Dynamical Systems and Geometric

Mechanics An Introduction Jared

Michael Maruskin
Visit to download the full and correct content document:
https://textbookfull.com/product/dynamical-systems-and-geometric-mechanics-an-intr
oduction-jared-michael-maruskin/
More products digital (pdf, epub, mobi) instant
download maybe you interests ...

Geometric Group Theory An Introduction draft Clara Löh

https://textbookfull.com/product/geometric-group-theory-an-
introduction-draft-clara-loh/

Geometric Group Theory An Introduction 1st Edition

Clara Löh

https://textbookfull.com/product/geometric-group-theory-an-
introduction-1st-edition-clara-loh/

Clinical Neuroendocrinology An Introduction Michael

Wilkinson

https://textbookfull.com/product/clinical-neuroendocrinology-an-
introduction-michael-wilkinson/

Microstates Entropy and Quanta An Introduction to

Statistical Mechanics Don Koks

https://textbookfull.com/product/microstates-entropy-and-quanta-
an-introduction-to-statistical-mechanics-don-koks/
An Introduction to Global Health Michael Seear

https://textbookfull.com/product/an-introduction-to-global-
health-michael-seear/

Psychometrics: An Introduction 3rd Edition Michael Furr

https://textbookfull.com/product/psychometrics-an-
introduction-3rd-edition-michael-furr/

Political Science An Introduction Michael G. Roskin

https://textbookfull.com/product/political-science-an-
introduction-michael-g-roskin/

An Introduction to Statistical Mechanics and

Thermodynamics 2nd Edition Robert H. Swendsen

https://textbookfull.com/product/an-introduction-to-statistical-
mechanics-and-thermodynamics-2nd-edition-robert-h-swendsen/

Dynamical Systems in Population Biology Xiao-Qiang Zhao

https://textbookfull.com/product/dynamical-systems-in-population-
biology-xiao-qiang-zhao/
Jared Michael Maruskin
Dynamical Systems and Geometric Mechanics
De Gruyter Studies in
Mathematical Physics

|
Edited by
Michael Efroimsky, Bethesda, Maryland, USA
Leonard Gamberg, Reading, Pennsylvania, USA
Dmitry Gitman, São Paulo, Brazil
Alexander Lazarian, Madison, Wisconsin, USA
Boris Smirnov, Moscow, Russia

Volume 48
Jared Michael Maruskin
Dynamical Systems
and Geometric
Mechanics

|
An Introduction

2nd edition
Physics and Astronomy Classification Scheme 2010
35-02, 65-02, 65C30, 65C05, 65N35, 65N75, 65N80

Author
Dr. Jared Michael Maruskin
San Jose CA 95128, USA
jared.maruskin@gmail.com

ISBN 978-3-11-059729-5
e-ISBN (PDF) 978-3-11-059780-6
e-ISBN (EPUB) 978-3-11-059803-2
ISSN 2194-3532

Bibliographic information published by the Deutsche Nationalbibliothek

The Deutsche Nationalbibliothek lists this publication in the Deutsche Nationalbibliografie;
detailed bibliographic data are available on the Internet at http://dnb.dnb.de.

© 2018 Walter de Gruyter GmbH, Berlin/Boston

Typesetting: VTeX UAB, Lithuania
Printing and binding: CPI books GmbH, Leck

www.degruyter.com
Contents
Preface | IX

Part I: Dynamical Systems

1 Linear Systems | 3
1.1 Eigenvector Approach | 3
1.2 Matrix Exponentials | 11
1.3 Matrix Representation of Solutions | 15
1.4 Stability Theory | 24
1.5 Fundamental Matrix Solutions | 28
1.6 Nonhomogeneous and Nonautonomous Systems | 35
1.7 Application: Linear Control Theory | 41

2 Linearization of Trajectories | 45
2.1 Introduction and Numerical Simulation | 45
2.2 Linearization of Trajectories | 47
2.3 Stability of Trajectories | 50
2.4 Lyapunov Exponents | 51
2.5 Linearization and Stability of Fixed Points | 56
2.6 Dynamical Systems in Mechanics | 66
2.7 Application: Elementary Astrodynamics | 68
2.8 Application: Planar Circular Restricted Three-Body Problem | 71

3 Invariant Manifolds | 81
3.1 Asymptotic Behavior of Trajectories | 81
3.2 Invariant Manifolds in ℝn | 85
3.3 Stable Manifold Theorem | 88
3.4 Contraction Mapping Theorem | 92
3.5 Graph Transform Method | 95
3.6 Center Manifold Theory | 101
3.7 Application: Stability in Rigid-Body Dynamics | 105

4 Periodic Orbits | 111

4.1 Summation Notation | 111
4.2 Poincaré Maps | 112
4.3 Poincaré Reduction of the State-Transition Matrix | 116
4.4 Invariant Manifolds of Periodic Orbits | 119
4.5 Families of Periodic Orbits | 121
VI | Contents

4.6 Floquet Theory | 122

4.7 Application: Periodic Orbit Families in the Hill Problem | 126

5 Bifurcations and Chaos | 131

5.1 Poincaré–Bendixson Theorem | 131
5.2 Bifurcation and Hysteresis | 134
5.3 Period Doubling Bifurcations | 138
5.4 Chaos | 141
5.5 Application: Billiards | 143

Part II: Geometric Mechanics

6 Differentiable Manifolds | 151

6.1 Differentiable Manifolds | 151
6.2 Vectors on Manifolds | 153
6.3 Mappings | 158
6.4 Vector Fields and Flows | 159
6.5 Jacobi–Lie Bracket | 161
6.6 Differential Forms | 167
6.7 Riemannian Geometry | 170
6.8 Application: The Foucault Pendulum | 177
6.9 Application: General Relativity | 179

7 Lagrangian Mechanics | 183

7.1 Hamilton’s Principle | 183
7.2 Variations of Curves and Virtual Displacements | 186
7.3 Euler–Lagrange Equation | 188
7.4 Distributions and Frobenius’ Theorem | 192
7.5 Mechanical Systems with Holonomic Constraints | 195
7.6 Nonholonomic Mechanics | 196
7.7 Application: Nöther’s Theorem | 208

8 Hamiltonian Mechanics | 213

8.1 Legendre Transform | 213
8.2 Hamilton’s Equations of Motion | 216
8.3 Hamiltonian Vector Fields and Conserved Quantities | 218
8.4 Routh’s Equations | 222
8.5 Symplectic Manifolds | 224
8.6 Symplectic Invariants | 229
8.7 Application: Optimal Control and Pontryagin’s Principle | 234
8.8 Application: Symplectic Probability Propagation | 240
 Contents | VII

9 Lie Groups and Rigid-Body Mechanics | 243

9.1 Lie Groups and Their Lie Algebras | 243
9.2 Left Translations and Adjoints | 248
9.3 Euler–Poincaré Equation | 253
9.4 Application: Rigid-Body Mechanics | 256
9.5 Application: Linearization of Hamiltonian Systems | 267

10 Moving Frames and Nonholonomic Mechanics | 271

10.1 Quasivelocities and Moving Frames | 271
10.2 A Lie Algebra Bundle | 274
10.3 Maggi’s Equation | 277
10.4 Hamel’s Equation | 279
10.5 Relation between the Hamel and Euler–Poincaré Equations | 289
10.6 Application: Constrained Optimal Control | 291

11 Fiber Bundles and Nonholonomic Mechanics | 297

11.1 Fiber Bundles | 298
11.2 The Transpositional Relation and Suslov’s Principle | 302
11.3 Voronets’ Equation | 305
11.4 Combined Hamel–Suslov Approach | 309
11.5 Application: Rolling-Without-Slipping Constraints | 313

Bibliography | 321

Index | 333
Preface
This book is based on a set of lecture notes from a graduate course in dynamical sys-
tems that I taught during spring 2010. The goal of this book is not to serve as an en-
cyclopedic reference for the fields of dynamical systems and geometric mechanics;
many wonderful and advanced books already exist, several of which I strive to direct
the reader to along the journey contained within these pages. Rather, my goal is to
provide a broad perspective on this beautiful subject in a brief and introductory man-
ner that can be taught within a single semester for advanced undergraduates or be-
ginning graduate students. The field of dynamical systems has grown to encompass
many branches of mathematics; this text will use notions from linear algebra, analy-
sis, topology, differential geometry, and Lie group theory. Though I strive to make such
a book as self-contained as possible, that magical ingredient referred to as mathemat-
ical maturity should be taken as prerequisite. This means that, as an eager student,
you do not have to have studied any of the particular fields of mathematics listed above
before taking on this text, but you should have a certain understanding of mathemat-
ical rigor and language that will facilitate your digestion of powerful ideas that are
presented in a concise fashion.

Theme

The theme of this book is systems of first-order differential equations. In the first part,
we describe various aspects of the theory of solutions of linear and nonlinear first-
order differential equations; in the second part, we take up the topic of differential
equations on smooth differentiable manifolds. Generally speaking, we are interested
in differential equations of the form ẋ = f (x). In Part I of the book, we take x ∈ ℝn . We
begin Chapter 1 by examining the case where f (x) = Ax, for some real-valued, n × n
matrix A. For the remaining chapters of Part I, we broaden our perspective by regard-
ing f (x) as a smooth vector field on ℝn . In Part II of the book, we are again interested
in the same equation, only x is regarded as a point on a manifold and f as a vector
field on that manifold. The solutions can be described as a smooth class of curves on
the manifold whose tangent vectors coincide with the vector field f at every point.
In addition to studying the mathematical and theoretical aspects of such systems,
we will pay keen attention to applications throughout the text, with an emphasis on
applications to mechanics. The second part of the book is particularly suited to the
modern theory of geometric mechanics. Newton’s laws of motion were originally writ-
ten down for systems in Euclidean spaces. However, starting with the works of La-
grange and Hamilton, dynamicists began to realize that a system’s configuration could
more accurately be viewed as a point on a smooth manifold: the simple pendulum

https://doi.org/10.1515/9783110597806-201
X | Preface

lives on a circle, the double pendulum lives on a torus, and a satellite in orbit about
the Earth lives on a certain matrix Lie group.

Map of the Book

To provide the greatest amount of flexibility, I have provided a map of this text in Fig-
ure 1. The text is divided into two parts: Dynamical Systems and Geometric Mechanics,
respectively. Chapters 1–3 form the core of Part I of the book and should be taken as
prerequisite for Part II. Similarly, Chapters 6 and 7 constitute the core of Part II and
should be taken as prerequisite from which one may spring directly into Chapter 8, 9,
or 10.

Figure 1: Map of this book.

The purpose of this book is to provide a sophisticated and concise introduction to the
fields of dynamical systems and geometric mechanics that can be covered in a single-
semester course. For shorter courses that aim to cover either the first or second half of
the text, I have included various applications at the end of each chapter that can serve
as springboards for further discussion, research, projects, or investigation.

Background and References

I have written this book aimed at graduate students in mathematics, physics, and engi-
neering with previous coursework experience in linear algebra and differential equa-
tions and an undergraduate course in dynamical systems, at the level of analyzing
 Preface | XI

two-dimensional systems and phase planes. This book, however, is self-contained, so

that astute students without a background in differential equations and planar dy-
namical systems should still manage to thrive. I kept this flexibility in mind when
writing the first part of the book. The first chapter on linear systems, for example, is
more advanced than students with an undergraduate background in dynamical sys-
tems would have previously encountered, so the material should not bore them, but
it is comprehensive enough to provide a complete picture of linear systems to new-
comers. A similar statement could be made with regard to the remaining chapters in
Part I: material that some students may have previously encountered will be presented
in a more sophisticated, yet complete, way, so that experience with the material is not
absolutely necessary for a student possessing mathematical maturity.
I should, however, recommend several supplements to brave students who wish
to tackle the text with no previous coursework in dynamical systems. Most of the gist
can be picked up here; however, for a lighter discussion and reprieve one might con-
sult an elementary book on differential equations such as [34], which also contains a
very nice and concise chapter on linear and nonlinear phase plane analysis (some stu-
dents might want to start with this chapter). A second book on nonlinear phase plane
dynamics, such as [115], [138], or [274], should also be referenced so as to form a com-
plete picture of the subject and its applications. The Strogatz text is particularly good
in that it abounds with a rich tapestry of applications and conceptual clarity, whereas
the other two contain a greater amount of mathematical rigor.
For novice graduate and ambitious undergraduate students taking on this text, it
will become paramount, as you begin your careers in research, to learn how to consult
a variety of sources on a given topic in order to fully appreciate and comprehend its
complexity. Different authors offer different perspectives on any particular topic, and
these varying perspectives, taken in aggregate, yield a more holistic understanding of
the subject. As such, I have strived to offer a variety of references for additional ma-
terial throughout the text. I would like to make special mention of several important
ones up front. For Part I of the text, [110], [211], [235], and [294] constitute a nice se-
lection of references to complement this text. In those works you will find additional
relevant discussions, references, and topics. For Part II, [8], [11], [104, 106], and [222]
constitute a set of classic references for Lagrangian, Hamiltonian, and nonholonomic
mechanics, whereas [24], [37], and [196] are excellent references that go into greater
depth on the geometric formulation of mechanics and that might also form a basis for
subsequent studies of this book’s subject matter.

Acknowledgments

I am incredibly grateful for and indebted to the many enjoyable and enlightening con-
versations I have had with Anthony M. Bloch, Daniel J. Scheeres, and Dmitry Zenkov
pertaining to many of the topics and discussions contained within these pages, as well
XII | Preface

as to the suggestions of the anonymous reviewers that led to the formation of a bet-
ter manuscript and to the copyediting work performed by Glenn Corey, who helped
improve the overall readability of the text and helped enforce global consistencies in
style.
Finally, I am very pleased that you have chosen to undertake the study of such an
exciting and elegant subject. I look forward to embarking on our adventure together
in the pages that lie ahead.

San José, California Jared M. Maruskin

September 2018
|
Part I: Dynamical Systems
1 Linear Systems
Linear systems constitute an important hallmark of dynamical systems theory. Not
only are they among the rare class of systems in which exact, analytic solutions can
actually be obtained and computed, they have also become a cornerstone even in the
analysis of nonlinear systems, including systems on manifolds. We will see later in this
book that the flow near a fixed point can be locally understood by its associated lin-
earization. The different solution modes of a linear system may be analyzed individu-
ally using linearly independent solutions obtained from eigenvalues and eigenvectors
or matrixwise using operators and operator exponentials. We begin our study of linear
systems with a discussion of both approaches and some fundamental consequences.

1.1 Eigenvector Approach

Once the early pioneers of linear dynamical systems realized they could represent
these systems in matrix form, it was not long before they realized that, when expressed
relative to an eigenbasis of the constant coefficient matrix, a spectacular decoupling
results, yielding the solution almost trivially. We begin our study of linear systems
with a review of this method, paying particularly close attention to the case of distinct
eigenvalues. We will consider the case of repeated eigenvalues later in the chapter.

Linear Systems and the Principle of Superposition

Throughout this chapter, we will consider linear systems of first-order ordinary differ-
ential equations of the form

ẋ = Ax, (1.1)
x(0) = x0 , (1.2)

where x : ℝ → ℝn is the unknown solution to the above initial value problem, x0 ∈ ℝn

is the initial condition, and A ∈ ℝn×n is a real-valued matrix called the coefficient ma-
trix. The solution, or flow, of system (1.1) is a smooth function φ : ℝ × ℝn → ℝn that
satisfies
dφ
= Aφ(t; x0 ), for all t ∈ ℝ, (1.3)
dt
φ(0; x0 ) = x0 , for all x0 ∈ ℝn . (1.4)

The singular curve x(t) = 0 trivially satisfies differential equation (1.1) with initial
condition x0 = 0 for all t ∈ ℝ. Hence the point x = 0 is referred to as a fixed point of
the system.

https://doi.org/10.1515/9783110597806-001
4 | 1 Linear Systems

We say that a curve z : ℝ → ℂn is a complex-valued solution to (1.1) if both its real

and imaginary parts are solutions to (1.1), i. e., if z(t) = x(t)+iy(t), where x, y : ℝ → ℝn ,
then z is a complex-valued solution to (1.1) if and only if both x and y are real-valued
solutions to (1.1). To facilitate our analytic understanding of system (1.1), we first state
a theorem dealing with linear combinations of particular solutions of the system.

Theorem 1.1 (Principle of Superposition). Suppose z1 , . . . , zk : ℝ → ℂn are k distinct

(possibly complex-valued) solutions to the system of differential equations (1.1). Then
the linear combination

z(t) = c1 z1 (t) + ⋅ ⋅ ⋅ + ck zk (t) (1.5)

is also a solution of (1.1) for arbitrary scalar coefficients c1 , . . . , ck ∈ ℂ.

Proof. This follows directly from the fact that both A and d/dt are linear operators.
First, suppose that z1 , z2 : ℝ → ℂn are two complex-valued solutions. To show that
an arbitrary linear combination satisfies the system of differential equations (1.1), we
simply check that
d
(c z + c2 z2 ) = c1 z1̇ + c2 z2̇ = c1 Az1 + c2 Az2 = A(c1 z1 + c2 z2 ).
dt 1 1
The final result holds by induction.

Now suppose that v ∈ ℂn is an eigenvector of coefficient matrix A, with λ ∈ ℂ

being its associated eigenvalue. It is clear that x(t) = veλt is a particular solution to dif-
ferential equation (1.1). Assuming the uniqueness of solutions to the preceding linear
system, we immediately obtain the following general solution.

Theorem 1.2. Suppose matrix A has n distinct eigenvalues λ1 , . . . , λn ∈ ℂ with associ-

ated eigenvectors v1 , . . . , vn ∈ ℂn . Then the general solution of the system of first-order
differential equations (1.1) is given by

x(t) = c1 v1 eλ1 t + ⋅ ⋅ ⋅ + cn vn eλn t , (1.6)

where c1 , . . . , cn ∈ ℂ are arbitrary constants.

Real, Distinct Eigenvalues

Now let us restrict our attention to systems that have real, distinct eigenvalues, so that
the general solution (1.6) may be regarded as a real-valued solution, where each of the
coefficients c1 , . . . , cn ∈ ℝ.

Example 1.1. Compute the general solution of system (1.1), where

3 −2
A=[ ].
0 −1
 1.1 Eigenvector Approach | 5

The eigenvalues and eigenvectors of A are λ1 = 3, λ2 = −1, v1 = (1, 0), and v2 = (1, 2),
respectively. Therefore, the general solution is given by
1 1
x(t) = c1 e3t [ ] + c2 e−t [ ] .
0 2
Observe that if the initial condition is x0 = αv1 , then the particular solution of the
initial value problem (1.1)–(1.2) is x(t) = αe3t v1 . Hence the solution trajectory remains
on the line defined by the span of v1 for all time and the solution grows exponentially
in the direction away from the origin.
On the other hand, if the initial condition is x0 = αv2 , then the particular solution
of the initial value problem is x(t) = αe−t v2 . Again, the solution remains on the line
defined by the span of v2 for all time, and the solution decays exponentially toward
the origin.
Since span{v1 , v2 } = ℝ2 , vectors v1 and v2 form a basis for ℝ2 . The constants c1 and
c2 are simply the components of the initial condition vector with respect to this basis.
Hence, the general solution flow will be a linear combination of the two previously
mentioned cases. The phase portrait, i. e., a graph of representative solution trajecto-
ries, is plotted in Figure 1.1. This type of fixed point is called a saddle point because of
the shape of the surrounding phase flow (it helps to imagine the level sets of the flow
in three dimensions).

Figure 1.1: Example 1.1: flow around a saddle

point.

Example 1.2. Consider linear system (1.1) with the coefficient matrix
1.5 0.5
A=[ ].
0.5 1.5
The eigenvalues and eigenvectors of A are λ1 = 2, λ2 = 1, v1 = (1, 1), and v2 = (−1, 1),
respectively. Hence the general solution is given by
1
x(t) = c1 e2t [ ] + c2 et [ ] .
−1
1 1
6 | 1 Linear Systems

Again, if the initial condition lies in the direction of either eigenvector, then the solu-
tion flow will remain on the line defined by that eigenvector for all time. For a general
initial condition, coefficients c1 and c2 will be the components of x0 with respect to
the eigenbasis {v1 , v2 }. Since both eigenvalues are positive, all nonzero solutions will
grow exponentially away from the origin as t → ∞. If x0 does not lie in the direction
defined by either eigenvector, then the solution will veer toward v1 as t → ∞ since the
solution component relative to v1 has an exponential growth constant λ1 that is greater
than the exponential growth constant λ2 relative to v2 . A fixed point with two eigen-
values of the same sign is referred to as a node or nodal point. Since solutions starting
near the origin exponentially diverge away from the origin, the fixed point in this ex-
ample is referred to as an unstable node. The phase flow is plotted in Figure 1.2.

Figure 1.2: Example 1.2: flow around an unstable

node.

Exercise 1.1. Compute the general solution of the linear system ẋ = Ax, where the
coefficient matrix is given by

0 −1
A=[ ].
4 −5

Plot the phase portrait of the flow.

Exercise 1.2. Compute the general solution of the linear system ẋ = Ax, where the
coefficient matrix is given by

−1 −1 3
A = [0
[ ]
−2 3] .
[0 0 1]

Plot a sketch of the solution flow in ℝ3 . What happens to trajectories that initially
hover above the x1 -x2 plane as t → ∞?
 1.1 Eigenvector Approach | 7

These examples exhibit a fundamental feature of linear systems: initial conditions

in a subspace that is spanned by the eigenvectors associated with all the positive or
negative eigenvalues result in trajectories that remain within that subspace. Next we
will define these concepts more rigorously and then return to them in §1.4.

Definition 1.1. Suppose A ∈ ℝn×n has k distinct negative eigenvalues λ1 , . . . , λk and

(n − k) distinct positive eigenvalues λk+1 , . . . , λn . The stable and unstable subspaces of
linear system (1.1) are defined by

E s = span{v1 , . . . , vk },
E u = span{vk+1 , . . . , vn },

respectively.

Definition 1.2. A subspace W ⊂ ℝn is said to be an invariant subspace under the flow

of linear system (1.1) if φ(t; x0 ) ∈ W for all t ∈ ℝ whenever x0 ∈ W.

Exercise 1.3. Show that if x0 ∈ E s , then φ(t; x0 ) ∈ E s for all t and, moreover, that
limt→∞ φ(t; x0 ) = 0. Similarly, show that if x0 ∈ E u , then φ(t; x0 ) ∈ E u for all t and,
moreover, that limt→−∞ φ(t; x0 ) = 0. Conclude that E s and E u are invariant subspaces
of linear system (1.1). The invariant subspaces E s and E u are referred to as the stable
and unstable subspaces of the system, respectively.

Complex, Distinct Eigenvalues

We now consider the case in which the coefficient matrix A in (1.1) has k ≤ n dis-
tinct, real-valued eigenvalues and (n − k) distinct, complex-valued eigenvalues. Since
complex eigenvalues and their corresponding eigenvectors always occur in complex
conjugate pairs, we may define the integer s = (n − k)/2 as the number of such pairs.
The eigenvalues may therefore be arranged so that λ1 , . . . , λk ∈ ℝ, and

λj = aj + ibj , λj = aj − ibj , for j = (k + 1), . . . , (k + s),

where aj , bj ∈ ℝ for j = (k + 1), . . . , (k + s). Similarly, the first k associated eigenvectors

may be taken as w1 , . . . , wk ∈ ℝn , whereas the associated complex eigenvectors, which
also occur in complex conjugate pairs, may be taken as

wj = uj + ivj , wj = uj − ivj , for j = (k + 1), . . . , (k + s),

where uj , vj ∈ ℝn for j = (k + 1), . . . , (k + s).

In other words, the eigenvalues of A are λ1 , . . . , λk , ak+1 +ibk+1 , ak+1 −ibk+1 , . . . , ak+s +
ibk+s , ak+s − ibk+s . Similarly, the eigenvectors of A are w1 , . . . , wk , uk+1 + ivk+1 , uk+1 −
ivk+1 , . . . , uk+s + ivk+s , uk+s − ivk+s .
8 | 1 Linear Systems

Definition 1.3. Given a set of k complex-valued vectors w1 , . . . , wk ∈ ℂn , we define the

real span of the vectors as

realspan{w1 , . . . , wk } = span{ℜ{w1 }, ℑ{w1 }, . . . , ℜ{wk }, ℑ{wk }} ⊂ ℝn ,

where ℜ{w} and ℑ{w} are the real and imaginary parts of w, respectively.

We expect the final solution of (1.1) to be real-valued. Hence the complex-valued

coefficients of the general solution (1.6) must be such that their sum is a real-valued
vector function in ℝn .
We know from (1.6) that eλj t wj and eλj t wj are distinct, complex-valued solutions of
(1.1) for j = (k+1), . . . , (k+s). From the principle of superposition we know that arbitrary
(complex) linear combinations of these solutions are solutions as well. Hence, their
real and imaginary parts are solutions. Since eλj t wj and eλj t wj are complex conjugates
of each other, it follows that

realspan{eλj t wj , eλj t wj } = span{ℜ{eλj t wj }, ℑ{eλj t wj }},

for j = (k + 1), . . . , (k + s). We may therefore “trade” the complex solutions eλj t wj and
eλj t wj for the real solutions ℜ{eλj t wj } and ℑ{eλj t wj } for each j = (k + 1), . . . , (k + s). Hence
the following theorem.

Theorem 1.3. Suppose linear system (1.1) has k ≤ n distinct, real-valued eigenvalues
λ1 , . . . , λk and s = (n − k)/2 complex conjugate eigenvalue pairs

λk+1 , λk+1 , . . . , λk+s , λk+s .

Then the general, real-valued solution of (1.1) is given by

k k+s
x(t) = ∑ cj eλj t wj + ∑ (cj ℜ{eλj t wj } + dj ℑ{eλj t wj }),
j=1 j=k+1

where c1 , . . . , ck+s , dk+1 , . . . , dk+s ∈ ℝ.

Notice that the real and imaginary parts of eλj t wj , up to a sign, are identical to the
real and imaginary parts of eλj t wj , for j = (k + 1), . . . , (k + s). This is why we are allowed
to replace complex conjugate pair solutions with their real and imaginary parts.

Example 1.3. Consider the linear system ẋ = Ax, where coefficient matrix A is given
by

−2 2
A=[ ].
−4 2

The eigenvalues and eigenvectors are λ1 = 2i, λ1 = −2i, w1 = (1, 1 + i), and w1 = (1, 1 −
i), respectively. Note that the eigenvalues are purely imaginary. By Theorem 1.3, the
 1.1 Eigenvector Approach | 9

general solution is given by

x(t) = c1 ℜ{eλ1 t w1 } + d1 ℑ{eλ1 t w1 },

where c1 , d1 ∈ ℝ. Using Euler’s equation

eiθ = cos θ + i sin θ, (1.7)

we may write

1
eλ1 t w1 = (cos(2t) + i sin(2t)) [ ]
1+i
cos(2t) + i sin(2t)
=[ ].
(cos(2t) − sin(2t)) + i(cos(2t) + sin(2t))

Hence the general solution is given by

cos(2t) sin(2t)
x(t) = c1 [ ] + c2 [ ].
cos(2t) − sin(2t) cos(2t) + sin(2t)

The phase portrait is plotted in Figure 1.3. Orbits form concentric, closed periodic loops
about the origin. This type of fixed point is referred to as a center.

Figure 1.3: Example 1.3: flow around a center.

Example 1.4. Consider the linear system ẋ = Ax, where coefficient matrix A is given
by

−1 2
A=[ ].
−2 −1
10 | 1 Linear Systems

The eigenvalues and eigenvectors are λ1 = −1 + 2i, λ1 = −1 − 2i, w1 = (1, i), w1 = (1, −i).
From Euler’s equation (1.7) we have

e(a+bi) = eat (cos(bt) + i sin(bt)) .

Hence,

1 e−t cos(2t) + ie−t sin(2t)

eλ1 t w1 = e−t (cos(2t) + i sin(2t)) [ ] = [ −t ].
i −e sin(2t) + ie−t cos(2t)

Therefore, by Theorem 1.3, the general solution is given by

cos(2t) sin(2t)
x(t) = c1 e−t [ ] + c2 e−t [ ].
− sin(2t) cos(2t)

Hence the oscillations due to the imaginary eigenvalues decay exponentially so that
solutions spiral in toward the origin. In this context, the origin is referred to as a stable
spiral. The phase portrait for this flow is plotted in Figure 1.4. Notice that the exponen-
tial decay is due to the real part of the eigenvalues. Had the real parts been positive,
the solutions would have spiraled away from the origin, and the origin would have
been referred to as an unstable spiral.

Figure 1.4: Example 1.4: flow around a stable spi-

ral.

Exercise 1.4. Determine the general solution of the system

−4 3 2
ẋ = [−6 6 ] x.
[ ]
2
[0 0 −2]

Sketch the flow in ℝ3 .

 1.2 Matrix Exponentials | 11

1.2 Matrix Exponentials

In a first course on ordinary differential equations, one doubtlessly encounters the fact
that the solution to the initial value problem

ẋ = ax, x(0) = x0 ,

for a ∈ ℝ, is x(t) = x0 eat . In §1.3, we will generalize this result to coupled linear systems
in ℝn , i. e., we will see that if ẋ = Ax, where A ∈ ℝn×n , then

x(t) = eAt x0 ,

given some appropriate understanding of the matrix eAt .

Before laying out the precise definition of a matrix exponential, we will make a
brief interlude to review a few key concepts from functional analysis, so that we may
obtain a more sophisticated understanding of this definition. These concepts will also
be useful later during our discussion of the graph transform method, which we will
use to prove the existence of stable and unstable manifolds of fixed points in nonlinear
systems. For more details on functional analysis, see, for example, [107], [171], or [257].

Normed Linear Spaces

We begin with the definition of normed linear space. After discussing notions of con-
vergence, we will further introduce the concept of a Banach space, which we will re-
quire later in §3.4, preceding our discussion of the contraction mapping theorem.

Definition 1.4. A real (or complex) vector space is a set V with the following opera-
tions:
1. Vector addition: any x, y ∈ V determines an element x + y ∈ V that satisfies the
following properties:
(a) x + y = y + x for all x, y ∈ V;
(b) x + (y + z) = (x + y) + z for all x, y, z ∈ V;
(c) there is an element of V called 0 such that 0 + x = x for all x ∈ V;
(d) given any element x ∈ V, there is an element called −x ∈ V such that x +
(−x) = 0.
2. Scalar multiplication: any x ∈ V and α ∈ ℝ (or ℂ) determines an element αx ∈ V
that satisfies the following properties:
(a) α(βx) = (αβ)x for all α, β ∈ ℝ (or ℂ) and x ∈ V;
(b) 1x = x for any x ∈ V;
(c) (α + β)x = αx + βx for all α, β ∈ ℝ (or ℂ) and x ∈ V;
(d) α(x + y) = αx + αy for all α ∈ ℝ (or ℂ) and x, y ∈ V.

Example 1.5. As the vectors in ℝn satisfy each of the above axioms, ℝn is clearly a real
vector space.
12 | 1 Linear Systems

Exercise 1.5. Show that C 0 ([0, 1]), the set of continuous functions on the interval [0, 1],
is a real vector space.

Exercise 1.6. Show that

Sδ ([0, 1]) = {f ∈ C 0 ([0, 1]) : |f (x) − f (y)| ≤ δ|x − y| for all x, y ∈ [0, 1]},

the set of Lipschitz continuous functions with Lipschitz constant δ > 0, is not a real
vector space.

Exercise 1.7. Show that L(ℝn ), the set of linear transformations A from ℝn → ℝn , is a
real vector space.

Definition 1.5. Let V be a real (or complex) vector space. Then a norm on V is a map-
ping ‖ ⋅ ‖ : V → ℝ that satisfies the following conditions:
1. ‖x‖ ≥ 0 for all x ∈ V, and ‖x‖ = 0 if and only if x = 0;
2. ‖αx‖ = |α| ‖x‖ for all x ∈ V and α ∈ ℝ (or ℂ);
3. ‖x + y‖ ≤ ‖x‖ + ‖y‖ for all x, y ∈ V (the triangle inequality).

A vector space with a norm is called a normed linear space.

Exercise 1.8. Show that

‖f ‖ = sup |f (x)| (1.8)

x∈J

defines a norm on C 0 (J), where J ⊂ ℝ (called the sup norm).

Given a normed linear space V, an ε-ball centered at the point x ∈ V is the set
Bε (x) = {y ∈ V : ‖x − y‖ < ε}. These ε-balls form a basis for a topology on V (see, for
example, [216]), i. e., arbitrary open sets can be defined as arbitrary unions of these
open balls. Moreover, a set is closed if its complement in V is open.
An important concept in analysis is that of convergence of a sequence of points in
a space. Many of the applications in this chapter rely on such a notion. We presume the
reader is familiar with convergence of sequences of real numbers, and we introduce
the notion of convergence in a normed linear space.

Definition 1.6. Let V be a normed linear space and let (xn )∞

n=1 = (x1 , x2 , . . .) be a se-
quence of points in V. The sequence (xn ) is said to converge to the element x ∈ V,
denoted

lim x = x,
n→∞ n

if the real numbers ‖xn − x‖ → 0 as n → ∞.

Definition 1.7. Let V be a normed linear space and let (xn )∞n=1 be a sequence of points
in V. The sequence (xn ) is said to be a Cauchy sequence if, for every ε > 0, there exists
an N such that ‖xn − xm ‖ < ε whenever n, m > N.
 1.2 Matrix Exponentials | 13

It is true of the real numbers that every Cauchy sequence converges to some point
in ℝ; see, for instance, [252]. However, the same is not necessarily true of a generic
normed linear space. The situation thus merits the following distinction.

Definition 1.8. A normed linear space V is said to be complete if every Cauchy se-
quence in V converges to some element of V. A Banach space is a complete normed
linear space.

Exercise 1.9. Show that C 0 ([0, 1]) with the sup norm is a Banach space.

Matrix Exponentials

Let L(ℝn ) be the space of linear transformations T : ℝn → ℝn . As we saw in Exer-

cise 1.7, L(ℝn ) is a real vector space. Before defining the matrix exponential, we must
first endow this space with a norm.

Definition 1.9. The operator norm of T ∈ L(ℝn ) is defined by

‖T‖ = sup{|T(x)| : x ∈ ℝn and |x| ≤ 1}.

Exercise 1.10. Show that the operator norm satisfies the properties that define a norm,
i. e., for S, T ∈ L(ℝn ), show that
(a) ‖T‖ ≥ 0 and ‖T‖ = 0 if and only if T = 0;
(b) ‖kT‖ = |k| ⋅ ‖T‖ for k ∈ ℝ;
(c) ‖S + T‖ ≤ ‖S‖ + ‖T‖.

Definition 1.10. A sequence of linear operators Tk ∈ L(ℝn ) is said to converge to the

linear operator T ∈ L(ℝn ) if, for all ε > 0, there exists an N ∈ ℕ such that ‖T − Tk ‖ < ε
for all k > N.

Definition 1.11 (The Matrix Exponential). Let A be a real-valued, n×n matrix. For t ∈ ℝ,
we define the matrix exponential of At as

∞
Ak t k
eAt = ∑ . (1.9)
k=0
k!

Our goal is to show that for any t0 > 0, the series defined in (1.9) converges uni-
formly on the interval [−t0 , t0 ]. But first we will require the following lemma.

Lemma 1.1. For S, T ∈ L(ℝn ) and x ∈ ℝn :

(i) |T(x)| ≤ ‖T‖ ⋅ |x|;
(ii) ‖TS‖ ≤ ‖T‖ ⋅ ‖S‖;
(iii) ‖T k ‖ ≤ ‖T‖k , for k ∈ ℤ+ .
14 | 1 Linear Systems

Proof. 1. It is obvious that (i) holds for x = 0. For x ≠ 0, define y = x/|x|. Then, by
the definition of the operator norm, we have

|T(x)|
‖T‖ ≥ |T(y)| = ,
|x|

from which (i) follows.

2. From (i) it follows that

|T(S(x))| ≤ ‖T‖ ⋅ |S(x)|

≤ ‖T‖ ⋅ ‖S‖ ⋅ |x|.

For |x| ≤ 1, it further follows that |T(S(x))| ≤ ‖T‖ ⋅ ‖S‖. We therefore obtain the
inequality

‖TS‖ = sup{|TS(x)| : x ∈ ℝn and |x| ≤ 1} ≤ ‖T‖ ⋅ ‖S‖.

3. The third inequality follows immediately from (ii) using mathematical induction.

Theorem 1.4. Given T ∈ L(ℝn ) and t0 > 0, the series

∞
T k tk
∑
k=0
k!

converges uniformly on the interval −t0 ≤ t ≤ t0 .

Proof. From the lemma we have, for |t| ≤ t0 ,

󵄩󵄩 T k t k 󵄩󵄩 ‖T‖k |t|k ‖T‖k t k

󵄩󵄩 󵄩󵄩 0
󵄩󵄩 󵄩󵄩 ≤ ≤ .
󵄩󵄩 k! 󵄩󵄩 k! k!
But
∞
‖T‖k t0k
∑ = e‖T‖t0 .
k=0
k!

Therefore, the series in question converges by the Weierstrass M-Test.

Our next theorem, regarding the relation of the matrix exponential of two similar
matrices, is the key to solving linear systems of differential equations. Recall that two
matrices S, T ∈ L(ℝn ) are similar if there exists an invertible matrix P ∈ L(ℝn ) such
that S = PTP −1 .

Theorem 1.5. Let S, T ∈ L(ℝn ) be similar matrices, and let P be an invertible matrix with
the property S = PTP −1 . Then

eS = PeT P −1 .
 1.3 Matrix Representation of Solutions | 15

Proof. This follows immediately from the definition of the matrix exponential and
from the fact that

Sk = PT k P −1

for every k ∈ ℕ. Specifically, we have

n n
(PTP −1 )k Tk
eS = lim ∑ = P [ lim ∑ ] P −1 = PeT P −1 .
n→∞
k=0
k! n→∞
k=0
k!

Exercise 1.11. Show that if S, T ∈ L(ℝn ) commute (i. e., ST = TS), then eS+T = eS eT .
Hint: Use the binomial theorem. Does the same result hold if S and T do not commute?

Exercise 1.12. Show that if D = diag{λ1 , . . . , λn }, where λ1 , . . . , λn ∈ ℝ, then eDt =

diag{eλ1 t , . . . , eλn t }.

Exercise 1.13. Show that

cos b − sin b a
eA = ea [
−b
], where A = [ ].
sin b cos b b a

Exercise 1.14. Show that

1 b a b
eA = ea [ ], where A = [ ].
0 1 0 a

1.3 Matrix Representation of Solutions

We now have the ability to discuss the solutions of linear systems in a more elegant and
succinct way than before. Our basic aim is to show how solutions may be represented
using matrix exponentials, diagonalization, and block diagonalization and to explore
the particular form solutions take for a variety of different types of eigenvalues. For an
early reference on such operator representations of solutions, see [193].

Real, Distinct Eigenvalues

Let us first revisit linear systems of the form (1.1), where coefficient matrix A has real,
distinct eigenvalues λ1 , . . . , λn ∈ ℝ and associated eigenvectors v1 , . . . , vn . The general
solution is given by

x(t) = c1 eλ1 t v1 + ⋅ ⋅ ⋅ + cn eλn t vn ,

where (c1 , . . . , cn ) are the components of the initial condition x0 relative to the eigen-
basis v1 , . . . , vn . Before we proceed, let us recall a theorem from linear algebra.
16 | 1 Linear Systems

Theorem 1.6 (Diagonalization). If the eigenvalues λ1 , . . . , λn ∈ ℂ of an n × n matrix A ∈

ℝn×n are distinct, then the matrix V ∈ ℂn×n defined by

| |
V = [v1 (1.10)
[ ]
⋅⋅⋅ vn ] ,
[| |]

where v1 , . . . , vn ∈ ℂn are the associated eigenvectors, is invertible and

V −1 AV = diag{λ1 , . . . , λn } = D. (1.11)

Proof. Equation (1.11) is equivalent to the matrix equation AV = VD. Now, the ith col-
umn of this equation is simply Avi = λi vi , which is true by definition of the eigenvalues
and eigenvectors. The invertibility of V follows from the fact that the eigenvalues are
distinct, which implies that their associated eigenvectors must be linearly indepen-
dent.

This theorem is equally applicable to any combination of real or complex eigen-

values. For the remainder of this section, we will restrict our attention to the case of
n real, distinct eigenvalues. Let us define matrix V as in (1.10) and D as the diagonal
matrix whose entries consist of the eigenvalues of A, as in (1.11). We then introduce
the change of variables y = V −1 x. The differential equations for vector y will then be
given by

ẏ = V −1 ẋ = V −1 Ax = V −1 AVy = Dy.

Hence, relative to eigenbasis V, system (1.1) decouples as

ẏ1 = λ1 y1 ,
..
.
ẏn = λn yn .

The general solution, in terms of the new variable y, is therefore given by

y1 (t) = c1 eλ1 t ,
..
.
yn (t) = cn eλn t ,

or, in matrix form, as

eλ1 t
[
y(t) = [ .. ]
] y(0) = eDt y0 ,
[ . ]
[ eλn t ]
Another random document with
no related content on Scribd:
 “Well, I saw the shortcake on the window, and I thought maybe it
was to be thrown away, so I picked it up. I didn’t know anybody
wanted it.”
“Well, you know now,” said Mr. Bunker grimly. “And you had better
not try any more tricks like that. Are you sure you didn’t take
anything more by mistake?”
“No, I take nothing more,” answered the boy sullenly, as he
fastened his box again, and, slinging that and his basket of wares
over his shoulder, away he walked. He was quite angry at being
caught, it appeared.
“Oh, I’m so glad I got my shortcake back!” cried Rose. “Now we
can eat it when we get back to the house.”
“Do you think it was kept clean?” asked her mother.
But they need not have worried on that score. Whatever else he
was, the peddler boy seemed clean, and he had wrapped a clean
paper about the short cake before putting it in his box. To be sure
some of the strawberries on top were crushed and a little of their red
juice had run down the sides of the cake.
“But that doesn’t matter, ’cause we got to smash it a lot more when
we eat it,” said Laddie.
Which, of course, was perfectly true.
So Rose’s shortcake came back to Farmer Joel’s and they sat
down to the table again and ate it. Dessert was a little late that
evening, but it was liked none the less.
“Busy day to-morrow, children!” said Farmer Joel, as the six little
Bunkers went up to bed.
“What doing?” asked Russ.
“Getting in the hay!” was the answer. “Those who can’t help can
ride on the hay wagon.”
There were whoops of delight from the six little Bunkers.
“Could I drive the horses?” asked Russ.
 “Well, we’ll see about that,” answered Farmer Joel slowly.
“I want to ride on the rake that makes the hay into heaps like
Eskimo houses,” announced Laddie.
“You’d better not do that,” his mother said. “You might fall off and
get raked up with the hay.”
“I’ll look after them, and so will Adam North!” chuckled Farmer
Joel. “So to bed now, all of you. Up bright and early! We must get the
hay in before it rains!”
 CHAPTER XV
AN EXCITING RIDE

Very seldom did the six little Bunkers need any one to call them to
get them out of bed. Generally they were up before any one else in
Farmer Joel’s house. The morning when the hay was to be gotten in
was no exception.
Almost as soon as Norah had the fire started and breakfast on the
way, Russ, Rose and the others were impatient to start for the hay
field.
“Why does he have to get the hay in before it rains?” asked Violet
of her father, remembering what Farmer Joel had said the night
before.
“Because rain spoils hay after it has been cut and is lying in the
field ready to be brought in,” answered Mr. Todd, who heard Vi’s
question. “Once hay is dried, it should be brought in and stored away
in the barn as soon as possible.
“After it is raked up and made into cocks, or Eskimo houses, as
Laddie calls them, if it should rain we’d have to scatter the hay all
over again to dry it out. For if it were to be put away in the haymow
when wet the hay would get mouldy and sour, and the horses would
not eat it.”
“Also if the hay gets rained on after it is cut and dried, and while it
is still scattered about the field, it must be turned over so the wet part
will dry in the hot sun before it can be hauled in. We have had
several days of hot weather and my hay is fine and dry now. That’s
why I am anxious to get it into the barn in a hurry.”
“Yes, I think we had better hurry,” said Adam North who, with a
couple of other hired men, was to help get in Farmer Joel’s hay.
“We’re likely to have thunder showers this afternoon.”
 “Then we must all move fast!” exclaimed Daddy Bunker, who liked
to work on the farm almost as much as did Adam and Mr. Todd.
The day before the hay had been raked into long rows by Adam,
who rode a large two-wheeled rake drawn by a horse. The rake had
long curved prongs, or teeth, which dragged on the ground pulling
the hay with them. When a large enough pile of hay had been
gathered, Adam would press on a spring with one foot and the teeth
of the rake would lift up over the long row of dried grass. This was
kept up until the field was filled with many rows of hay, like the waves
on the seashore.
Then men went about piling the hay into cocks, or cone-shaped
piles, which, as Laddie said, looked like the igloos of the Eskimos.
Now all that remained to be done was to load the hay on a big, broad
wagon and cart it to the barn.
Laddie was a bit disappointed because all the hay was raked up,
for he wanted to ride on the big machine which did this work. But
Farmer Joel said:
“We always have a second raking after we draw in the hay, for a
lot of fodder falls off and is scattered about. You shall ride on the
rake when we go over the field for the second time.”
So Laddie felt better, and he was as jolly as any of the six little
Bunkers when they rode out to the field on the empty wagon. Once
the field was reached there was a busy time. There was little the
children could do, for loading hay is hard work, fit only for big, strong
men.
But Russ, Rose and the others watched Adam, Farmer Joel, their
father, and the two hired men dig their shiny pitchforks deep into a
hay cock. Sometimes two men, each with a fork, would lift almost a
whole cock up on the wagon at once. When one man did it alone he
took about half the cock at a time.
As the hay was loaded on the wagon, which was fitted with a rick,
going over the wheels, the pile of dried grass on the vehicle became
higher and higher. So high it was, at last, that the men could hardly
pitch hay up on it.
 “I guess we’ll call this a load,” said Farmer Joel, as he looked up at
the sky. “The road is a bit rough and if we put on too much we’ll have
an upset. Adam, I think you’re right,” he went on. “We’ll have thunder
showers this afternoon. Have to hustle, boys, to get the hay in!”
When the horses were ready to haul the first load back to the barn,
to be stored away in the mow, the six little Bunkers were put up on
top of the load to ride.
“Oh, this is lovely!” cried Rose.
“Like being on a hundred feather beds!” added Russ.
“And you don’t feel the jounces at all!” added Laddie, for as the
wagon went over rough places in the field the children were only
gently bounced up and down, and not shaken about as they would
have been had there been no hay on the wagon.
But the rough field caused one little accident which, however,
harmed no one.
The first load of hay was almost out of the field when, as it
approached the bars, Mun Bun suddenly yelled:
“I’m slippin’! I’m slippin’!”
And Margy followed with a like cry.
“Oh, I’m fallin’ off!” she shouted.
And, surely enough, Russ and Rose also felt the top of the load of
hay beginning to slip to one side. Adam North was riding with the
children, Farmer Joel and Daddy Bunker having remained in the
field, while one hired man drove the team of horses.
“I guess they didn’t load this hay evenly,” said Adam. “Part of it is
going to slip off. But don’t be frightened, children,” he said kindly.
“You can’t get hurt falling with a load of hay.”
Just as Adam finished speaking part of the top of the load slid off
the wagon and fell into the field, and with it fell the six little Bunkers
and Adam himself.
“Oh! Oh!” screamed Margy and Mun Bun.
 “Keep still!” ordered Russ. “You won’t get hurt!”
“Look out for the pitchforks—they’re sharp!” warned Rose.
Laddie and Violet laughed with glee as they felt themselves
sliding.
Down in a heap went the hay, the six little Bunkers and Adam. The
hay was so soft it was like falling in a bed of feathers. The man
sitting in front to drive the horses did not slide off.
“All over! No damage!” cried Adam, with a laugh, as he leaped up
and picked the smallest of the little Bunkers from the pile of hay. “But
we’ll have to load the hay back on the wagon.”
This was soon done, and once more the merry party started for
the barn, which was reached without further accident.
Farmer Joel had many things on his place to save work. Among
these was a hay fork which could pick up almost half a load of hay at
once and raise it to the mow.
A hay fork, at least one kind, looks like a big letter U turned upside
down. The two arms are made of iron, and from their lower ends
prongs come out to hold the hay from slipping off the arms.
A rope, running through a pulley is fastened to the curved part of
the U, and a horse, pulling on the ground end of the rope, hoists into
the air a big mass of hay.
The wagon was driven under the high barn window and from a
beam overhead the hay fork was lowered. Adam North plunged the
two sharp arms deep into the springy, dried grass.
All but Russ had gotten down off the load of hay to wait for the ride
back to the field. But Russ remained there. He wanted to see how
the hay fork worked.
So when Adam plunged the arms into the fodder Russ was near
by. Adam pulled on the handle that shot the prongs out from the
arms to hold the hay from slipping off as the fork was raised.
Then, suddenly, Russ did a daring thing. Seeing the mass of hay
rising in the air, pulled by the horse on the ground below, the boy
made a grab for the bunch of dried grass. He caught it, clung to it
and up in the air he went, on an exciting and dangerous ride.
“Oh, look at Russ! Look at Russ!” cried Rose.
“Hi there, youngster, what are you doing?” shouted Adam.
“I—I’m getting a ride!” Russ answered. But his voice had a
frightened tone in it as he swung about and looked down below. He
began to feel dizzy.
 CHAPTER XVI
OFF ON A PICNIC

While Russ swung to and fro in the mass of hay lifted by the hay
fork and was kept over the load itself there was little danger. If he fell
he would land on the hay in the wagon.
But the hay fork had to swing to one side, when high up in the air,
so the hay could be placed in the window opening into the storage
mow. And it was this part of Russ’s ride that was dangerous.
The man on the ground, who had charge of the horse that was
hitched to the pulley rope, knew nothing of what was going on above
him, for the load of hay was so large that it hid Russ and the fork
from sight. But this man heard the shout of Adam, and he called up:
“Is anything the matter?”
“No! No!” quickly answered Adam, for he feared if the horse
stopped the shock might throw Russ from his hold. “Keep on, Jake!”
he called to the hired man. “You’ll have to hoist a boy up as well as a
fork full of hay. Hold on tight there, Russ!” Adam warned the Bunker
lad.
“I will,” Russ answered. He was beginning to wish that he had not
taken this dangerous ride. It was done on the impulse of the
moment. He had seen the mass of hay being lifted with the fork and
he felt a desire to go up with it—to get a ride in the air. So he made a
grab almost before he thought.
Up and up went the fork full of hay with Russ on it. Now he was
swung out and away from the wagon, and was directly over the bare
ground, thirty or forty feet below. In the barn window of the mow
overhead a man looked out.
“What’s this you’re sending me?” called this man to Adam.
 “It’s Russ! Grab him when he gets near enough to you,” Adam
answered.
“I will,” said the man who was “mowing away,” as the work of
storing the hay in the barn is called.
“NOW WATCH HER WHIZZ!” CRIED RUSS.
 Six Little Bunkers at Farmer Joel’s. (Page 160)

Higher and higher up went Russ, while Rose and the other little
Bunkers on the ground below gazed at him in mingled fright and
envy.
“Will he fall and be killed?” asked Vi.
“No, I guess not. Oh, no! Of course not!” exclaimed Rose.
A moment later the fork load of hay with Russ clinging to it, one
hand on the lifting rope, swung within reach of the man in the mow
window. Russ was caught, pulled inside to safety, and as he sank
down on the pile of hay within the barn the man said:
“You’d better not do that again!”
“I won’t!” promised Russ, with a little shiver of fear and excitement.
Rose and the other children breathed more easily now, and Adam
North, wiping the sweat from his forehead, murmured:
“You never know what these youngsters are going to do next!”
Back to the hay field went the empty wagon, the six little Bunkers
riding on it. The trip back was not as comfortable as the one on the
load of hay had been. For the wagon was rickety and the road was
rough and jolty. But the six little Bunkers had a jolly time, just the
same.
The men were working fast now, and Daddy Bunker was helping
them, for dark clouds in the west and distant muttering of thunder
seemed to tell of a coming storm, and Farmer Joel did not want his
hay to get wet.
Another big load was taken to the barn, no upset happening this
time. And you may be sure Adam made certain that Russ did not
cling to the hay fork.
After three loads had been put away most of the hay was in.
Scattered about the field, however, were little piles and wisps of the
fodder—perhaps half a load in all—and this must be raked up by the
big horse rake.
“Oh, may I have a ride?” cried Laddie, when he saw the machine
being brought out from a corner of the rail fence where it had been
standing.
“Yes, I’ll give you each a ride in turn,” kindly offered Adam North,
who was to drive the horse hitched to the big rake. And as Laddie
had asked first he was given the first ride, sitting on the seat beside
Adam.
The curved iron teeth of the rake gathered up a mass of hay until
they could hold no more. Then Adam “tripped” it, as the operation is
called. The teeth rose in the air and passed over the mass of hay
which was left on the ground.
Working in this way, more hay was raked up until there were
several windrows and cocks to be loaded upon the wagon. As a
special favor Russ and Rose were allowed to pitch small forkfuls of
the hay on the wagon. And when all the dried grass had been
gathered up the children piled on and rode to the barn for the last
time.
“Hurray! Hurray! Hurray for the hay!” they sang most merrily.
“And it’s a good thing we got it in to-day,” said Farmer Joel, with a
chuckle, as the last forkful was raised to the mow. “For here comes
the rain!”
And down pelted the big drops. There was not much thunder and
lightning, but the rain was very hard and the storm pelted and
rumbled all night.
“It’s a good thing I got in my hay,” said Farmer Joel, as he went to
bed that night. “Now I can sleep in peace.”
For there is nothing more worrying to a farmer than to hear it rain,
knowing it is spoiling his hay. Hay, once wet, is never quite so good
as that which has not been soaked.
Though it rained all night, the sun came out the next day, and the
six little Bunkers could play about and have fun. Russ and Laddie
were glad of the storm, for the rain had made the brook higher, and
water was now for the first time running over the little dam they had
made so their water wheel could be turned.
“She’ll splash like anything now!” cried Laddie, as he and his
brother hastened down to the brook.
The water wheel was made of some flat pieces of wood fastened
together and set in a frame work. The water, spouting over the dam,
fell on the blades of the paddle wheel and turned it. On the axle of
the wheel was a small, round pulley, and around this there was a
string, or a belt, running to a small mill that the boys had made. It
had taken them quite a while to do this.
“Now watch her whizz!” cried Russ to his brothers and sisters, who
had gathered on the bank of the brook.
The water wheel was shoved back so the overflow from the dam
would strike the paddles. Around they went, turning the pulley,
moving the string belt, and also turning the wheel of the “mill.”
“Oh, isn’t that fine!” exclaimed Rose.
“Could I have a ride on it?” Mun Bun wanted to know.
“Hardly!” laughed Russ. “If you sat on it the wheel would break.”
“And you’d get all wet!” added Rose.
The six little Bunkers had much fun that day, and more good times
were ahead of them, for that evening when they made ready for bed,
tired but happy, their mother said:
“To-morrow we are going on a picnic to the woods.”
“A really, truly picnic?” Vi wanted to know.
“Of course.”
“With things to eat?” asked Russ.
“Surely,” said his mother. “Now off to bed with you! Up early, and
we’ll have a fine picnic in the woods.”
 You may be sure that not one of the six little Bunkers overslept the
next day. Bright and early they were up, and soon they started for
the picnic grounds in the big hay wagon, on which some straw had
been scattered to make soft seats.
“I wonder if anything will happen to-day?” said Rose to Russ, as
they rode off with their lunches.
“What do you mean?” he inquired.
“I mean anything like an adventure.”
“Oh, maybe we’ll find a—snake!” and Russ laughed as he saw his
sister jump, for Rose did not like snakes.
“You’re a horrid boy!” she murmured.
But an adventure quite different from finding a snake happened to
the six little Bunkers.
 CHAPTER XVII
THE ICE CAVE

Along the road, through pleasant fields, and into the woods
rumbled the big farm hay wagon, driven by Adam North. In the
wagon sat the six little Bunkers with their father and mother and
Farmer Joel. For Farmer Joel had decided that, after the haymaking,
he was entitled to a holiday. So he stopped work and went on the
picnic with the six little Bunkers.
“How much farther is it to the picnic grounds?” asked Vi, after they
had ridden for perhaps half an hour.
“Not very far now,” answered Farmer Joel.
“Is it a nice picnic grounds?” went on the little girl who always
asked questions. “And is there——”
“Now, Vi,” interrupted her mother, “suppose you wait until we get
there and you can see what there is to see. You mustn’t tire Farmer
Joel by asking so many questions.”
“Well, I only wanted to ask just one thing more,” begged Vi.
“Go ahead. What is it?” chuckled the good-natured farmer.
“Is there a swing in the picnic woods?” asked Vi, after a moment’s
pause to decide which question was the most important.
“Well, if there isn’t we can put one up, for I brought a rope along,”
answered Adam North. He liked to see the six little Bunkers have fun
as much as the children loved to play.
“Oh, a swing! Goodie!” cried Violet.
“I want to swing in it!” exclaimed Mun Bun.
“So do I!” added Margy.
 “I can see where there’s going to be trouble, with only one swing,”
murmured Daddy Bunker, smiling at his wife.
“Oh, they can take turns,” she said.
The wagon was now going through the woods. On either side
were green trees with low-hanging branches, some of which met in
an arch overhead, drooping down so far that the children could reach
up and touch the leaves with their hands.
“Oh, it’s just lovely here,” murmured Rose, who liked beautiful
scenery.
“I see something that’s lovelier,” said Russ.
“What?” asked Rose. “I don’t see anything. You can’t get much of
a view down here under the trees, but it’s beautiful just the same.”
“Here’s the view I was looking at,” said Russ, with a laugh, and he
pointed to the piles of lunch boxes and baskets in the front part of
the hay wagon. “That’s a better view than just trees, Rose.”
“Oh, you funny boy!” she laughed. “Always thinking of something
to eat! Don’t you ever think of something else?”
“Yes, right after I’ve had something to eat I think of when it’s going
to be time to eat again,” chuckled Russ.
Deeper into the woods went the picnic wagon. The six little
Bunkers were talking and laughing among themselves, and Farmer
Joel was speaking to Mr. and Mrs. Bunker and Adam about
something that had happened in the village that day.
Suddenly there was a cry from the children, who were in the rear
of the wagon, sprawled about in the straw.
“Laddie’s gone!” exclaimed Rose.
“Did he fall out?” asked Mrs. Bunker.
“No, it looks more as if he fell up!” shouted Russ.
And that, indeed, is almost what happened. For, looking back, Mr.
and Mrs. Bunker, saw Laddie hanging by his hands to the branch of
a tree he had grasped as the wagon passed beneath it. The little
fellow was swinging over the roadway, the wagon having passed
from beneath him.
“Hold on, Laddie! I’ll come back and get you!” shouted Mr. Bunker.
“In mischief again!” murmured Russ.
“Whoa!” called Adam, bringing the horses to a halt.
“Hold on, Laddie! I’ll come and get you!” called Mr. Bunker again,
as he leaped from the hay wagon.
“I—I can’t hold on!” gasped Laddie. “My—my hands are slipping!”
The green branch was slowly bending over and Laddie’s hands
were slipping from it. Then, when he could keep his grasp no longer,
he let go, and down to the ground he fell, feet first.
Luckily Laddie was only a short distance above the ground when
he slipped from the branch, so he did not have far to fall. He was
only jarred and shaken a little bit—not hurt at all.
“Laddie, why did you do that?” his father asked him, when he had
reached the little fellow and picked him up. As Mr. Bunker carried
Laddie back to the waiting wagon Russ remarked:
“I guess he thought maybe he could pull a tree up by the roots
when he caught hold of the branch like that.”
“I did not!” exclaimed Laddie. “I just wanted to pull off a whip for
Mun Bun to play horse with. But when I got hold of the branch I
forgot to let go and it lifted me right out of the wagon.”
“It’s a mercy you weren’t hurt!” exclaimed his mother.
“I should say so!” added Farmer Joel. “It’s safer for you to think up
riddles, Laddie, than it is to do such tricks as that. Come now, sit
quietly in the wagon and think of a riddle.”
“All right,” agreed Laddie, as again he took his place in the straw
with the other little Bunkers. But he did not ask any riddles for a long
time. Perhaps he had been too startled. For surely it was rather a
startling thing to find himself dangling on a tree branch, the wagon
having gone out from under him.
 However, in about fifteen minutes more Laddie suddenly cried:
“Oh, now I know a riddle! Why is a basket——”
But before he could say any more the other children broke into
cries of:
“There’s the picnic ground! There’s the picnic ground!”
And, surely enough, they had reached the grove in the woods
where lunch was to be eaten and games played.
It was a beautiful day of sunshine, warm and pleasant. Too warm,
in fact, for Mrs. Bunker had to call to the children several times:
“Don’t run around too much and get overheated. It is very warm,
and seems to be getting warmer.”
“Yes,” agreed Farmer Joel, as he looked at the sky. “I think we’ll
have a thunder shower before the day is over.”
“We didn’t bring any umbrellas,” said Mrs. Bunker.
“If it rains very hard we can take shelter in a cave not far from here
that I know of,” said Mr. Todd.
“Oh, a cave! Where is it?” asked Russ, who was lying down in the
shade, having helped put up the swing. “Could we go and see it?” he
inquired.
“After a while, maybe,” promised Farmer Joel.
Rose helped her mother spread out the good things to eat. They
found some flat stumps which answered very well for tables, and
after Mun Bun and Margy and Laddie and Violet had swung as much
as was good for them, and when they had raced about, playing tag,
hide-and-seek, and other games, the children were tired enough to
sit down in the shade.
“We’ll eat lunch after you rest a bit,” said Mrs. Bunker.
“Ah, now comes the best part of the day!” murmured Russ.
“Silly! Always thinking of something to eat!” chided Rose. But she
smiled pleasantly at her brother.
 How good the things eaten in the picnic woods tasted! Even plain
bread and butter was almost as fine as cake, Laddie said. He was
trying to think of a riddle about this—a riddle in which he was to ask
when it was that bread and butter was as good as cake—when
suddenly there came a low rumbling sound.
“What’s that?” asked Margy.
“Thunder, I think,” was the answer.
Mun Bun, who was playing a little distance away, came running in.
“I saw it lighten,” he whispered.
“Yes, I think we’re in for a storm,” said Farmer Joel.
The thunder became louder. The sun was hidden behind dark
clouds. The picnic things were picked up. Mrs. Bunker was glad
lunch was over.
Then down pelted the rain.
“Come on!” cried Farmer Joel. “We’ll take shelter in the cave!”
He led the way along a path through the woods. The others
followed, Mr. Bunker carrying Mun Bun and Adam North catching up
Margy. The trees were so thick overhead that not much rain fell on
the picnic party.
“Here’s the cave!” cried Farmer Joel, pushing aside some bushes.
He showed a dark opening among some rocks. In they rushed, for it
was a welcome shelter from the storm.
“Oh, but how cool it is in here,” said Rose.
“Yes,” answered Farmer Joel. “This is an ice cave.”
“An ice cave!” exclaimed Russ. “Is there really ice in here in the
middle of summer?”
Before Farmer Joel could answer a terrific crash of thunder
seemed to shake the whole earth.
 CHAPTER XVIII
A BIG SPLASH

There was silence in the dark cave of ice following that big noise
from the sky. Then came a steady roar of sound.
“It’s raining cats and dogs outside,” said Farmer Joel. “We got here
just in time.”
Suddenly Margy began to whimper and then she began to cry.
“What’s the matter, my dear?” asked her mother.
“I—I don’t like it in here!” sobbed Margy.
“I—I don’t, either, an’—an’ I’m goin’ to cry, too!” snuffled Mun Bun.
“Oh, come, children!” exclaimed Mr. Bunker, with a laugh. “Don’t
be babies! Why don’t you like it in here?”
“I—I’m ’fraid maybe we’ll be struck by lightning,” whimpered
Margy.
“Oh, nonsense!” replied her mother.
“No lightning ever comes in here,” said Farmer Joel. “Why, if
lightning came in there couldn’t be any ice. The lightning would melt
the ice, and it hasn’t done that. I’ll show you a big pile of it back in
the cave. Of course no lightning ever comes in here! Don’t be afraid.”
The thunder was not so loud now, and as no lightning could be
seen because the Bunkers were far back in the dark cave, the two
smallest children stopped their crying.
“Is there really ice in here?” asked Russ.
“It feels so,” said Rose, with a little shiver.
“Yes, there’s ice here,” went on the farmer. “It comes every year,
and stays until after the Fourth of July. Come, I’ll show you.”