Arnolds Cat Map

Gabriel Peterson
Math 45 Linear Algebra
Fall 1997

Abstract
The purpose of this paper is to discuss and explore the properties of Arnolds cat
map, a chaotic mapping of the pixels of an image. By way of this, the introductory
principles of the exciting new science of chaos are addressed.

Introduction: Chaos
Before undertaking a discussion of Arnolds cat map, a working understanding of chaos must
first be established. We enter into this discussion already in the possession of some intuitive sense
regarding chaos, some nebulous sense that it involves disorder, randomness, disorganization,
entropy... Who among us has not observed that, over time, systems tend to become disorganized or
chaotic. Does a bedroom not evolve from an initial state of order to disorder with indifferent
cleaning? A jar containing marbles of two different colors, initially group by color, promptly
distributes the marbles randomly when vigorously shaken. Indeed, the inviolable and sacrosanct
Second Law of Thermodynamics tells as much: the entropy, the measure of disorder, of the
Universe increases over time.
However, the chaos that concerns this discussion, while inextricably tied to the above, is that of
the fledgling science of chaos. This science has leapt into the public consciousness in recent years
with popularization in motion pictures, such as Jurassic Park, and books, such as James Gleicks
Chaos: Making a New Science. Chaos evolved from the work of Edward Lorenz, a meteorologist
at Massachusetts Institute of Technology. In 1960, equipped with a newly purchased electronic
computer, he created a simulation of the weather using a simple system of equations; the machine
had neither the computational power nor the memory for a more sophisticated model. Initial
conditions were entered, and the weather in this electronic world would unfold, its binary denizens
enduring downpours, blizzards, droughts, and other meteorological adversities. Wanting to examine
a particular run in greater detail, Lorenz reenter its initial conditions. However, the data from the
first run had been entered to six decimal places of accuracy, whereas the second only to three. His
discovery was startling: comparing the second run to the first which should have been virtually
identical he found that the second rapidly diverged from the first. Surely, the small error
introduced by truncating the initial conditions to three decimal places could not have induced this
troubling behavior. But it did. Complex systems even a simplified complex system such as
Lorenzs are extremely sensitive to initial conditions, so-called sensitivity dependence upon initial
conditions. Small errors rapidly propagate through such systems. Lorenz called this the Butterfly
Effect: the beating of a butterflys wings in China could cause a blizzard in Chicago.
It was not until the publishing of the paper mischievously titled Period Three Implies Chaos in
1975 by James Yorke and Tien-Yien Li that the word chaos was coined mathematically to give these
ideas a name. Broadly defined, chaos pertains to those mathematical or physical phenomena that are

apparently random yet possess underlying order. As James Gleick commented,

Over the last decade, physicists, biologists, astronomers, and economists have
created a new way of understanding the growth of complexity in nature. This new
science, called chaos, offers a way if seeing order and pattern where formerly only
the random, erratic, the unpredictable in short, the chaotic had been observed
(Chaos and Fractals, vii).
In a now famous series of debates with Niels Bohr concerning quantum physics, Albert Einstien
proclaimed God does not play dice with the Universe, the Universe is not governed by chance.
However, chaos implies, in the words of Joseph Ford of the Georgia Institute of Technology,
God does play dice with the Universe. But they are loaded dice. And the main
objective of physics now is to find out by what rules were they loaded and how can
we use them for our own end (Chaos, 314).

Introduction: Images
As matter is composed of discrete units called atoms (which are themselves composed of
discrete units), so too images are composed of discrete units called pixels. A pixel is a small square
representing some color value, which when taken together form the mosaic that is the image. The
image is a m x n matrix, where m represents the number of rows of pixels and n the number of
columns of pixels, with each entry in the matrix being a numeric value that represents a given color.
For example, consider the 175 x 175 image of a caffeine molecule below.

20

40

60

80

100

120

140

160

20

40

60

80

100

120

140

160

Let the image be the matrix X, and we can examine selected entries in X.

217 217 217 217 ...

217 217 217 217

251 251 251 251 ...

251 251 251 251

251 251 251 251 ...

251 251 251 251

251 251 251 251 ...

251 251 251 251

251 251 251 251 ...

251 251 251 251

251 251 251 251 ...

251 251 251 251

251 251 251 251 ...

251 251 251 251

217 217 217 217 ...

217 217 217 217

The numeric entries merely represent some color value.

Arnolds Cat Map

The particular instance of chaos we will explore in this discussion is the chaotic mapping called
Arnolds cat map in recognition of Russian mathematician Vladimir I. Arnold, who discovered it
using an image of a cat. It is a simple and elegant demonstration and illustration of some of the
principles of chaos namely, underlying order to an apparently random evolution of a system. An
image (not necessarily a cat) is hit with a transformation that apparently randomizes the original
organization of its pixels. However, if iterated enough times, as though by magic, the original image
reappears.
If we let X

x
y

be a n x n matrix of some image, Arnolds cat map is the transformation

where mod is the modulo of the

150 mod 100 50 or

123

x
y

xy
x 2y
mod 100

xy
x 2y

mod n

and n . For example, 3.142 mod 1 .142 or

23

. Since the signs of both arguments are the

154
54
same sign in this exercise, the modulo will simply be the remainder of the long division of
xy
and n.
x 2y

To better understand the mechanism of the transformation , let us decompose it into its
elemental pieces.

1. Shear in the x-diection by a factor of 1.

x
y

xy
y

2. Shear in the y-direction by a factor of 1.

x
y

x
xy

3. Evaluate the modulo.

x
y

x
y

mod n

Included below is a visual aide illustrating these steps. The first step shows the shearing in the xand y-directions, followed by evaluation of the modulo and reassembly of the image.

The Experiment
Having developed the mathematical tools to address the heart of this discussion, an experiment
may be performed to determine if, indeed, there exists any order underlying or otherwise to the
chaotic mapping of Arnolds cat map. For this purpose, a program was developed for MATLAB by
the author (see accompanying program catbox.m), and the following 124 x 124 image of the
Earth was iterated with the transformation .

Original

Iteration 1

Iteration 2

Iteration 3

Iteration 4

Iteration 5

Iteration 6

Iteration 7

Iteration 8

Iteration 9

Iteration 10

Iteration 11

Iteration 12

Iteration 13

Iteration 14

Final

The pixels rapidly degenerate into a television-static of chaos by iteration number five, with some
unintelligible order prominent in a number of iterations prior to the original image reappearing on
the fifteenth iteration. Therefore, the image is said to have a period of fifteen. That periodicity
should be observed in a system such as this is quite extraordinary. Its not unlike a coherent image
spontaneously leaping out of the salt-and-pepper of television static, a cogent message spoken from
the howl of radio static loaded dice, indeed!
Experimentally, no elegant model could be developed for the relationship between the period of
an image and n, its number of rows or columns. In general, it may be claimed that as the value of n
increases, the period tends to increases. However, this is not always true. For example, a
101 x 101 image has a period of twenty-five; whereas, a 124 x 124 image, as we just learned, has a
period of fifteen. Other luminaries have found a relationship where this experimenter failed but it
certainly cannot be claimed to be elegant nor robust . Let the period be .
1. n 3n if and only if n 2 5 k for k 1, 2, ...
2. n 2n if and only if n 5 k or n 6 5 k for k 1, 2, ...
3. n t

12n
7

for all other choices of n

Let us address that ageless question of Mankind: Why? Why does order emerge out of this
apparently chaotic mapping? The best approach to answer this question is, perhaps, to examine the
behavior of a single pixel, to determine what effect Arnolds cat map has upon it. Consider the
52
ordinary pixel
of the 124 x 124 image considered previously. It takes the following path:
13

32
13

67

64

65

57

32
13

121

73

54

13
2 32
v

45

mod 124

81
30
51
105

58

111

17
v

103

37
4
21

25
46

16
53
v

69
122

71
117

32
13

After fifteen iterations, the pixel as would any other pixel in the image has returned to its initial
position, and it would continue eternally along this circle if iterated accordingly. This agrees with
the earlier observation that the 124 x 124 image has a period of fifteen.

Conclusions
In closing, it is important to consider the question that invariably and naturally occurs: What is
the utility, what are the practical applications of Arnolds cat map? Taken by itself, the utility is
limited. However, if taken as a small piece in the whole gestalt of chaos; as an illustration of the

power of this new science; as a glimpse into the machinery of the Universe; or as evidence that God
does play dice with the Universe, but they are loaded dice the utility is immense. Chaos gives us
the tools to approach mathematical and physical phenomena namely, complex, chaotic systems
with confidence or at least competence where before we were at a loss. Examples of chaos in
Nature abound: the beating of a heart, the eccentricity of the planet Plutos orbit, the Great Red Spot
of Jupiter, the turbulence of fluids, the capricious weather, the swinging of a pendulum, the
formation of a snowflake... Chaos brings us another step closer in our long journey to understanding
the Universe we live in.

Acknowledgments
In writing this, the following texts were invaluable: Linear Algebra and Applications by
Howard Anton; Chaos and Fractals by Heinz-Otto Peitgen, Hartmut Jurgens, and Dietmar Saupe;
and Chaos: Making a New Science by James Gleick. For anyone interested in an introduction the
chaos, the latter two texts are highly recommended. I would like to acknowledge Leon Poon of the
University of Maryland for generously allowing me to make use of the graphic illustrating the
different steps involved in one iteration of Arnolds cat map; Dr. David Mills of College of the
Redwoods for guiding me to this project; and, of course, David Arnold of College of the Redwoods,
without whose patient assistance and guidance while developing the companion program to this
paper (catbox.m), this project would not have been possible.

References
Anton, Howard, Linear Algebra and Applications 7th Edition. New York, New York: John
Wiley and Sons, 1994.
Gleick, James, Chaos: Making a New Science. New York, New York: Penguin Books, 1988.
Peitgen, Heinz-Otto, Jurgens, Hartmut, Saupe, Dietmar, Chaos and Fractals. New York, New
York: Springer-Verlag, 1992.