations, he threw up his hands in defeat and wondered if
anyone would ever understand the complexity he was finding. Thus, “chaos
theory,” as it is now called, really dates back over 130 years to the work of
Henri Poincaré.
8 A Visual and Historical Tour

Poincaré’s achievements in mathematics went well beyond the field of dy-

namical systems. His advocacy of topological and geometric techniques opened
up whole new subjects in mathematics. In fact, building on his ideas, math-
ematicians turned their attention away from dynamical systems and toward
these related fields in the ensuing decades. Areas of mathematics such as al-
gebraic and differential topology were born and flourished in the twentieth
century. But nobody could handle the chaotic behavior that Poincaré had
observed, so the study of dynamics languished.
There were two notable exceptions to this. One was the work of the French
mathematicians Pierre Fatou and Gaston Julia in the 1920s on the dynamics
of complex analytic maps. They too saw chaotic behavior, this time on what
we now call the Julia set. Indeed, they realized how tremendously intricate
these Julia sets could be, but they had no computer graphics available to see
these sets, and as a consequence, this work also stopped in the 1930s.
A little later, the American mathematician G. D. Birkhoff adopted
Poincaré’s qualitative point of view on dynamics. He advocated the study of
iterative processes as a simpler way of understanding the dynamical behavior
of differential equations, a viewpoint that we will adopt in this book.
The second major development in dynamical systems occurred in the
1960s. The American mathematician Stephen Smale reconsidered Poincaré’s
crossing stable and unstable manifolds from the point of view of iteration and
showed by an example now called the “Smale horseshoe” (which we will cover
in detail in Section 28) that the chaotic behavior that baffled his predecessors
could indeed be understood and analyzed completely. The technique he used
to analyze this is called symbolic dynamics and will be a major tool for us in
this book. At the same time, the American meteorologist E. N. Lorenz, using a
very crude computer, discovered that very simple differential equations could
exhibit the type of chaos that Poincaré observed. Lorenz, who actually had
been a Ph.D. student of Birkhoff, went on to observe that his simple meteoro-
logical model (now called the Lorenz system) exhibited what is called sensitive
dependence on initial conditions. Roughly speaking, this means that a very
small change in initial conditions in the system could lead to vastly different
outcomes. Lorenz characterized this by saying that the flap of a butterfly’s
wings in Brazil could possibly set off a tornado in Texas. For him, sensitive
dependence meant that long-range weather forecasting was all but impossible
and showed that the mathematical topic of chaos was important in all other
areas of science.
Then the third major development occurred in 1975 when T.Y. Li and
James Yorke published an amazing paper called Period Three Implies Chaos
[35]. In this paper, they showed that, if a simple continuous function on the
real line has a point which cycles with period three under iteration, then
this function must also have cycles of all other periods. Moreover, they also
showed that this function must behave chaotically on some subset of the line.
Perhaps most importantly, this paper was essentially the first time the word
“chaos” was used in the scientific literature, and this motivated a huge number
A Brief History of Dynamics 9

of mathematicians, scientists, and engineers to begin investigating this phe-

nomenon.
Curiously, the Li-Yorke result was preceded by a research paper that had
much more substantial results. In a 1964 paper [47], Oleksandr Sharkovsky
determined that, if such a map on the real line had a cycle of period n,
then he could list exactly all of the other periods that such a map must
have. Unfortunately, this paper was published in Ukranian and hence was not
known at all in the west. However, after the Li-Yorke paper appeared, this
result became one of the most interesting results in modern dynamical systems
theory. All of these results are described in Section 10.
These advances led to a tremendous flurry of activity in nonlinear dynamics
in all areas of science and engineering in the ensuing decade. For example, the
ecologist Robert May found that very simple iterative processes that arise in
mathematical biology could produce incredibly complex and chaotic behavior.
The physicist Mitchell Feigenbaum, building on Smale’s earlier work, noticed
that, despite the complexity of chaotic behavior, there was some semblance
of order in the way systems became chaotic. Physicists Harry Swinney and
Jerry Gollub showed that these mathematical developments could actually be
observed in physical applications, notably in turbulent fluid flow. Later, other
systems, such as the motion of the former planet Pluto or the beat of the
human heart, have been shown to exhibit similar chaotic patterns. In mathe-
matics, meanwhile, new techniques were developed to help understand chaos.
John Guckenheimer and Robert F. Williams employed the theory of strange
attractors to explain the phenomenon that Lorenz had observed a decade
earlier. And tools such as the Schwarzian derivative, symbolic dynamics, and
bifurcation theory—all topics we will discuss in this book—were shown to play
an important role in understanding the behavior of dynamical systems.
The fourth and most recent major development in dynamical systems was
the availability of high-speed computing and, in particular, computer graphics.
Foremost among the computer-generated results was Mandelbrot’s discovery
in 1980 of what is now called the Mandelbrot set. This beautiful image im-
mediately reawakened interest in the old work of Julia and Fatou. Using the
computer images as a guide, mathematicians such as Bodil Branner, Adrien
Douady, John Hubbard, Dennis Sullivan as well as Fields Medalists John Mil-
nor, Bill Thurston, Curt McMullen, Artur Avila, and Jean-Christophe Yoccoz
jumped into the field and greatly advanced the classical theory. Other com-
puter graphics images such as the orbit diagram and the Lorenz attractor
also generated considerable interest among mathematicians and led to further
advances.
One of the most interesting side effects of the availability of high-speed
computing and computer graphics has been the development of an experimen-
tal component in the study of dynamical systems. Whereas the old masters in
the field had to rely solely on their imagination and their intellect, now math-
ematicians have an invaluable additional resource to investigate dynamics: the
10 A Visual and Historical Tour

computer. This tool has opened up whole new vistas for dynamicists, some of
which we will sample in this book.
Finally, as mentioned earlier, even the iteration of simple functions on the
real line can behave quite chaotically. Indeed, for years the behavior of simple
quadratic functions such as x2 + c or kx(1 − x) when iterated was not com-
pletely understood. It was only with the appearance of the Mandelbrot set as
well as with techniques from complex analysis that we finally understood this
behavior in the 1990s. Think about it: We finally understood the iteration of
quadratic functions on the real line in the 1990s! That being said, the behav-
ior of cubic functions (and higher degree polynomials) is still not understood!
Clearly, there are lots of open problems in the area of dynamical systems,
many of which you will begin to see in this book. Please do solve several of
these major open problems, and enjoy your Fields Medal!
A Brief History of Dynamics 11

Plate 1: Douady’s rabbit Julia set and several magnifications.

Plate 2: Dancing rabbits Julia set.

12 A Visual and Historical Tour

Plate 3: A dragon and the basilica Julia set.

Plate 4: Three and four-eared rabbit Julia sets.

(a) (b)
Plate 5: A dendrite and a Cantor set Julia set.
A Brief History of Dynamics 13

Plate 6: The Mandelbrot set.

Plate 7: The tail of the Mandelbrot set.

14 A Visual and Historical Tour

Plate 8: A period 3 and 7 bulb in the Mandelbrot set.

Plate 9: A period 17 bulb and a magnification.

Plate 10: Another period 17 bulb and a magnification.

A Brief History of Dynamics 15

Plate 11: Another region in the Mandelbrot set.

16 A Visual and Historical Tour

Plate 12: Julia sets of (1 + 0.2i) sin(z) and (.61 + .81i) sin(z).

Plate 13: The Julia set of 2.95 cos(z).

 A Brief History of Dynamics 17

Plate 14: Complex Newton’s Method for z 3 − 1.

Plate 15: Sierpinski curve Julia sets for z 3 + (−0.25 + .03i)/z 3 and z 2 − .004/z 2.
18 A Visual and Historical Tour

Plate 16: The Julia sets of 0.36 exp(z) and 0.38 exp(z) exhibiting a
major “bifurcation.”

Plate 17: The parameter plane for λ exp(z).

References
Ahlfors, L. Complex Analysis: An Introduction to the Theory of Analytic Functions of One
Complex Variable. McGraw-Hill, 1979.
Barnsley, M. Fractals Everywhere, Academic Press, 1988.
Banks, J. et. al. On Devaney's Definition of Chaos. Amer. Math. Monthly. 99 (1992), 332–334.
Blanchard, P. , Devaney, R. L. , and Hall, G. R. Differential Equations. Cengage Learning, 2011.
Block, L. , Guckenheimer, J. , Misiurewicz, M. , and Young, L. S. Periodic Points and
Topological Entropy of One Dimensional Maps. In Global Theory of Dynamical Systems.
Springer 819 (1980), 18–34.
Branner, B. and Fagella, N. Quasiconformal Surgery in Holomorphic Dynamics. Cambridge
University Press, 2014.
Branner, B. and Hubbard, J. The Iteration of Cubic Polynomials. Acta. Math. 66 (1988),
143–206.
Brjuno, A. D. Convergence of Transformations of Differential Equations to Normal Forms. Dokl.
Akad. Nauk. USSR 165 (1965), 987–989.
Conway, J. B. Functions of One Complex Variable I. Springer-Verlag, 1995.
Curry, S. One-dimensional Nonseparating Plane Continua with Disjoint ε-dense Subcontinua.
Topology and its Applications 39 (1991), 145–151.
Devaney, R. L. Singular Perturbations of Complex Polynomials. Bulletin of the AMS 50 (2013),
391–429.
Devaney, R. L. The McMullen Domain: Satellite Mandelbrot Sets and Sierpinski Holes.
Conformal Geometry and Dynamics 11 (2007), 164–190.
Devaney, R. L. and Garijo, A. Julia Sets Converging to the Unit Disk. Proceedings of the AMS
136 (2008), 981–988.
Devaney, R. L. and Goldberg, L. Uniformization of Attracting Basins for Exponential Maps. Duke
Math. J. 55 (1999), 269–287.
Devaney, R. L. and Jarque, X. Indecomposable Continua in Exponential Dynamics. Conformal
Geometry and Dynamics 6 2002, 1–12.
Devaney, Killer . Dynamical Systems for Puppies. Advances in Puppy Math 123 (2099),
1–8000.
Devaney, R. L. and Krych, M. Dynamics of Exp(z). Ergodic Theory and Dynamical Systems 4
(1984), 35–52.
Devaney, R. L. , Look, D. M. , and Uminsky, D. The Escape Trichotomy for Singularly Perturbed
Rational Maps. Indiana Univ. Math. J. 54 (2005), 1621–1634.
Devaney, R. L. and Moreno-Rocha, M. Geometry of the Antennas in the Mandelbrot Set.
Fractals 10 (2002), 39–44.
Douady, A. and Hubbard, J. On the Dynamics of Polynomial-like Maps. Ann. Scient. Ec. Norm.
Sup. 18 (1985), 287.
Douady, A. and Hubbard, J. Itération des Polynomes Quadratiques Complexes. C. R. Acad.
Paris 294 (1982), 123–126.
Douady, A. and Hubbard, J. A Proof of Thurston's Topological Characterization of Rational
Functions. Acta Math. 171 (1993), 263–297.
Eremenko, A. and Lyubich, M. Iterates of Entire Functions. Soviet Math. Dokl. 30 (1984),
592–594.
Fatou, P. Sur l'itération des fonctions transcendantes entières. Acta Math. 47 (1926), 337–370.
Feigenbaum, M. Universality in Complex Discrete Dynamics. Los Alamos Theoretical Division
Annual Report 1975-76.
Gamelin, T. Complex Analysis. Springer, 2001.
Goldberg, L. and Keen, L. A Finiteness Theorem for a Dynamical Class of Entire Functions.
Ergodic Theory and Dynamical Systems 6 (1986), 183–192.
Guckenheimer, J. and Holmes, P. Nonlinear Oscillations, Dynamical Systems, and Bifurcations
of Vector Fields. Springer-Verlag, 1983.
Hirsch, M. W. , Smale, S. , and Devaney, R. L. Differential Equations, Dynamical Systems, and
an Introduction to Chaos. Elsevier Academic Press, 2013.
Julia, G. Mémoire sur l'itération des fonctions rationelles . J. Math. 8 (1918), 47–245.
Karpinska, B. Hausdorff Dimension of Hairs Without Endpoints for λ e z . C. R. Acad. Sci. 328
(1999), 1039–1044.
Karpinska, B. Area and Hausdorff Dimension of the Set of Accessible Points in the Julia Set of λ
e z and λ sin ( z ) . Fund. Math. 159, 269–287.
Krasinkiewicz, J. On Homeomorphisms of Sierpinski Curves. Prace Mat. (Comment Math.) 12
(1969), 255–257.
Kuratowski, C. Theorie des Continus Irreducibles Entre Deux Points I. Fund. Math. 3 (1922),
200.
Li, T.-Y. and Yorke, J. Period Three Implies Chaos. American Mathematical Monthly 82 (1975),
985–992.
Mandelbrot, B. The Fractal Geometry of Nature. W. H. Freeman, 1983.
Mayer, J. An Explosion Point for the Set of Endpoints of the Julia set of λ exp ( z ) . Ergodic
Theory and Dynamical Systems 10 (1990), 177–184.
May, R. M. Theoretical Ecology: Principles and Applications. Blackwell, 1981.
McMullen, C. Automorphisms of Rational Maps. In Holomorphic Functions and Moduli. Springer
(1988), 31–60.
McMullen, C. Complex Dynamics and Renormalization. Princeton University Press, 1994.
McMullen, C. Area and Hausdorff Dimension of Julia Sets of Entire Maps. Trans. AMS 300
(1987), 329–342.
Milnor, J. Dynamics in One Complex Variable. Princeton University Press, 2006.
Milnor, J. Periodic Points, External Rays, and the Mandelbrot Set: An Expository Account. In
Géométrie Complexe et Systèmes Dynamiques, Colloque en l'honneur d'Adrien Douady
Astérisque 261 (2000), 277–333.
Misiurewicz, M. On Iterates of e z . Ergodic Theory and Dynamical Systems. 1 (1981), 103–106.
Robinson, C. Dynamical Systems: Stability, Symbolic Dynamics, and Chaos. CRC Press, 1995.
Roesch, P. On Capture Zones for the Family z 2 + λ / z 2 . In Dynamics on the Riemann Sphere
, Eur. Math. Soc. (2006), 121–129.
Sharkovsky, O. M. Co-Existence of Cycles of a Continuous Mapping of a Line onto Itself.
Ukranian Math. Z. 16 (1964), 61–71.
Shishikura, M. The Hausdorff Dimension of the Boundary of the Mandelbrot Set and Julia Sets.
Annals of Math. 147 (1998), 225–267.
Siegel, C. L. Iteration of Analytic Functions. Annals of Math. 43 (1942), 607–612.
Siegel, C. L. and Moser, J. Lectures on Celestial Mechanics. Springer-Verlag, 1971.
Singer, D. Stable Orbits and Bifurcations of Maps of the Interval. SIAM J. Appl. Math 35 (1978),
260–267.
Smale, S. Diffeomorphisms with many periodic points. In Differential and Combinatorial
Topology. Princeton University Press (1965), 63.
Steinmetz, N. Rational Iteration. Berlin: de Gruyter, 1993.
Strogatz, S. Nonlinear Dynamics and Chaos. CRC Press, 2019.
Sullivan, D. Conformal Dynamical Systems. In Geometric Dynamics. Springer-Verlag 1007
(1983), 725–752.
Sutherland, S. Bad Polynomials for Newton's Method. In Linear and Complex Analysis.
Springer-Verlag (1994).
Whyburn, G. T. Topological Characterization of the Sierpinski Curve. Fundamenta Math. 45
(1958), 320–324.
Yoccoz, J.-C. Linéarisation des Germes de Difféomorphismes Holomorphes. C. R. Acad. Sci.
Paris 306 (1988), 55–58.

View publication stats