Working With Dynamical Systems: A Toolbox For Scientists and Engineers (Series in Computational Biophysics) 1st Edition Len Pismen

Working with Dynamical Systems: A
Toolbox for Scientists and Engineers

(Series in Computational Biophysics)
1st Edition Len Pismen
Visit to download the full and correct content document:
https://ebookmeta.com/product/working-with-dynamical-systems-a-toolbox-for-scienti
sts-and-engineers-series-in-computational-biophysics-1st-edition-len-pismen/
More products digital (pdf, epub, mobi) instant
download maybe you interests ...
Patterns and Interfaces in Dissipative Dynamics 2nd

Edition Len Pismen
https://ebookmeta.com/product/patterns-and-interfaces-in-
dissipative-dynamics-2nd-edition-len-pismen/
Intelligent Systems for Engineers and Scientists: A

Practical Guide to Artificial Intelligence, 4th Edition
Adrian A. Hopgood
https://ebookmeta.com/product/intelligent-systems-for-engineers-
and-scientists-a-practical-guide-to-artificial-intelligence-4th-
edition-adrian-a-hopgood/
Applied Numerical Methods with Python for Engineers and

Scientists 1st Edition Steven Chapra
https://ebookmeta.com/product/applied-numerical-methods-with-
python-for-engineers-and-scientists-1st-edition-steven-chapra/
Applied Numerical Methods with MATLAB for Engineers and

Scientists, 5th Edition Steven Chapra
https://ebookmeta.com/product/applied-numerical-methods-with-
matlab-for-engineers-and-scientists-5th-edition-steven-chapra/
Feedback Systems: An Introduction for Scientists and
Engineers Second Edition Karl Johan Åström
https://ebookmeta.com/product/feedback-systems-an-introduction-
for-scientists-and-engineers-second-edition-karl-johan-astrom/
Machine Learning Toolbox for Social Scientists 1st

Edition Yigit Aydede
https://ebookmeta.com/product/machine-learning-toolbox-for-
social-scientists-1st-edition-yigit-aydede/
Essential Mathematics for Engineers and Scientists 1st

Edition Pence
https://ebookmeta.com/product/essential-mathematics-for-
engineers-and-scientists-1st-edition-pence/
Machine Learning with Neural Networks: An Introduction

for Scientists and Engineers Bernhard Mehlig
https://ebookmeta.com/product/machine-learning-with-neural-
networks-an-introduction-for-scientists-and-engineers-bernhard-
mehlig/
Advanced Mathematics for Engineers and Scientists

Duchateau Paul
https://ebookmeta.com/product/advanced-mathematics-for-engineers-
and-scientists-duchateau-paul/
Working with Dynamical
Systems
Series in Computational Biophysics
Series Editor: Nikolay Dokholyan
Molecular Modeling at the Atomic Scale

Methods and Applications in Quantitative Biology
Ruhong Zhou
Coarse-Grained Modeling of Biomolecules

Garegin A. Papoian
Computational Approaches to Protein Dynamics

From Quantum to Coarse-Grained Methods
Monika Fuxreiter
Modeling the 3D Conformation of Genomes

Guido Tiana, Luca Giorgetti
Biomembrane Simulations
Computational Studies of Biological Membranes
by Max L. Berkowitz
Working with Dynamical Systems

A Toolbox for Scientists and Engineers
LM Pismen
For more information about this series, please visit:

[www.crcpress.com/Series-in-Computational-Biophysics/book-series/CRCSERCOMBIO]
Working with Dynamical
Systems
A Toolbox for Scientists and Engineers
Professor LM Pismen
Department of Chemical Engineering, Technion - Israel Institute of Technology
First edition published 2021
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
© 2021 Taylor & Francis Group, LLC
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 U.S. 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: 9781138591714 (hbk)

ISBN: 9780429488856 (ebk)
Typeset in Computer Modern font

by Cenveo Publisher Services
Visit the [companion website/eResources]: [www.routledge.com/9781138591714]

To my daughter Anne,
my best success in teaching math
Contents
Preface xi
1 Whence Dynamical Systems 1

1.1 Classical Mechanics . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
1.1.1 Conservative Equations of Motion . . . . . . . . . . . . . . . 2
1.1.2 Systems with One Degree of Freedom . . . . . . . . . . . . . 4
1.1.3 Symmetries and Conservation Laws . . . . . . . . . . . . . . 6
1.1.4 Interacting Particles . . . . . . . . . . . . . . . . . . . . . . . 8
1.1.5 Dissipative Motion . . . . . . . . . . . . . . . . . . . . . . . 9
1.2 Chemical Kinetics . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
1.2.1 Mass Action . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
1.2.2 Adsorption and Catalysis . . . . . . . . . . . . . . . . . . . . 13
1.2.3 Autocatalysis and Self-Inhibition . . . . . . . . . . . . . . . . 16
1.2.4 Thermal Effects . . . . . . . . . . . . . . . . . . . . . . . . . . 16
1.3 Biological Models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
1.3.1 Population Dynamics . . . . . . . . . . . . . . . . . . . . . . 18
1.3.2 Epidemiological Models . . . . . . . . . . . . . . . . . . . . . 20
1.3.3 Neural and Genetic Networks . . . . . . . . . . . . . . . . . 21
1.4 Electric Currents . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
1.4.1 Electric Circuits . . . . . . . . . . . . . . . . . . . . . . . . . 22
1.4.2 Electrochemical Reactions . . . . . . . . . . . . . . . . . . . . 24
1.4.3 Membrane Transport . . . . . . . . . . . . . . . . . . . . . . . 25
1.5 Spatially Extended Systems . . . . . . . . . . . . . . . . . . . . . . 26
1.5.1 From Time to Coordinate Dependence . . . . . . . . . . . . 26
1.5.2 Fourier Decomposition . . . . . . . . . . . . . . . . . . . . . . 27
1.6 Continuous vs. Discrete . . . . . . . . . . . . . . . . . . . . . . . . . 29
1.6.1 Iterative Systems . . . . . . . . . . . . . . . . . . . . . . . . 29
1.6.2 From Continuous to Discrete . . . . . . . . . . . . . . . . . . 30
1.6.3 Poincaré Maps . . . . . . . . . . . . . . . . . . . . . . . . . . 31
vii
viii CONTENTS
2 Local Bifurcations 33
2.1 Bifurcation of Stationary States . . . . . . . . . . . . . . . . . . . . 34
2.1.1 Branches of Stationary States . . . . . . . . . . . . . . . . . 34
2.1.2 Bifurcation Expansion . . . . . . . . . . . . . . . . . . . . . . 35
2.1.3 Fold and Transcritical Bifurcations . . . . . . . . . . . . . . 37
2.1.4 Cusp Singularity . . . . . . . . . . . . . . . . . . . . . . . . . 38
2.1.5 Higher Singularities . . . . . . . . . . . . . . . . . . . . . . . 40
2.2 Stability and Slow Dynamics . . . . . . . . . . . . . . . . . . . . . . 41
2.2.1 Linear Stability Analysis . . . . . . . . . . . . . . . . . . . . 41
2.2.2 Stable and Unstable Manifolds . . . . . . . . . . . . . . . . . 44
2.2.3 Exchange of Stability . . . . . . . . . . . . . . . . . . . . . . 45
2.2.4 Amplitude Equations . . . . . . . . . . . . . . . . . . . . . . 47
2.3 Bifurcations of Periodic Orbits . . . . . . . . . . . . . . . . . . . . . 49
2.3.1 Hopf Bifurcation . . . . . . . . . . . . . . . . . . . . . . . . . 49
2.3.2 Derivation of the Amplitude Equation . . . . . . . . . . . . . 52
2.3.3 Instabilities of Periodic Orbits . . . . . . . . . . . . . . . . . 54
2.4 Example: Exothermic Reaction . . . . . . . . . . . . . . . . . . . . . 56
2.4.1 Bifurcation of Stationary States . . . . . . . . . . . . . . . . 56
2.4.2 Hopf Bifurcation . . . . . . . . . . . . . . . . . . . . . . . . . 58
2.4.3 Branches of Periodic Orbits . . . . . . . . . . . . . . . . . . 61
2.5 Example: Population Dynamics . . . . . . . . . . . . . . . . . . . . 62
2.5.1 Prey–Predator Models . . . . . . . . . . . . . . . . . . . . . 62
2.5.2 Stability and Bifurcations . . . . . . . . . . . . . . . . . . . . 64
2.5.3 Periodic Orbits . . . . . . . . . . . . . . . . . . . . . . . . . 65
3 Global Bifurcations 69
3.1 Topology of Bifurcations . . . . . . . . . . . . . . . . . . . . . . . . 70
3.1.1 More Ways to Create and Break Periodic Orbits . . . . . . . 70
3.1.2 Bifurcations in a System with Three Stationary States . . . 72
3.2 Global Bifurcations in the Exothermic Reaction . . . . . . . . . . . . 74
3.2.1 Basin Boundaries . . . . . . . . . . . . . . . . . . . . . . . . . 74
3.2.2 Saddle-Loop Bifurcations . . . . . . . . . . . . . . . . . . . . 75
3.2.3 Sniper Bifurcation . . . . . . . . . . . . . . . . . . . . . . . . 78
3.3 Bifurcation at Double-Zero Eigenvalue . . . . . . . . . . . . . . . . . 80
3.3.1 Locating a Double Zero . . . . . . . . . . . . . . . . . . . . . 80
3.3.2 Quadratic Normal Form . . . . . . . . . . . . . . . . . . . . 81
3.3.3 Expansion in the Vicinity of Cusp Singularity . . . . . . . . 82
3.4 Almost Hamiltonian Dynamics . . . . . . . . . . . . . . . . . . . . . 85
3.4.1 Weak Dissipation . . . . . . . . . . . . . . . . . . . . . . . . 85
3.4.2 Hopf and Saddle-Loop Bifurcations . . . . . . . . . . . . . . 87
3.4.3 Bifurcation Diagrams . . . . . . . . . . . . . . . . . . . . . . 88
3.4.4 Basin Boundaries . . . . . . . . . . . . . . . . . . . . . . . . 91
CONTENTS ix
3.5 Systems with Separated Time Scales . . . . . . . . . . . . . . . . . . 92

3.5.1 Fast and Slow Variables . . . . . . . . . . . . . . . . . . . . . 92
3.5.2 Van der Pol Oscillator . . . . . . . . . . . . . . . . . . . . . . 93
3.5.3 FitzHugh–Nagumo Equation . . . . . . . . . . . . . . . . . . 94
3.5.4 Canards . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 96
3.6 Venturing to Higher Dimensions . . . . . . . . . . . . . . . . . . . . 97
3.6.1 Dynamics Near Triple-Zero Eigenvalue . . . . . . . . . . . . 97
3.6.2 Double Hopf Bifurcation . . . . . . . . . . . . . . . . . . . . 100
3.6.3 Blue Sky Catastrophe . . . . . . . . . . . . . . . . . . . . . . 101
4 Chaotic, Forced, and Coupled Oscillators 103

4.1 Approaches to Hamiltonian Chaos . . . . . . . . . . . . . . . . . . . 104
4.1.1 Hiding in Plain Sight . . . . . . . . . . . . . . . . . . . . . . . 104
4.1.2 Resonances and Small Divisors . . . . . . . . . . . . . . . . . 106
4.1.3 Example: Hénon-Heiles Model . . . . . . . . . . . . . . . . . 108
4.1.4 Quantitative Measures of Chaos . . . . . . . . . . . . . . . . 111
4.2 Approaches to Dissipative Chaos . . . . . . . . . . . . . . . . . . . . 114
4.2.1 Distilling Turbulence into Simple Models . . . . . . . . . . . 114
4.2.2 Chaotic Attractors . . . . . . . . . . . . . . . . . . . . . . . . 116
4.2.3 Period-Doubling Cascade . . . . . . . . . . . . . . . . . . . . 117
4.2.4 Strange, Chaotic, or Both? . . . . . . . . . . . . . . . . . . . 120
4.3 Chaos Near a Homoclinic . . . . . . . . . . . . . . . . . . . . . . . . 123
4.3.1 Shilnikov’s Snake . . . . . . . . . . . . . . . . . . . . . . . . 123
4.3.2 Complexity in Chaotic Models . . . . . . . . . . . . . . . . . 125
4.3.3 Lorenz Model . . . . . . . . . . . . . . . . . . . . . . . . . . 129
4.4 Weakly Forced Oscillators . . . . . . . . . . . . . . . . . . . . . . . . 136
4.4.1 Phase Perturbations . . . . . . . . . . . . . . . . . . . . . . . 136
4.4.2 Forced Harmonic Oscillator . . . . . . . . . . . . . . . . . . . 138
4.4.3 Weakly Forced Hamiltonian System . . . . . . . . . . . . . . 140
4.5 Effects of Strong Forcing . . . . . . . . . . . . . . . . . . . . . . . . 142
4.5.1 Universal and Standard Mappings . . . . . . . . . . . . . . . 142
4.5.2 Forced Dissipative Oscillators . . . . . . . . . . . . . . . . . 145
4.5.3 Forced Relaxation Oscillator . . . . . . . . . . . . . . . . . . 147
4.6 Coupled Oscillators . . . . . . . . . . . . . . . . . . . . . . . . . . . 152
4.6.1 Phase Dynamics . . . . . . . . . . . . . . . . . . . . . . . . . 152
4.6.2 Coupled Pendulums . . . . . . . . . . . . . . . . . . . . . . . 153
4.6.3 Coupled Relaxation Oscillators . . . . . . . . . . . . . . . . . 156
4.6.4 Synchronization in Large Ensembles . . . . . . . . . . . . . . 158
x CONTENTS
5 Dynamical Systems in Space 163

5.1 Space-Dependent Equilibria . . . . . . . . . . . . . . . . . . . . . . . 163
5.1.1 Basic Equations . . . . . . . . . . . . . . . . . . . . . . . . . 163
5.1.2 Stationary Solution in One Dimension . . . . . . . . . . . . . 165
5.1.3 Systems with Mass Conservation . . . . . . . . . . . . . . . . 168
5.2 Propagating Fronts . . . . . . . . . . . . . . . . . . . . . . . . . . . . 170
5.2.1 Advance into a Metastable State . . . . . . . . . . . . . . . . 170
5.2.2 Propagation into an Unstable State . . . . . . . . . . . . . . 175
5.2.3 Pushed Fronts . . . . . . . . . . . . . . . . . . . . . . . . . . 178
5.3 Separated Time and Length Scales . . . . . . . . . . . . . . . . . . . 180
5.3.1 Two-Component Reaction–Diffusion System . . . . . . . . . 180
5.3.2 Stationary and Mobile Fronts . . . . . . . . . . . . . . . . . 183
5.3.3 Stationary and Mobile Bands . . . . . . . . . . . . . . . . . . 187
5.3.4 Wave Trains . . . . . . . . . . . . . . . . . . . . . . . . . . . 192
5.4 Symmetry-Breaking Bifurcations . . . . . . . . . . . . . . . . . . . . 198
5.4.1 Amplitude Equations . . . . . . . . . . . . . . . . . . . . . . 198
5.4.2 Bifurcation Expansion . . . . . . . . . . . . . . . . . . . . . 200
5.4.3 Interacting Modes . . . . . . . . . . . . . . . . . . . . . . . . 202
5.4.4 Plane Waves and their Stability . . . . . . . . . . . . . . . . . 204
5.5 Resonant Interactions . . . . . . . . . . . . . . . . . . . . . . . . . . 207
5.5.1 Triplet Resonance . . . . . . . . . . . . . . . . . . . . . . . . 207
5.5.2 Stripes–Hexagons Competition . . . . . . . . . . . . . . . . . 210
5.5.3 Standing Waves . . . . . . . . . . . . . . . . . . . . . . . . . 212
5.6 Nonuniform Patterns . . . . . . . . . . . . . . . . . . . . . . . . . . . 215
5.6.1 Propagation of a Stationary Pattern . . . . . . . . . . . . . . 215
5.6.2 Self-Induced Pinning . . . . . . . . . . . . . . . . . . . . . . . 218
5.6.3 Propagating Wave Pattern . . . . . . . . . . . . . . . . . . . 221
5.6.4 Nonuniform Wave Patterns . . . . . . . . . . . . . . . . . . . 223
Bibliography 227
Online Files 233
Illustration Credits 235

Preface
Why Dynamical Systems?
Things change in time. They tend to change gradually and continuously, though
sometimes an abrupt transition occurs. Ordinary differential equations (ODE) are
natural models of gradual change. One starts with assigning a set of variables. It
should be possible to define them in such a way that the rate of change at any moment
is determined in a unique way by their instantaneous values. Then the evolution is
deterministic, and future states of the system can be predicted by integrating these
equations starting from known initial conditions.
However, it is not that simple in the real world. Any model containing a finite num-
ber of variables is imperfect, and no values can ever be measured precisely. Therefore
“solving” equations is not enough.
Dynamical systems may exhibit complex temporal behavior, and sometimes, even
though evolution remains deterministic, unreachable precision is required to predict
the outcome of a disorderly process. Dynamical systems also serve as a tool for
understanding still more complex spatially distributed systems when a few dominant
modes are sufficient for adequate approximate description of spatio-temporal dynamics
or when a single spatial coordinate plays the role of time.
People may be interested in dynamical systems for different reasons. For the
mathematician, it is a logical arena and a way to look into the abyss of chaos. For
the physicist, it is a station on the way of inquiry into ever more complex dynamics.
The engineer may pursue the opposite aim of taming the system and eliminating all
dangerous quirks of unruly nonlinearities and feedback loops.
We do not look at a dynamical system as an abstract object. The first chapter of
this book emphasizes the origins of the systems we study. We wish to see common
features in systems of different origins, and to apply common methods of study without
losing sight of possible complications and uncertainties related to their physical origin.
Why This Book?

Do we need another book on dynamical systems? Hasn’t it been thoroughly explored
and explained by the great mathematicians of the last century, from Henri Poincaré
to Vladimir Arnold? Have not the books on the topic filled long meters of shelves in
xi
xii PREFACE
academic libraries, now thinned out by the librarians to clear space for cozy nooks
where the students play with their smartphones and laptops, while the faculty browse
arxiv.org on huge desktop screens in their offices?
I embarked on writing this book nevertheless, with some pushing by the editors.
Nonlinearities and bifurcations have accompanied me throughout my variegated aca-
demic career, from chemical reactors to fluids to topological singularities and back to
wetting and soft matter. Dynamical systems is an old and well-studied topic. This
book is essentially a 20th-century book spiced by 21st-century computer graphics. The
20th century was the summit of analytical insight into complex dynamics, but it has
not yet penetrated undergraduate curricula, still discussing paradoxes of infinitesimals
that perplexed ancient Greeks, and advancing only as far as linear equations of the
18th century. This leaves the students to play blindly with 21st century computers
when encountering nonlinear problems of real life.
I am trying here to present advanced classical tools as a template for exploring the
complexity hidden in run-of-the-mill physical and engineering models briefly reviewed
in the first chapter. In the following two chapters, the analysis of nonlinear dynamical
systems, with their plethora of local and global bifurcations among stationary and
periodic attractors and repellers, is carried out, in a way that the mathematicians may
characterize as formal, helped by the classical tools of perturbations, multiscaling, and
asymptotic matching, before turning to numerics as the last resort. Chaos makes an
appearance in the fourth chapter, where dynamics of forced and interacting oscillators
is discussed as well. The last chapter is devoted to extending the methods of dynamical
systems to problems involving spatial patterns and waves. This is a door to a much
wider world of dynamical patterns, which is beyond the scope of this book.
The presentation style is not suitable for lovers of theorems. I am trying to use
plain English as far as it is possible in a text interspersed with formulas. The material
develops from elementary to more advanced, but we have to stop when encountering
more complicated and specialized topics. The accompanying Mathematica notebooks
demonstrating the application of general methods to specific, apparently simple but
hiding inherent complexity, problems may serve both as illustration of general deriva-
tions and as templates for quantitative analysis of similar systems by the devoted
reader.
Rules of Engagement
We intend to study phenomena in their simplest form, using dimensionless variables
and parameters to get wider applicable results with least clutter. The material of
this book is supposed to be self-contained. The reader should be able to reproduce
all computations rather than rely on an outside authority. After some doubts, I
decided against including elementary exercises usually given in textbooks, which are
mostly helpful to teachers of elementary courses for assigning tests but otherwise may
be just a waste of time. The opposite extreme are exercises in the famous course
by Landau and Lifshitz, which are based on old journal publications. In our time,
PREFACE xiii
journal publications are usually far more complicated and specialized, and are read
only by those continuing the same or a related line of research. The best strategy
for the actively engaged reader would be to reproduce the derivations in the text
applying them to similar problems, in the best case, related to their own line of study
or research.
The online Mathematica notebooks complementing the text can be helpful in this
respect. They are open and transparent; the user has to understand the inner workings
of mathematical operations rather than to press buttons and get results out of a black
box. It is not essential to understand the programming; thus, the user should be
able to change program inputs, such as equations and boundary conditions, without
taking care of a panoply of commands. The notebooks may be also helpful, though in
a lesser degree, to readers not having access to Mathematica, as they still could use
free Wolfram Player to follow the derivations in more detail than in the main text,
and may be able to reproduce the computations in a more familiar or better accessible
program.
The notebooks are written in the latest Mathematica 12 version; the reader is
warned that changes in programming in updated versions sometimes make some com-
ponents of old notebooks incompatible. The logical structure of Mathematica 1 was a
brilliant creation of the genius of Stephen Wolfram and his early collaborators, while
further development went along the common path of adding more features of conve-
nience, first of all, fine graphics, the source of almost all illustrations in this book.
Bibliographical Notes
There is a limited number references in this book, confined to specific problems treated
in detail and wider surveys. Most prominent names are mentioned but I do not ex-
pect the reader to look into original publications in obscure journals, nor to engage in
priority issues. Neither do I wish to endorse some monographs or review papers over
the others. The inquisitive readers interested in going further along their own line of
inquiry will be not spared the adventure of googling the amazing maze of scientific
literature or, more safely, restricting the search to more reliable sources in the Web of
Science. Relevant references in the text can serve as a seed for this search.
L. Pismen
Haifa, Israel
May 2020
https://www.routledge.com/Working-with-Dynamical-Systems-A-Toolbox-for-
Scientists-and-Engineers/Pismen/p/book/9781138591714.
1
Whence Dynamical Systems
Chapter Overview
What is a dynamical system? A wide definition would be that it is any system devel-
oping in time. This definition is much too wide, as it encompasses spatially distributed
systems with an infinite number of degrees of freedom. A working definition should
be more restrictive. The core of the subject lies in initial-value problems for ordinary
differential equations (ODE). Discrete evolving systems should be included as well, as
they are intimately connected to continuous ones, as we shall see. A dynamical system
defined in this way is a relatively simple mathematical object, but more complex phe-
nomena, taking place in spatially distributed systems, may be sometimes reduced to
simpler descriptions without losing their essential features, and the study of boundary
value ODE problems may also benefit from the techniques developed in the theory of
dynamical systems.
The essential feature of problems we are interested in, whether discrete or contin-
uous is nonlinearity. A linear system is bland and easily molded by external inputs.
On the opposite, a nonlinear system has a character of its own, tends to be attracted
to certain states, and may exhibit mutable, and even irregular, behavior. It is the
changes in behavior, that we are mostly interested in.
This introductory chapter has a wide scope but just touches upon the problems
to be discussed at length in the sequel. The material of different sections of this
chapter might be very familiar (and therefore superfluous) to some readers but novel to
others. Our aim here is to make clear that dynamical systems are not a mathematical
abstraction but are ubiquitous in nature and technology. We will see that similar
equations may come from entirely different backgrounds, and the essential difference
may lie in their general properties rather than in their origin.
What are these essential properties? There is an apparently clear divide between
conservative and dissipative systems, but this distinction is blurred in open systems
interacting with their environment. Our attention is focused on dynamical systems
of different origin (mechanical, chemical, and biological) that are likely to display
interesting (complex, nontrivial) behavior. The tools for studying the various dynamic
behaviors will be developed in the following chapters.
1
2 CHAPTER 1. WHENCE DYNAMICAL SYSTEMS
1.1 Classical Mechanics

1.1.1 Conservative Equations of Motion
Classical mechanics is the oldest – classical indeed – source of dynamical systems.
Problems of classical mechanics are not a centerpiece of this book, but we will start
with them anyway, since they provided methods and ideas for many other applications.
Lagrangian formalism. Classical mechanics deals with nondissipative motion of point

particles. The state of a mechanical system at any given moment is determined by po-
sitions of these particles qn , which are vectors that can be represented by components
qnj in a coordinate system fitting the geometry of the problem. To predict positions
at future times, it is sufficient to know instantaneous velocities vn = dqn /dt ≡ q̇n .
The index n marks here the particles, and the index j, vector components. One can
arrange all variables in arrays Q, Q;˙ together, they span 2N -dimensional phase space
of the system; N is the number of degrees of freedom equal to the number of particles
multiplied by the number of spatial dimensions1 .
Derivation of equations of motion starts from the Lagrangian L(Q, Q): ˙
1
L= mn |q̇n |2 − V (Q). (1.1)
2 n
The first term is kinetic energy, and the second term is potential energy; mn is the
mass of nth particle. The equations of motion are obtained by varying the action
integral S = Ldt:
d ∂L ∂L
− = 0. (1.2)
dt ∂q̇n ∂qn
This leads to the Newton law
mn q̈n = Fn (Q), (1.3)
where Fn = −∂V /∂qn is the force acting on the nth particle. The equations of motion
imply conservation of energy:
d ∂L 1
E= q̇n · −L= mn |q̇n |2 + V (Q). (1.4)
n
dt ∂q̇n 2 n
Trajectories of a conservative mechanical system lie on the manifold E = const in

˙ the dimensionality of this manifold can be further reduced if
the phase space (Q, Q);
there are additional conservation laws, which may follow from the symmetries of the
system. Equilibria are extrema of the potential V (Q) wherein all forces vanish.
1 Arrays are printed in bold italics to distinguish them from vectors. The dot product may stand
for either vector or array contraction, i.e., the sum over the respective components.
1.1. CLASSICAL MECHANICS 3
Hamiltonian formalism. The Hamiltonian formalism involves, instead of velocities

q̇n , momenta pn = ∂L/∂q̇n . The Hamiltonian function is defined as

H= pn · q̇n − L. (1.5)
n
The coordinates and momenta are conjugate variables, and the equations of motion
equivalent to (1.3) are obtained as
∂H ∂H
q̇n = , ṗn = − , (1.6)
∂pn ∂qn
or, in the component notation,
∂H ∂H
q̇nj = , ṗnj = − . (1.7)
∂pnj ∂qnj
We have to keep in mind that the momentum is a covariant vector, unlike contravari-
ant position q and velocity v = q̇ vectors. Components of a covariant vector are
written with lower vector indices, to distinguish them from upper-index contravariant
components. The straightforward relation p = mv can be used in Cartesian coordi-
nates only. Otherwise, this relation has to be written explicitly as pi = mgij v j , where
gij is an applicable metric tensor, and summation over repeating upper and lower
indices is implied. For example, in the spherical coordinates r, θ, φ, the metric tensor
is diagonal with the elements {1, r2 , r2 sin2 θ}.
Provided the Hamiltonian does not depend on time explicitly, its conservation
follows directly from the equations of motion:
dH ∂H dqn ∂H dpn

= · + · = 0. (1.8)
dt n
∂qn dt ∂pn dt
The Hamiltonian is related to the action integral defined by integrating the Lagrangian
in time along a certain trajectory, so that S˙ = L. Using this in (1.5) yields

δS = pn dqn − Hdt. (1.9)
According to this relation, the partial derivatives of action with respect to the coordi-
nates are equal to the respective momenta, ∂S/∂qn = pn , while ∂S/∂t = −H. This
leads to the Hamilton–Jacobi equation
∂S
+ H(Q, SQ ) = 0, (1.10)
∂t
where SQ is an array of partial derivatives ∂S/∂qn . The advantage of this form is
that it includes only first-order derivatives; it will be particularly useful in Sect. 4.1.2.
1.1.2 Systems with One Degree of Freedom

Basic equations. The simplest case is one-dimensional (1D) motion of a single particle
(N = 1). The Lagrangian and Hamiltonian, with the coterie of indices dropped, are
1 1
L= mq̇ 2 − V (q), H= mp2 + V (q). (1.11)
2 2
The respective equation of motion is
mq̈ = F (q), F = −V (q). (1.12)
Generally, the force may vanish at several positions corresponding to stable equilibria
(s) (u)
q = qj [minima of V (q) with F (q) < 0] and unstable equilibria q = qj [maxima

of V (q) with F (q) > 0]. Evidently, these equilibria must alternate. The dynamics
can be visualized as frictionless motion of a massive point particle on a slide with the
profile proportional to V (q) under the action of gravity.
Nonlinear oscillator. The simplest nontrivial example is a nonlinear oscillator with a

symmetric double-well potential, which we shall write in a dimensionless form V (q) =
1 2 2
4 (1−q ) . The mass m can be incorporated in the energy scale, and in this formulation
there is no difference between the velocity q̇ and the momentum p (note, however, the
basic distinction explained in Sect. 1.1.1). The orbits in the phase plane q, p are curves
of constant energy levels:
1
E= mp2 + V (q) = const. (1.13)
2
One can see in Fig. 1.1 two kinds of phase space orbits separated by a figure-eight
passing through the unstable equilibrium q = p = 0. This orbit, corresponding to
(a) (b)

()
-
-
-
-
- - - - - -

Figure 1.1. (a): A symmetric double-well potential. (b) The phase plane of a nonlinear
oscillator showing three kinds of orbits

-
-
- - -
/
Figure 1.2. The phase plane of a pendulum
E = E0 = 1/4, is called a homoclinic. It has an infinite period – we’ll have much

more to say on this in the following two chapters. At 0 < E < E0 , the orbits are
disconnected, and correspond to oscillations around one of stable equilibria q = ±1.
At higher energies, E > 1/4, the orbits lie outside the homoclinic and surround
all three
equilibria. All these orbits can be easily drawn by plotting the function
p = ± 2(E − V (q)) for some values of energy below and above E0 .
Ideal pendulum. Another example – physical but as highly abstract as all textbook
examples – is a pendulum with a massive bob tied to a fixed pivot by a weightless
noninertial stiff rod of length l and moving without friction in a fixed plane in the
field of gravity. There is a single position variable – the deviation angle θ from the
rest position, and the potential is V (θ) = mgl cos θ. The Lagrangian is
1 2 ˙2
L= ml θ + mgl cos θ. (1.14)
2
The parameters – mass m and acceleration
of gravity g, as well as l, can be removed
by choosing mgl as the energy scale and l/g as the time scale. The dimensionless
Lagrangian and Hamiltonian rewritten in these units are
1 ˙2 1 2
L= θ + cos θ, H= p − cos θ. (1.15)
2 2
The respective dimensionless equation of motion is
θ̈ + sin θ = 0. (1.16)
The equilibria are θ = nπ with any integer n, but those with n even are stable (minima
of V ) and those with n odd are unstable (maxima of V ). The orbits in the phase plane
θ, p =θ̇ (Fig. 1.2) are level curves E = const expressed, as before, by the function
p = ± 2(E − cos θ).
One can observe that there is no difference in positions with angles differing by
2π. There is, however, a qualitative difference between undulation, with the angle
oscillating around a stable equilibrium, and rotation, with the angle increasing or
decreasing monotonically. The two kinds of trajectories are separated by an infinite
period orbit passing through unstable equilibria, which corresponds to the level E = 1
(Fig. 1.2). This distinction would not be as clear if, say, the vertical position z =
−l cos θ, rather than angle, was chosen as a dynamic variable. On the other hand,
since θ modulo 2π corresponds to the same spatial position, one could hesitate whether
to call an infinite period orbit a homoclinic (one starting and ending at the same
equilibrium state) or heteroclinic (one connecting two distinct equilibria).
1.1.3 Symmetries and Conservation Laws

Spherical pendulum. The above example can be slightly modified by removing the
restriction of planar motion. This is a spherical pendulum (Fig. 1.3a); its motion
should be viewed in spherical coordinates θ, φ. In this case, there are four phase
variables, but equations of motion remain simple, since the potential depends on the
azimuthal angle θ only.
The Lagrangian can be written in the dimensionless form scaled as in (1.15):
1 2
L= θ̇ + φ˙ 2 sin2 θ + cos θ. (1.17)
2
Take note that sin2 θ is the element of the metric tensor that has to be entered when the
respective coordinates are multiplied. The equations of motion are more transparent
when derived from the Hamiltonian
1 2
H = pθ θ˙ + pφ φ˙ − L = p + p2φ − cos θ. (1.18)
2 θ
The Hamiltonian is independent of the polar angle φ; this symmetry results in con-
servation of the respective momentum pφ = ∂L/∂φ˙ = φ˙ sin2 θ. Notably, this does not
˙ as the rotation rate in the horizontal plane varies
imply a constant angular velocity φ,
with changing polar angle.
The motion, combining azimuthal undulation with rotation around the symmetry
axis, looks superficially rather complex, but remains predictable. Since the system is
constrained by two conservation laws (energy and angular momentum), phase space
trajectories cover, generally, a 2D surface in the 4D phase space. Any continuous
symmetry implies a conservation law (Noether’s theorem). In particular, conserva-
tion of energy is a consequence of the symmetry to translations in time, whereby
the Lagrangian does not depend on time explicitly. The conservation of the angular
momentum follows from the symmetry to azimuthal translations.
Integrable systems. A symmetry may be not as obvious as that. It might be possible,

however, to transform the coordinates and momenta to action-angle variables Jn , Θn ,
such that the Hamiltonian acquires a normal form dependent on Jn only. Then
Figure 1.3. (a): A spherical pendulum. (b): Coupled pendulums tied by an elastic spring.
(c): A double pendulum. (d): Long exposure of a double pendulum exhibiting chaotic motion
(tracked with a LED)
the equations of motion become

dΘn ∂H dJn
= , = 0. (1.19)
dt ∂Jn dt
Thus, the action variables are integrals of motion, while the angle variables rotate
with a constant speed. When such a canonical transformation is possible, the system
is integrable, and can be fully resolved analytically. The dynamics reduces to rotation
around an N -dimensional torus (where N the number of degrees of freedom). Since
the rotation periods are, generally, incommensurate, the motion is quasiperiodic: the
system never returns to the original state. This is also true in the simple case of a
spherical pendulum. Integrable systems are nongeneric; they can be useful, however,
for studying more complex dynamics when conservation laws are slightly perturbed
(see Sect. 4.1).
1.1.4 Interacting Particles

Coupled pendulums. Motion becomes far more complicated when interactions be-
tween particles are taken into account. In this case, coordinates of different particles
in V (Q) cannot be separated, so that the force Fn = −∂V /∂qn acting on nth particle
depends on other particles’ positions.
As an example, consider two coupled pendulums with the rods of the same length
l moving in parallel planes separated by the length d, and with identical bobs tied by
an elastic spring, which is unstrained in the rest position (Fig. 1.3b). The squared
distance between the bobs is d2 + 2l2 [1 − cos(θ1 − θ2 )], where θn are deviation angles,
and the Lagrangian, brought to the dimensionless form as in Sect. 1.1.2, is
1
k
2
L= θ̇ + cos θn − [1 − cos(θ1 − θ2 )], (1.20)
n
2 2
where k is the dimensionless elastic constant of the spring, which controls the strength
of interaction. The dynamic behavior is influenced thereby in contradictory ways.
At k → 0, the two pendulums are independent, while at k → ∞, they undulate
synchronously as a single body. Thus, it is at intermediate strength of interaction
that nontrivial dynamics is to be expected.
Elastic pendulum. A simpler device apt to behave chaotically is an elastic pendulum

hung on a weightless spring rather than on a stiff rod. Here the pendulum and a spring
are “interacting particles”, and the dynamic variable additional to the deviation angle
θ is the relative elongation of the spring x. The dimensionless Lagrangian is
1 k
L= (1 + x)2 θ̇2 + ẋ2 + (1 + x) cos θ − x2 . (1.21)
2 2
Similar to the case of coupled pendulum, in the limit of a very large elastic constant
the system reverts to a simple pendulum, but a very small elastic constant would allow
the bob to drop indefinitely far from the pivot.
Double pendulum. Another example is a double pendulum with massive bobs and
weightless stiff rods (Fig. 1.3c). There are four variables, angles of both rods θn , n =
1, 2 and the corresponding velocities θ˙n or momenta pn , but no symmetries at hand
to simplify the problem, even when the pendulums are identical. Assuming this is
true, we can use the same units as in Sect. 1.1.2 to write the Lagrangian in the form
(Landau and Lifshitz, 1960)
1
L = θ˙12 + θ˙22 + θ˙1 θ˙2 cos(θ1 − θ2 ) + 2 cos θ1 + cos θ2 . (1.22)
2
On the road to chaos. In all above cases, the source of complications lies in the
interaction terms containing cos(θ1 − θ2 ) or both x and cos θ. The levels H = const
are now (generically) 3D hypersurfaces, and the motion can be chaotic – and, for
all practical purposes, unpredictable, even though the equations of motion remain,
of course, fully deterministic. The entanglement is stronger in the case of a double
pendulum, where not only coordinates but their derivatives or momenta appear in
the same nonlinear term. A chaotic trajectory traced in the experiment is shown
in Fig. 1.3d. Coupled and elastic pendulums are easier to analyze, since the weak
interaction limit can be tested in these cases. Chaotic dynamics of these models can
be analyzed in the same way as a more abstract example in Sect. 4.1.3.
Henri Poincaré has understood this more than a century ago, before quantum un-
certainty shocked the science, but nobody paid attention. His motivation was a study
of a three-body planetary interaction – a more realistic and practically important
problem than frictionless mathematical pendulums with weightless rods or springs.
Learned philosophers as well as common folk kept viewing classical mechanics as a
model of determinism, even though a die and a roulette, tools of chance games, are
respectable macroscopic mechanical systems. The reason was that only an unrepre-
sentative minority of mechanical systems – those that can be resolved analytically –
could be studied and understood before computers became commonplace (more on
this in Sect. 4.1.1).
1.1.5 Dissipative Motion

Ubiquity of dissipation. As conservation laws are woven into the canvas of physics,
dissipative (nonconservative) models always appear as approximate or coarse-grained
models describing the averaged behavior of complex systems. Dissipation may appear
in a mechanical system as an additional term taking into account friction, which is,
in fact, the result of disordered motion of a great number of microscopic particles
lumped by a macroscopic model characterized by few degrees of freedom. Similar
mechanisms underlie dissipation due to resistance in electric circuits or losses in optical
transmission lines. Mean field models of thermodynamic or chemical origin are always
dissipative, as they describe averaged dynamics characterized by a small number of
thermodynamic variables. In fluid mechanics, dissipation is brought in by viscosity
and diffusion, stemming, like friction, from disordered motion on a microscopic level.
Dissipative models may be also derived from a mass, energy, or momentum con-
servation law, but contain sources or sinks that destroy conservation of the respective
integrals. Besides dissipation in the common sense of the word – loss of energy, mo-
mentum, etc. – nonconservative systems may include sources due to external pumping,
or “antidissipation”. Behavior of such systems, constrained neither by conservation
laws nor by the boring laws of thermodynamics, is particularly rich and interesting;
most dynamical systems we shall study belong to this class. From a formal point of
view, conservative systems are nongeneric: a randomly generated set of equations will
have no conservation laws.
Driven mechanical systems with friction. Dissipation of me-

chanical energy by friction makes long-time dynamics trivial,
since the system always evolves to a stable equilibrium (though
not necessarily to the one corresponding to the lowest energy
level). Losses can be, however, compensated by external energy
supply. This, in essence, increases dimensionality of the avail-
able phase space. The energy of the system is not conserved
anymore and can be lost and gained at a variable rate in the
course of evolution, so that the system is not constrained to a
constant energy level.
Adding to (1.12) a friction term proportional to the velocity
and an external force F0 (q, t) brings it to the form
mq̈ + kq̇ − F (q) = F0 (q, t), (1.23)
where k > 0 is a friction coefficient.

As an example, consider a planar pendulum with a pivot
Figure 1.4. A pen- rotating clockwise in the same vertical plane with the ampli-
dulum on a rotating tude a and frequency ω (Fig. 1.4). The horizontal and vertical
pivot coordinates of its bob are
x = a cos ωt + l sin θ, y = −(a sin ωt + l cos θ), (1.24)
so that the kinetic energy is

m 2 m 2 2
x˙ + ẏ 2 = a ω + l2 θ˙2 − 2alωθ˙ sin(θ + ωt) . (1.25)
2 2
This expression can be used to write down the Lagrangian, but the latter looks more
neat if the term linear in θ˙ is removed by extracting the time derivative of alω cos(θ +
ωt), which would not affect the equation of motion to be obtained by varying the
Lagrangian. The constant term a2 ω 2 can be removed as well, and the θ-dependent
part of the potential energy V (y) = −gy → gl cos θ remains the same as for a pendulum
hanging on a fixed point. Transforming to the dimensionless variables as in Sect. 1.1.2,
we write the Lagrangian in the form
1 ˙2
L= θ + γ sin(θ − ωt) + cos θ, (1.26)
2
where γ = aω 2 /g. Varying this expression and adding friction yields the dimensionless
equation of motion
θ¨ + kθ˙ + sin θ = γ cos(θ − ωt). (1.27)
This equation can be further simplified assuming undulations to be small, θ 1.
Then the angle in the driving term can be neglected, and the nonlinear term expanded
to the lowest nonvanishing order, sin θ = θ 1 − 13 θ2 . Removing also the numerical
1.2. CHEMICAL KINETICS 11
√
coefficient by setting θ = 3q yields a version of the Duffing equation – the simplest
model of a driven dissipative nonlinear oscillator:
q̈ + kq̇ − (1 − q 2 )q = γ cos ωt. (1.28)
The same equation in a modified form describes the motion of a nonlinear spring with
friction under the action of a periodic force. It comes there with the reverse sign of
the linear term:
q̈ + kq̇ + ω02 (1 + αq 2 )q = γ cos ωt. (1.29)
√
The parameters√ α, ω0 can be eliminated by rescaling q → q/ α, t → t/ω0 , k →
ω0 k, γ → γω02 / α, ω → ωω0 . Only the sign of α is relevant: α = 0 corresponds to a
linear spring, while with the positive sign, the spring strengthens with strain. The case
α < 0 that corresponds to a weakening spring √ is problematic, as the solutions then
diverge when q escapes from the interval |q| < α. We will return to this equation,
as one of examples of chaos in forced dissipative systems, in Sect. 4.5.2.
1.2 Chemical Kinetics

1.2.1 Mass Action
Stoichiometric equations. A different kind of dynamical systems is based on appli-
cation of the mass action law lying in the foundation of chemical kinetics. Production
rates of different species due to a chemical reaction are related by the stoichiometry
of the reaction. Any chemical reaction is characterized by its stoichiometric equation,
which has a general form
νi Ai = 0, (1.30)
where Ai are symbols of chemical species, and νi are stoichiometric coefficients defining
proportions between changes of the amount of reactants and products in the course
of the reaction. Stoichiometric coefficients can be multiplied by any number, but it
is reasonable to choose the lowest possible set of integers. The signs are customarily
chosen in such a way that stoichiometric coefficients of reactants are negative, and
those of products are positive. However, the reaction may be reversible and proceed
in either direction; then the choice depends on the net direction of chemical change.
A more conventional notation is

|νi |Ai ν i Ai , (1.31)
i∈I− i∈I+
where I− is a set of reactants and I+ is a set of products. It may be more suggestive

when one of the reactants is also a reaction product. It happens in chain reactions,
the most celebrated example of which – stepping outside chemistry, but retaining the
same formalism – is the nuclear chain reaction multiplying the number of neutrons in
explosive nuclear fission.
Chemical kinetics. The mass action law assumes that the reaction rate is proportional
to the product of reactant concentrations. This, indeed, should be true if reactant
molecules, present in macroscopic amounts, are uniformly distributed in space, and
the reaction might occur when they collide at random. The probability of the reaction
upon collision is measured by the reaction rate constant k. Thus, the change of the
concentration ci of a chemical species due to a reversible reaction defined by (1.31) is
⎛ ⎞
|ν |
ċi = νi ⎝k+ ci i − k − cνi i ⎠ , (1.32)
i∈I− i ∈I +
where k± are rate constants of the direct and reverse reactions.

This general formula is unnecessarily clumsy, since triple and higher-order colli-
sions are highly unlikely, and the expression should not contain terms higher than
quadratic if the stoichiometric equation reflects a true reaction mechanism. This, in
fact, almost never happens. Chemical reactions, which require restructuring of sta-
ble molecules, proceed, as a rule, through a number of stages involving short-lived
intermediate species. Some chemical stages rapidly reach equilibrium, which enforces
certain algebraic relations among the concentrations of reacting species. Even when
no partial equilibrium is reached, concentrations of intermediate species taking part
in fast reactions will arrive at certain quasistationary values dependent on instanta-
neous concentrations of long-lived species. Using either equilibrium or quasistationary
relations in kinetic equations of slower stages leads to nonpolynomial expressions for
reaction rates even when all original equations are based on the mass action law. Some
examples are given in the next subsection.
Reaction rate expressions, measured experimentally or deduced from a detailed
mechanistic model after eliminating intermediate species, may have variegated forms
involving some functions of the array of concentrations C:
ċi = νi f (C), (1.33)
For a single reaction, one can eliminate concentrations of all species except one with the
help of stoichiometric relations (ci −c0i )/νi = idem, where c0i are initial concentrations.
For a general reaction system, the number of independent equations equals to the
rank of the matrix of stoichiometric coefficients. This number defines the effective
dimensionality of the phase space of the system.
Chemical reactors. Chemical reactions in a closed system (what is called a “batch”

reactor) always proceed to a state of thermodynamic equilibrium in such a way that
free energy of the system decreases monotonically. This appears to rule out any non-
trivial dynamic behavior. When Belousov observed back in the mid-20th century the
famous oscillatory reaction, known now as the Belousov–Zhabotinsky (BZ) reaction,
he was prevented from publishing this, ostensibly, rude violation of thermodynamic
laws – or, in fact, of thermodynamic misconceptions, as was later realized. Oscillations
are indeed impossible close to equilibrium, but far from equilibrium these constraints
are invalidated, and a monotonic decrease of the free energy contradicts oscillatory
dynamics no more than a monotonic decrease of the energy of a not perfectly fric-
tionless pendulum prevents it from undulating for hours before coming to rest at the
equilibrium.
A chemically reacting mixture can be sustained in a stationary nonequilibrium
state or in a persistent nonstationary, e.g., oscillatory state when reactants are supplied
and products removed from the system; energy supply or removal are needed as well
when the reactions have a heat effect as they usually do. Both chemical reactors and
living cells operate in this way, whatever complex reactions might be involved. A
simplest type of a chemical reactor is a well stirred tank with one or more inlets and
outlets. The reacting mixture within the tank, as well as in the outlet streams has a
uniform composition and temperature, so that the material balance equations become
ċi = νi f (C) + β c0i − ci , (1.34)
where c0i
is the concentration in the inlet stream, and β is the inverse of the average
residence time, i.e., the ratio of the volumetric flow rate to the reactor volume.
Another kind of an industrial chemical reactor is a plug flow reactor (PFR). An
ideal PFR is a flow-through reactor, where each volume element of the inlet flow spends
in the reactor identical time. The flow elements entering the PFR at different moments
of time do not mix, and therefore each one can be considered as an infinitesimal batch
reactor. Therefore in a PFR operating in a stationary regime, the clock time is replaced
by the residence time that has passed from the moment of entry of a particular flow
element. The residence time t can be related to the current spatial position x by the
differential relationship dx = v dt, where v is the local flow velocity. Although a batch
reactor and a PFR are not at all similar physically, a stationary PFR is described by
exactly the same equations as a batch reactor, and we shall consider both together.
The similarity is lost when nonstationary operation of a PFR is considered, or when
deviations of a real flow pattern in the reactor from the ideal model are taken into
account.
1.2.2 Adsorption and Catalysis

Quasistationary relations. Fast intermediate reaction stages are commonly encoun-
tered in catalytic or enzyme reactions. Any catalytic reaction is a multistep process
that involves stages of adsorption, chemical transformation of the adsorbed species,
and desorption of products. An enzyme is just a biochemical catalyst, attaching a
reactant in order to change its structure and facilitate its transformation. Usually, the
adsorption and desorption stages are much faster than the reaction stage, and can be
assumed to be at a quasi-equilibrium corresponding to local reaction conditions at a
particular moment of time.
A simple monomolecular reaction proceeds according to the following mechanism:
A1 + A0 A∗1 , A∗1 → P + A0 , (1.35)
where A0 is a free catalytic site or an enzyme molecule, P is a nonadsorbing product,

and the asterisk denotes the adsorbed forms of the reactant A1 . The adsorption rate
is proportional to the fraction of free sites c0 = 1 − c∗ , where c∗ is the fraction of
occupied sites. Assuming adsorption equilibrium, the latter is determined by the
algebraic relation
k + c (1 − c∗ ) = k − c∗ , (1.36)
where k ± are adsorption and desorption rate constants (which should both far exceed
the reaction rate constant k). This relation defines c∗ , which further determines the
chemical reaction rate per active site kc∗ , i.e., the rate of production of P . Since the
intermediate adsorption stage is fast, the rate of consumption of A1 is the same, as
can be also checked algebraically:
kσKc
ċ = − . (1.37)
1 + Kc
where K = k+ /k − is the adsorption equilibrium constant, and σ is the concentration
of active sites. The reaction rate saturates at c K −1 (Fig. 1.5a). This mechanism is
known as Langmuir–Hinshelwood kinetics in heterogeneous catalysis and Michaelis–
Menten kinetics in enzymatic reactions. A similar equation with a sharper saturation
threshold (introduced on formal grounds), called the Hill equation, is often used in
biochemical applications:
kσ(Kc)n
ċ = − . (1.38)
1 + (Kc)n
Competitive and cooperative adsorption. More varied expressions are obtained when
a second species competing for the same adsorption sites is added. Then the fraction
of free sites is c0 = 1 − c∗1 − c∗2 , where c∗i is the fraction of sites occupied by the
respective species. The adsorption equilibrium relations analogous to (1.36) become
ki+ ci (1 − c∗1 − c∗2 ) = ki− c∗i (i = 1, 2), (1.39)
where ki± are adsorption and desorption constants of the two species. The ratio
c∗1 K1 c 1 ki+
= , Ki = , (1.40)
c∗2 K2 c 2 ki−
dependent on adsorption equilibrium constants Ki , is obtained when the two relations
(1.39) are divided one by the other. Combining this with either of (1.39) yields
Ki ci
c∗i = (i = 1, 2). (1.41)
1 + K1 c1 + K2 c2
The resulting kinetic expression depends on the role of the second species. If it is just
a passive inhibitor, (1.37) is modified to
kσK1 c1
ċ1 = −kσc∗i = − . (1.42)
1 + K1 c1 + K2 c2
(a) ) b)
Figure 1.5. (a): Saturation of the chemical reaction rate: the Langmuir–Hinshelwood equa-
tion (n = 1) and the Hill equation with n = 2 and n = 4. (b): A typical dependence of the
reaction rate of a bimolecular catalytic reaction on the concentration of a competing (blue
curve) or cooperating (orange curve) reactant at a fixed level of the other concentration,
(made larger for the competitive case to make scaling compatible).
Alternatively, A2 may take part together with A1 in a bimolecular reaction, so

that the detailed mechanism is
Ai + A0 A∗i (i = 1, 2), A∗1 + A∗2 → P + 2A0 . (1.43)
Then, according to the mass action law, the chemical reaction rate per active site is
kc∗1 c∗2 , and (1.42) yields kinetic equations for the concentrations of both reactants:
kσK1 K2 c1 c2
ċi = − (i = 1, 2). (1.44)
(1 + K1 c1 + K2 c2 )2
The important feature of this reaction rate expression is a possibility of its decreasing
with the concentration of one of the reactants (Fig. 1.5b).
Interaction of adsorbed species may be cooperative rather than competitive. For
example, a simplified model of glycolysis (biochemical oxidation of glucose) involves
adsorption of the reactant by the adsorbed reaction product rather than by an unbound
enzyme. The chemical scheme of the process is
A2 + A0 A∗2 , A1 + A∗2 A∗1 , A∗1 → A2 + A∗2 . (1.45)
If the two adsorption stages are at equilibrium, the equations defining the fraction of
sites c∗i occupied by the reactant and the product are
K1 c1 c∗2 = c∗1 , K2 c2 (1 − c∗1 − c∗2 ) = c∗2 . (1.46)
Using this to find c∗i , we arrive at the overall reaction rate expression
kσK1 K2 c1 c2
ċi = −kσc∗1 = − . (1.47)
1 + K 2 c 2 + K1 K 2 c 1 c 2
The contrast between both cases is clearly seen in Fig. 1.5b.
1.2.3 Autocatalysis and Self-Inhibition

Far from equilibrium, the function f (C) may have a form qualitatively different from
the mass action law (1.32); in particular, the reaction may be slowed down by re-
actants (self-inhibition) or accelerated by reaction products (autocatalysis). Either
property is essential for nontrivial dynamics capable of producing multiple stationary
states and oscillations, as we shall further see. An extreme example of autocatalysis
is a branched chain reaction, such as the one occurring in nuclear fission where a neu-
tron causing a heavy nucleus to break up generates, alongside splinter nuclei, more
neutrons. Chemical chain reactions involving free radicals are common and had been
studied way before the nuclear fission was discovered.
Branching should be moderated by removal of the autocatalytic agent through
transport or other reactions if we want to observe long-time dynamics rather than
having the system explode on us or somebody else. An example of an autocatalytic
system contained within a finite phase space volume is a celebrated artificial qua-
sichemical system called, after its birthplace fused with its prominent feature, the
Brusselator model. The dynamic equations are constructed as mass conservation
equations describing two fictitious chemical reactions
2A1 + A2 → 3A1 , A1 → A2 , (1.48)
with kinetics obeying the mass action law and the reactant A1 supplied at a constant
rate and removed at a rate proportional to its concentration. Scaling the concen-
trations and time in an ingenuous way, one can bring the dynamic equations to the
form
u̇ = 1 − (μ + 1)u + μu2 v, γv̇ = u − u2 v, (1.49)
where the only stationary solution u = v = 1 is independent of the two remaining
dimensionless parameters μ, γ. This system, though being rather unrealistic, is used
as a convenient model of oscillations and pattern formation due to its algebraic sim-
plicity. Other artificial models, some of which we shall encounter further in this book,
have been constructed likewise to illustrate various aspects of oscillatory and chaotic
dynamics.
Natural examples of self-inhibition are found in catalytic or enzymatic reactions. In
the bimolecular reaction characterized by the kinetic expression (1.44), self-inhibition
arises under conditions when strong adsorption of one of the reactants reduces the re-
action rate by suppressing adsorption of the other reactant; thus, if K1 c1 far exceeds
both K2 c2 and unity, the nonlinear expression in (1.44) is approximately proportional
to c2 /c1 , so that the reaction rate decreases when concentration of the strongly ad-
sorbing reactant increases, as seen in Fig. 1.5b.
1.2.4 Thermal Effects

Arrhenius law. A common autocatalytic agent in chemical processes is heat. Heat
released in an exothermic reaction increases the temperature of the reacting mixture,
which, in turn, accelerates the reaction. The ratio of the reaction heat to the heat
capacity of the reacting mixture can be viewed as a “stoichiometric coefficient” of
temperature. Denoting it as ν0 , we can write, analogous to (1.34), the dynamic
equation for temperature, complementing kinetic equations of chemical species:
Ṫ = ν0 f (C, T ). (1.50)
The temperature dependence in f (C), now extended to f (C, T ), comes from the tem-
perature dependence of reaction rate constants, commonly expressed by the Arrhenius
law, which assumes that only sufficiently energetic molecules exceeding a certain acti-
vation energy E may undergo chemical transformation. If energy and temperature are
measured in the same units (setting the Boltzmann constant to unity), the fraction of
such molecules in an ensemble obeying the Maxwell distribution is e−E/T . The acti-
vation energy usually far exceeds the thermal energy, so the fraction in the exponent
is very small, and the reaction rate constant,
k = k0 e−E/T , (1.51)
is a sharply increasing function of temperature reaching saturation only at unrealisti-

cally high temperatures T ∼ E. For a stirred reactor, (1.50) is extended to the energy
balance equation analogous to (1.34):
Ṫ = νT f (C, T ) + βT (T0 − T ) . (1.52)
The constants βT , T0 may combine the effect of different ways of cooling and heating,
either by adjusting the temperature of the inlet stream or by heat exchangers, or both.
A representative example. A standard example, which we shall work out in the se-
quel, is a monomolecular (first-order) exothermic reaction A1 → A2 with the simplest
kinetic function f (c1 ) = k(T )c1 and ν1 = −1. Instead of concentration, it is conve-
nient to use the conversion ratio u = 1−c1 /c01 . The inverse rate constant at a reference
temperature k(T0 ) serves as the time scale. The activation energy E is not fit to be
used as a temperature scale, being, as mentioned above, too large. A more convenient
scale suitable to a common temperature range is T02 /E. Defining the reduced dimen-
sionless temperature as v = E (T − T0 ) /T02 brings (1.34), (1.52) to the dimensionless
form

v
u̇ = exp (1 − u) − μu, (1.53)
1+ v

v
γv̇ = ν exp (1 − u) − μv, (1.54)
1+ v
containing four parameters
β1 νT c01 Eβ1 β1 T0
μ= , ν= , γ= , = . (1.55)
k(T 0 ) βT T02 ) βT E
This system, as well as (1.44), generates multiple stationary states and oscillations.
The parameter γ affects the dynamics only; if it is small, temperature is less inertial
than conversion. The parameter 1 is not essential and can be set to zero, thereby
approximating the Arrhenius temperature dependence by a simple exponent. This
results in the loss of a nonphysical solution with v ∼ −1 . The two essential parameters
affecting stationary states are μ, the ratio of characteristic times of chemical reaction
and reactant supply, and ν, expressing thermal effects. A slightly simplified system
containing only these parameters still exhibits all essential generic features of two-
component dynamical systems, and will be presented in Sects. 2.4 and 3.2 as a template
for a detailed study of their behavior.
1.3 Biological Models

1.3.1 Population Dynamics
Interacting populations. Equations of population dynamics are constructed, as in
chemical kinetics, based on the mass action law. Birth, death, and interactions among
populations can be presented, similar to (1.30), as “reactions” involving one, two, or
more species, e.g.,
Ai → (νi − 1)Ai , 2Ai → 0, (1.56)

Ai + Aj → (νi + 1)Ai + (νj + 1)Aj . (1.57)
In the first “reaction”, νi is the excess of the birth over death rate for an ith species,
both assumed being linear in the population density ci . The second “reaction” in
(1.56) expresses the effect of congestion. In (1.57), Ai is a predator, and Aj is a prey
when νi > 0 and νj < 0. If both νi and νj are positive, it expresses the effect of
symbiosis, and when both are negative, the effect of competition.
Kinetic equations. A general equation system including linear growth or decay and
quadratic interactions has the form

ċi = ci ki + kij cj . (1.58)
These equations can be further extended to a general form, dependent on an array of

variables
ċi = ci f (C), (1.59)
where f (C) is some function of the array C of population densities ci . This may look
similar to the general kinetic expression (1.33), but the substantial difference is that
ci = 0 always remains a solution, and the dynamics is restricted to ci ≥ 0. Another
common generalization is introducing delays, so that past population densities enter
(1.58) or (1.59).
1.3. BIOLOGICAL MODELS 19
(a) (b)

- -

Figure 1.6. (a) Logistic growth. Inset: the growth rate (1.60). (b) Effect of the cubic
growth rate. The solid curves corresponding to c− = 0.5 (above) and c− = 0.2 (below) with
initial concentrations above these values approach the same limit c+ = 1, but the population
decreases when the starting value is below c− (dashed curve at c− = 0.5)
A single population with a linear growth rate would either die away or grow ex-
ponentially to its Malthusian doom, depending on whether ν is negative or positive.
The growth can be limited by adding a quadratic term expressing congestion effects.
This leads to the logistic growth equation
ċ = kc(c+ − c), (1.60)
where c+ is the maximum population. We can see in Fig. 1.6a that the population
density, initially growing exponentially, saturates at the limiting value c+ .
The kinetic expression can be further modified by allowing for negative growth
rates at population densities below some critical value c− (Allee effect), leading to a
cubic expression
ċ = kc(c − c− )(c+ − c). (1.61)
The dynamics of population growth shown in Fig. 1.6b, if started at c slightly larger
than c− , is not dissimilar to that of the logistic curve but the population is extinguished
if drops below c− .
Lotka–Volterra system. A simple example of interaction of two populations is the

Lotka–Volterra system involving “autocatalytic” prey and predator which dies away
linearly but is supported by a quadratic interaction term. The general prey–predator
model describes as well plant–herbivore or host–parasite dynamics. We write this
system in a dimensionless form
u̇ = u(1 − v), v̇ = μv(u − 1), (1.62)
where u, v are scaled population densities, and the only parameter μ expresses the ratio
of the predator death rate to the net growth rate of the prey. A peculiar feature of this
system is a Hamiltonian structure revealed by transforming to the new variables x =

ln u, y = ln v. The transformed dynamic equations are derivable from the Hamiltonian
H = μ(ex − x) + ey − y:
ẋ = 1 − ey ≡ −∂H/∂y, ẏ = μ(ex − 1) ≡ ∂H/∂x. (1.63)
Conservation of H necessitates oscillatory dynamics, with trajectories surrounding
the only nontrivial equilibrium u = v = 1. The amplitude of oscillations increases for
levels of H rising above its minimal value H0 = 1 + μ (see Fig. 1.7).
Unlike problems of classical mechanics,
there is no intrinsic reason for a conserva-
tion law to appear in population dynamics.
The Hamiltonian structure of the Lotka–
Volterra system is purely accidental, and
breaks down when any change is introduced
into the model to make it more realistic,
e.g., accounting for congestion or competi-

tion, as in the models we will consider in
detail in Sect. 2.5. Although the conserva-
tion law simplifies the analysis, it points out,
in essence, to a deficiency of this nongeneric
model. All this said, the model had impor-
tant practical consequences. It made it clear
that eliminating a predator may not help
prey: when done at prey’s population den-
Figure 1.7. Trajectories of the Lotka– sities below the unstable stationary state, it
Volterra system.
moves the system to higher levels of H and
just increases the amplitude of oscillations.
This is a reason why wolves, previously hunted to extinction, are reintroduced in
Europe and North America.
1.3.2 Epidemiological Models

A different scheme operates in models of infection and immune response. Viral infec-
tion can be described by the equivalent reaction scheme
A1 + A2 → A1 + A3 , A3 → νA1 . (1.64)
The first reaction is infection by virus A1 of a healthy cell A2 turning it to an infected
cell A3 . The second reaction is production of the virus by an infected cell. Adding
here linear growth or death terms we arrive at the dynamical system
ċ1 = νk13 c3 − k1 c1 ,
ċ2 = −k12 c1 c2 + k2 c2 ,
ċ3 = k12 c1 c2 − k3 c3 . (1.65)
1.3. BIOLOGICAL MODELS 21
This system reduces to the Lotka–Volterra model if infected cells are short-lived, so
that the last equation reaches a quasistationary state.
A basic epidemiological model involves susceptible (A1 ), infected (A2 ), and immune
(A3 ) populations and the binary interaction A1 + A2 → 2A2 . Assuming that all
populations give birth to susceptible individuals with the rate constant k0 , and adding
a linear death (with the rate constants ki , k1 = k3 < k2 ), infection (with the rate
constant k12 ) and immunization (with the rate constant k23 ) terms, we arrive at the
system
ċ1 = k0 (c1 + c2 + c3 ) − k1 c1 − k12 c1 c2 ,

ċ2 = k12 c1 c2 − k23 c2 − k2 c2 ,
ċ3 = k23 c2 − k3 c3 . (1.66)
It is easy to integrate these, as well as chemical, kinetic equations for a particular set
of parameters but we need more advanced tools of bifurcation analysis to identify the
various possible regimes as the parameters are varied.
1.3.3 Neural and Genetic Networks

Interactions of various proteins and morphogens in living cells generally follow the
rules of chemical kinetics, but expression of a gene is basically a logical operation: it
is switched on or off when a certain concentration level of a particular combination of
proteins is reached within the cell. In its turn, the production of proteins, which is
dependent on the state of a particular gene, affects expression of other genes. Likewise,
a neuron is fired when a potential induced at its synapses by other neurons reaches a
certain threshold. The term “neural network” applies as well to computational devices
and algorithms based on the same principle.
A number of neurons or genes interacting in this way form a neural or genetic
network. These networks are unfathomably complex in living organisms, but complex
dynamic behavior may be generated even by relatively small networks. A general
equation system of a network of this kind can be written in the form
⎛ ⎞

u̇i = μi f ⎝ aij uj − ηi ui ⎠ − λi ui . (1.67)
j
In the genetic context, ui denotes the concentration of the ith protein and λ its decay
rate; f (x) is a monotonic function changing from 0 at x → −∞ to 1 at x → ∞; the
simplest option is the Heaviside (step) function jumping from H(x) = 0 at x < 0 to
H(x) = 1 at x > 0. The values ηi denote then the thresholds of the linear combinations
of protein concentrations, characterized by the coefficients aij , above which the gene
governing the production of the ith protein becomes active; μi denotes the production
rate above this threshold. In the neural context, ui denotes the potential of the ith
synapse and ηi , the excitation threshold of a neuron affecting this synapse.
The coefficients aij can be of either sign, with the positive ones stimulating and
the negative ones repressing activation or expression. The sum sign in (1.67) should
not be interpreted literally. It is more like a logical operation: in Wolpert’s (2011)
words, “it is rather like voting: the unit looks at all the activation and inhibition and
their weightings, and decides who wins”.
In order to make a complex network more manageable, it can be separated in
several “layers”. Some ui may be considered as “inputs” subject to outside control,
which can be exercised by affecting the respective thresholds or by adding source
terms to their dynamic equations. Other ui are viewed as “outputs” available for
observation; the rest constitute a “hidden” layer. The layered structure can be made
more restrictive by allowing links leading from “lower” to “upper” layers (with the
input at the bottom and the output on the top) but not in the opposite direction.
1.4 Electric Currents

1.4.1 Electric Circuits
Linear circuits. Electric circuits contain power sources, resistors, capacitors, and in-
ductors. The variables are currents and voltages. The voltage drop VLR on a segment
with the resistance R and inductance L is
˙
VLR = RI + LI, (1.68)
˙ while the charge q on a capacitor
dependent both on current I and its time derivative I,
with the capacitance C and the related voltage drop VC evolve as
q̇ = C −1 V˙ C = I. (1.69)
For an elementary circuit shown in Fig. 1.8a, the sum VLR + VC equals to the voltage
V0 supplied by the power source. Taking q as the dynamic variable and combining
(1.68), (1.69), one can write the dynamic equation
Lq̈ + Rq̇ + Cq = V0 . (1.70)
This equation has exactly the same structure as the equation of dissipative motion of
a driven dissipative mechanical system with a single degree of freedom (1.23) with a
linear function F (q) and the imposed voltage V0 playing the role of a constant external
force.
Nonlinear circuits. For a long time, electrical engineers dealt with circuits described
by linear equations like (1.70) applied to larger circuits with more degrees of freedom.
This does not interest us: a linear system does not have a life of its own, it just reacts
to external inputs. The most straightforward way to make the system nonlinear is to
introduce some dependence of the capacitance on the charge. Choosing the simplest
cubic function and applying harmonic driving V0 = γ cos ωt yields, after removing
1.4. ELECTRIC CURRENTS 23
(a) (b)
R
R
V0 C L C2 C1 NR
Figure 1.8. (a): Elementary electric circuit. (b): Chua’s circuit containing a nonlinear
negative resistance NR
superfluous parameters by scaling, the same Duffing equation (1.28) that we derived
in a totally different way before.
Another possibility is to introduce a nonlinear variable resistance. A vacuum tube
may amplify the current, i.e., act as a negative resistance, at low and damp it at high
charges. A driven circuit containing such a nonlinear resistor is modeled by the forced
van der Pol equation
q̈ − k(1 − q 2 ) q̇ + q = γ cos ωt. (1.71)
Nonlinearities may enter in many other ways; they may be sustained by realistic
physical mechanisms, e.g., the resistance increasing with the current due to heating,
or may be introduced in a more sophisticated way by creating special combinations
of elements with desired behavior, such as operational amplifiers, to be plugged into
a circuit. A well-studied circuit with complex dynamics is Chua’s circuit (Fig. 1.8b)
containing a nonlinear negative resistance – an amplifier with a positive feedback. It is
described by the system of three equations for the voltages V1 , V2 across the capacitors
C1 , C2 and the electric current I through the inductor L:
V̇1 = α[V2 − V1 − f (V1 )],

RC2 V˙ 2 = V1 − V2 − RI, I˙ = −βV2 . (1.72)
The function f (V1 ) describes the electrical response of the nonlinear resistor and α, β
are constants. This circuit is known for generating chaotic dynamics (see Sect. 4.3.2).
More coupled equation systems can be obtained for branching circuits with the
help of Kirchhoff’s laws following from conservation of energy and electric charge:
the algebraic sum of currents should vanish at each junction, and the voltage drops
around any closed contour should add up to zero. The interest to dynamics of nonlinear
electric circuits was at its peak in the pre-digital age when analog computers – suitably
constructed circuits where currents and voltages could be easily measured – were used
as differential equations solvers.
1.4.2 Electrochemical Reactions

Electrochemical cells. An electric circuit may include an electrochemical element
where the passing current is accompanied or caused by chemical reactions on elec-
trodes involving electron transfer. Electrochemical elements may consume power, like
electrolysers, or produce it, as fuel cells. The stoichiometric scheme of an electrochem-
ical reaction written in the form (1.30) or (1.31) includes, among the symbols Ai , ions
and the electron, as well as electroneutral species.
Kinetics of electrochemical reactions. If a

voltage difference is applied between two
electrodes (anode and cathode) placed in an
electrolyte solution (Fig. 1.9) but no elec-
trochemical reaction takes place, no current
passes through the electrolyte, and the en-
tire voltage drop is concentrated in electric
double layers formed in the immediate vicin-
ity of the electrodes. The thickness of dou-
ble layers usually lies in a nanometer range,
and can only become macroscopic in non-
Figure 1.9. A basic electrochemical cell conducting liquids containing a very small
concentration of ions.
The activation barrier of an electrochemical reaction is overcome with the help of a
surface overpotential Φ, i.e., the difference between the actual potential drop between
the electrode and the electrolyte solution and its equilibrium value at zero current.
Therefore, the reaction rate depends on Φ exponentially:
k = k0 eαΦ/T . (1.73)
The electrode current per unit area is proportional to the electrochemical reaction
rate:
I = νe k0 f (C)eαΦ/T , (1.74)
where the “stoichiometric coefficient” νe = nF is the Faraday number F (i.e., the
electron charge times the Avogadro number) multiplied by the number of electrons n
produced as the result of the reaction.
This relation may be applicable only in a limited range of Φ. As the overpotential
increases, the physical conditions of the electrode surface may change. Most com-
monly, a protective film is formed, and, as a result, the electrode is passivated, i.e.,
the value of k0 drops abruptly. The current may increase exponentially with Φ again
in the transpassive region.
Electrochemical circuits. The inverse of the derivative of the current-voltage relation,

Re = (dI/dΦ)−1 can be interpreted as the effective resistance of the electrode. The
electrode has a substantial capacitance as well, which is inversely proportional to
1.4. ELECTRIC CURRENTS 25
the double-layer thickness. Thus, it can be modeled by a resistor and a capacitor

connected in parallel. The dynamic equation for the overpotential Φ is written then
as
˙ = −I(C, Φ) + V0 − Φ ,
γΦ (1.75)
AR
where γ is the double-layer capacitance per unit area, V0 is the externally applied volt-
age, and A is the electrode area. This equation can be solved together with dynamic
equations of the type (1.32) for chemical species taking part in the reaction, with the
rate constants k± dependent on the overpotential Φ. If the reactants are supplied and
removed from the electrochemical cell, the applicable equations are (1.34). For a single
reaction, the chemical rates in (1.33) or (1.34) are related to the current per unit area
I as f (C) = Iσ/νe , where σ = A/V is the electrode area per unit volume. These
equations are, of course, applicable only when the electrolyte is well stirred, so that
the concentrations are uniform everywhere except, perhaps, a diffusion layer adjacent
to the electrode. In the latter case, the expression for the current or electrochemical
reaction rate should be modified to account for diffusional limitations.
One can notice that (1.75) has the same form as (1.34) or (1.52), and the exponen-
tial dependence of the current on the overpotential is rather similar to the Arrhenius
temperature dependence of the reaction rate of an endothermic reaction. Indeed,
(1.75) can be reduced to the same dimensionless form (1.54), but with −v standing
as the argument of the exponent.
If the current increases with growing overpotential as in (1.74), the effective re-
sistance is positive and, assuming the concentrations to be fixed, (1.75) has a single
stable solution. Instabilities and steady state multiplicity may appear when the effec-
tive resistance is negative in some range of Φ, as it happens in the passivation region.
Otherwise, instabilities may be caused, as in chemical reactions, by autocatalysis or
self-inhibition.
1.4.3 Membrane Transport

The performance of electrochemical cells and, in particular, of modern fuel cells also
depends on the transport of reactants toward electrodes and the change of potential
in the electrolyte. Ion transport and electrochemical reactions play as well a major
role in cellular membrane transport and propagation of signals in neural networks.
A membrane can be modeled as a capacitor in parallel with a resistor. The voltage
drop across the membrane obeys the same relation (1.69) as in a standard circuit,
but the current-voltage dependence I(V ), determined by the ion transport within the
membrane may be far more involved.
The ion transport combines diffusion, driven by the concentration gradient ∇ci of
a particular ion, and migration, driven by the potential gradient ∇φ. The ion flux j
is defined, therefore, by the Nernst–Planck equation
ji = −D∇ci − T −1 nF ci ∇φ, (1.76)

where n is the ion charge and F is the Faraday number. This equation contains
already a spatial dependence of the concentration and potential, and the concentration
distribution can be computed by solving the equation
ċi = ∇ · ji . (1.77)
The same equation applies to the concentration distribution in the electrolyte separat-
ing the electrodes in an electrochemical cell, unless it is well mixed by convection as
in a stirred tank discussed in Sects. 1.2.1, 1.2.4. This is already a partial differential
equation (PDE), which lies outside the scope of dynamical systems.
The description is, however, simplified if the ions do not undergo any chemical
transformations within the membrane, and stationary concentration levels c± i are
maintained at both sides of the membrane, say, at x = ±L/2. The flux must be
then constant and directed along the coordinate x across the membrane. Then (1.77),
with the term containing the time derivative dropped, reduces to an ordinary differen-
tial equation (ODE) defining the ion concentration distribution within the membrane.
It has to be complemented by an equation of φ, which depends also on other ions
present and, possibly, on nonionic conduction mechanisms. The simplest case is a
constant potential gradient, which will be achieved if the electric current is largely
carried by a “background electrolyte” present at a constant concentration far exceed-
ing ci or by an efficient nonionic mechanism. Then ∇φ = φ (x) = V /L, and (1.77)
becomes
dci nF V j
+ ci + = 0. (1.78)
dx DLT D
This first-order equation is subject to two boundary conditions, and the flux j is
computed to satisfy a solvability condition of this overdetermined problem:
D nF V c+ − c−
i exp (nF V /T )
j= ci i . (1.79)
L T 1 − exp (nF V /T )
This flux can be entered in the equations describing chemical transformations in vol-
umes separated by the membrane bringing them to the form similar to (1.34) but with
the coefficients βi dependent on the potential V applied across the membrane.
Outside chemical and electrical engineering, there is enduring interest in electro-
chemical processes in neural networks. Nerve axons can be viewed as elements of a
sophisticated electric circuit, but chemical reactions and molecular transport play no
lesser role in the activity of neural networks than electric currents.
1.5 Spatially Extended Systems

1.5.1 From Time to Coordinate Dependence
Real-life problems of physics, biology, and engineering involve systems varying in space
as well as in time. Spatial dependence may be totally suppressed only in simplest cases,
1.5. SPATIALLY EXTENDED SYSTEMS 27
when we deal with a small number of rigid bodies approximated as point particles or
when spatial dependence of concentrations or population densities is suppressed by
intensive mixing or rapid motion in a habitat of a limited extent.
Dynamics of spatially distributed systems is described by partial rather than or-
dinary differential equations and, in more complicated cases, by integral equations
involving nonlocal interactions between different locations. We will not touch upon
such complicated topics in the framework of this book, but restrict to simpler cases
when a space-dependent problem can be either reduced to ODEs or replaced by a
finite system of dynamic equations.
In the simplest 1D setup, stationary solutions obey an ODE with time replaced by
a spatial coordinate. Such “dynamical systems in space”, to be considered in Ch. 5,
differ in important aspects from time-dependent dynamical systems but retain helpful
analogies. In this way, it is also possible to understand some basic spatially-dependent
dynamical problems.
Spatially extended nonequilibrium systems are capable, in principle, of exhibiting
extremely complex dynamics. With a great (virtually, infinite) number of degrees
of freedom present, one could expect chaotic behavior of unfathomable complexity
to be a norm. However, this does not necessarily happen. More often than not,
the latent complexity of behavior is not realized, and the prevailing state can be
well characterized by a few properly chosen dynamic variables. This suggests that
a dynamical system involving a limited number of variables might be sufficient to
faithfully describe the dynamic behavior in lieu of a full partial differential equation
system, at least approximately and within a certain range of parameters.
1.5.2 Fourier Decomposition

Fourier series. The choice of such a representative set of variables is by no means
obvious. In the absence of other hints, one can expand the variables in a Fourier
series, say,
u(x, t) = a0 (t) + aj (t) exp(ikj · x). (1.80)
The amplitude aj (t) of the mode with a wave vector kj is complex, and if the variable
u(x, t) is real, the Fourier expansion should include also complex conjugate terms with
the amplitudes aj (t). When the expansion is used in the original system of equations,
different modes couple through nonlinear terms, and, generally, no finite number of
modes would ever suffice, as, given two modes with wave vectors ki , kj , their sum
and difference will always be generated. The expansion can be, however, truncated,
leaving only a small number of modes and neglecting the rest.
The resulting dynamical system with polynomial functions fi (A) dependent on the
array A of amplitudes aj (t) of the chosen modes may look like “chemical” systems of
Sect. 1.2.1 derived from the mass action law. The semblance is, however, superficial,
since completely different principles restrict possible combinations. The product of the
amplitudes aj , ak , . . . may appear in the equation for the amplitude ai if the respective
wave vectors add up to zero, ki + kj + kk + . . . = 0. This condition ensures that the

modes in question are in resonance and, generally, makes the dynamics dependent on
a combination of their phases.
Which modes are relevant? One can never be sure which, if any, features of the
original system are preserved by such a bold action, but, with a good luck, it may bring
success. In this way, Lorenz (1963) constructed a three-variable dynamical system
by retaining three modes of the Navier–Stokes and convective diffusion equations
describing thermal convection in a fluid heated from below. The resulting system
was not actually faithful to the original, but turned instead (after some 15 years of
obscurity) into a celebrated paradigm of chaotic dynamics (see Sect. 4.3.3).
A more systematic (though empirical) approach to spectral truncation is offered by
the Karhunen–Loewe method used, in particular, for compressing images and discern-
ing “coherent structures” in turbulent flow and complex chemically reacting systems.
This method employs statistical analysis of measured or computed data to find out an
optimal combination of modes sufficient to reproduce the data in a faithful but eco-
nomic way. This combination would, of course, change when the underlying system is
modified or driven to a different regime.
It is far more appealing to be able to discern relevant modes analytically; this may,
indeed, be possible under favorable conditions. The key is given by scale separation:
some modes may evolve on far longer characteristic times than do the others. Fast
modes relax to their quasistationary values dependent on the current values of slow
modes, which remain virtually unchanged during this relaxation period. Following
this, slow modes evolve on a longer time scale, while fast modes follow them quasis-
tationarily, being bound by quasiequilibrium algebraic relations – much in the same
way as rapidly reacting chemical species reaching a partial equilibrium (Sect. 1.2.3).
This is expressed by Haken’s (2004) slaving principle: “fast modes are slaved by slow
modes”.
Clearly, only slow modes are relevant for long-time evolution. Therefore, the orig-
inal system, however complex it might be, can be reduced to a dynamical system for
the amplitudes of slow modes. Exact derivation of such a system of amplitude equa-
tions is possible when the amplitudes of slow modes are small, so that the state of the
system deviates only slightly from a known “trivial” homogeneous stationary state.
We shall study some systems of this kind in Sect. 5.4. Dynamical systems derived
through this approach can reproduce the dynamics of the original underlying system
faithfully only within a limited range of parameters and within a limited region of the
phase space, since slow modes would, generally, be modified or cease to be slow when
either parameters or values of dependent variables change. Nevertheless, they may
reproduce dynamics correctly in a qualitative sense even outside the domain where
their derivation can be justified formally.
1.6. CONTINUOUS VS. DISCRETE 29
1.6 Continuous vs. Discrete

1.6.1 Iterative Systems
What is more complicated – continuous or discrete? The answer appears to be straight-
forward. Ancient Greeks had trouble with the notion of continuum, musing over Zeno’s
paradoxes of Achilles unable to overtake a tortoise or an arrow unable to move. Mod-
ern freshmen, introduced to calculus as small children looking on a speedometer in a
family car, must still struggle with limit theorems invented to overcome the doubts of
the ancients, and acquire aversion to mathematics after passing calculus 101. We are
accustomed to think in terms of a continuum, yet we are solving continuum problems
using digital computers, which, by their nature, are capable to work with discrete en-
tities only. After all, digital computing, translated literally, is just counting on fingers,
which our computers have aplenty.
Discrete systems appear to be far simpler – yet a discrete dynamical system, such
as a 1D iterative map
un+1 = f (un ), (1.81)
generates far more complicated dynamics than a 1D continuous dynamical system.
This is an intrinsic reason for numerical instabilities plaguing digital computations.
A continuous dynamical system is constrained topologically: trajectories may never
intersect. This is why chaotic dynamics is impossible in less than three dimensions. A
discrete system, free of this constraint, may be chaotic already in a single dimension
(more on this in Sect. 4.2).
An iterative map may arise as a mathematical procedure, e.g., finding a root of
an algebraic equation f (u) = 0 using Newton’s method. In 1D, it is expressed by the
iterative map
f (un )
un+1 = un − . (1.82)
f (un )
The method works because the iterations converge to an equilibrium, but this never
can be warranted when the dimension is higher than one. This is why we need expert

- -
- -
Figure 1.10. Newton’s method. Iterations, shown by dashed lines, promptly converge to a
root, but not necessarily to the closest one.
computer scientists fighting numerical instabilities. Even in 1D, the root may be not
the one we are looking for if it is not unique (see Fig. 1.10).
In population dynamics, an iterative map may express a relation between num-
bers of individuals or population densities in successive generations. Such a relation,
whether derived from a model or measured empirically, would reflect, in fact, the result
of a very complex continuous process. Thus, a discrete dynamical system, if it is taken
from the physical world rather than introduced as a mathematical abstraction, will be
usually a result of a reduction of a continuous system. Examples of such reductions
are given further in this section. Such a reduction, if it is possible, is always welcome,
as it greatly facilitates computations. On the other hand, a reductive procedure hid-
den in the background justifies the complexity of behavior of a discrete system, which
reflects the complexity of a continuous system hidden underneath.
1.6.2 From Continuous to Discrete

Reduction of a continuous dynamic system to a map is possible when an analytically
predictable behavior is interrupted by discrete events. A simple example is a bouncing
ball. As with other mechanical systems (Sect. 1.1.5), taking account of dissipation
and compensating it by pumping energy makes the system both more realistic and
more interesting. We consider therefore a ball bouncing with an imperfect restitution
on a vibrating plate. We shall neglect air resistance, another dissipation mechanism,
assuming it to be relatively weak.
Between two consecutive impacts, the ball jumping off vertically at a moment tn
with a velocity un decelerates in the gravity field, so that its velocity and vertical
displacement are
1
u(t) = un − gt, z(t) = un t − gt2 . (1.83)
2
(a) (b)

Figure 1.11. The bouncing velocity as a function of time. (a): The map (1.86) with α =
0.99, γ = 0.05; the solid line shows the change of the bouncing velocity on a quiescent plate.
(b): The map (1.87) with α = 0.9, γ = 0.1.
Another random document with
no related content on Scribd:
which was worm-eaten in the step; and the Etoile shortened hers,
the head of it being sprung. We likewise took in, from on board the
store-ship, the flour and biscuit which still belonged to us, in
proportion to our number. There were fewer pulse than we at first
thought, and I was obliged to cut off above a third part of the
allowance of the (gourganes) pease or caravanses for our soup: I
say ours, for every thing was equally distributed. The officers and the
sailors had the same nourishment; our situation, like death, rendered
all ranks of men equal. We likewise profited of the fair weather, to
make good observations.
Observation of On the 11th, in the morning, M. Verron brought his
longitude. quadrant and pendulum on shore, and employed
them the same day, to take the sun’s altitude at noon. The motion of
the pendulum was exactly determined by several corresponding
altitudes, taken for two days consecutively. On the 13th, there was
an eclipse of the sun visible to us, and we got every thing in
readiness to observe it, if the weather permitted. It was very fair; and
we saw both the moment of immersion, and that of emersion. M.
Verron observed with a telescope of nine feet; the chevalier du
Bouchage with one of Dollond’s acromatic telescopes, four feet long;
my place was at the pendulum. The beginning of the eclipse was to
us, on the 13th, at 10h. 5′ 45″ in the morning, the end at 00 h. 28′ 16″
true time, and its magnitude 3′ 22″. We have buried an inscription
under the very spot where the pendulum had been; and we called
this harbour Port Praslin.
This observation is so much the more important, as it was now
possible, by its means, and by the astronomical observations, made
upon the coast of Peru, to determine, in a certain fixed manner, the
extent of longitude of the vast Pacific Ocean, which, till now, had
been so uncertain. Our good fortune, in having fair weather at the
time of the eclipse, was so much the greater, as from that day to our
departure there was not a single day but what was dreadful. The
continued rains, together with the suffocating heat, rendered our stay
here very pernicious to us. On the 16th, the frigate had completed
her works, and we employed all our boats to finish those of the
Etoile. This store-ship was quite light, and as there were no stones
proper for ballast, we were obliged to make use of wood for that
purpose; this was a long troublesome labour, which in these forests,
where an eternal humidity prevails, is likewise unwholesome.
Description of Here we daily killed snakes, scorpions, and great
two insects. numbers of insects, of a singular sort. They are
three or four inches long, and covered over with a kind of armour;
they have six legs, projecting points on the sides, and a very long
tail. Our people likewise brought me another creature, which
appeared extraordinary to us all. It is an insect about three inches
long, and belongs to the Mantis genus. Almost every part of its body
is of such a texture, as one would take for a leaf, even when one
looks closely at it. Each of its wings is one half of a leaf, which is
entire when the two wings are closed together; the under side of its
body resembles a leaf, of a more dead colour than the upper one.
The creature has two antennæ and six legs, of which the upper joints
are likewise similar to parts of leaves. M. de Commerçon has
described this singular insect; and I placed it in the king’s cabinet,
preserved in spirits.
Here we found abundance of shells, many of them very fine. The
shoals offered treasures for the study of Conchology. We met with
ten hammer-oysters in one place, and they are said to be a scarce
species[120]. The curiosity of some of our people was accordingly
raised to a great pitch; but an accident happening to one of our
sailors abated their zeal. He was bit in the water by Sailor bit by a
a kind of snake as he was hauling the seine. The water-snake.
poisonous effects of the bite appeared in half an hour’s time. The
sailor felt an excessive pain all over his body. The spot where he had
been bit, which was on the left side, became livid, and swelled
visibly. Four or five scarifications extracted a quantity of blood, which
was already dissolved. Our people were obliged to lead the patient
walking, to prevent his getting convulsions. He suffered greatly for
five or six hours together. At last the treacle (theriaque) and flower
de luce water which had been given him, brought on an abundant
perspiration, and cured him.
This accident made every one more circumspect and careful in
going into the water. Our Taiti-man curiously observed the patient
during the whole course of his sickness. He let us know that in his
country were snakes along the sea-shore, which bit the people in the
sea, and that every one who was thus bit died of the wound. They
have a kind of medicinal knowledge, but I do not believe it is
extensive at all. The Taiti-man was surprised to see the sailor return
to his work, four or five days after the accident had happened to him.
When he examined the productions of our arts, and the various
means by which they augment our faculties, and multiply our forces,
this islander would often fall into an extatic fit, and blush for his own
country, saying with grief, aouaou Taiti, fy upon Taiti. However, he
did not like to express that he felt our superiority over his nation. It is
incredible how far his haughtiness went. We have observed that he
was as supple as he was proud; and this character at once shews
that he lives in a country where there is an inequality of ranks, and
points out what rank he holds there.
Bad weather On the 19th in the evening we were ready to sail,
which but it seemed the weather always grew worse and
persecutes us. worse. There was a high south wind, a deluge of
rain, with thunder and tempestuous squalls, a great sea in the offing,
and all the fishing birds retired into the bay. On the Earthquake.
22d in the morning, towards half an hour past ten o’clock, we
sustained several shocks of an earthquake. They were very sensibly
felt on board our ships, and lasted about two minutes. During this
time the sea rose and fell several times consecutively, which greatly
terrified those who were fishing on the rocks, and made them retreat
to the boats. It seems upon the whole, that during this season the
rains are uninterrupted here. One tempest comes on before the
other is gone off, it thunders continually, and the nights are fit to
convey an idea of chaotic darkness. Notwithstanding this, we daily
went into the woods in search of thatch palms and cabbage trees,
and endeavouring to kill some turtle doves. We Unsuccessful
divided into several bodies, and the ordinary result endeavours to
of these fatiguing caravans, was, that we returned find provisions.
wet to the skin, and with empty hands. However, in these last days,
we found some mangle-apples, and a kind of fruit called Prunes de
Monbin[121]. These would have been of some service to us, had we
discovered them sooner. We likewise found a species of aromatic
ivy, in which our surgeons believed they had discovered an
antiscorbutic quality; at least, the patients who used an infusion of it,
and washed with it, found themselves better.
Description of a We all went to see a prodigious cascade, which
fine cascade. furnished the Etoile’s brook with water. In vain would
art endeavour to produce in the palaces of kings, what nature has
here lavished upon an uninhabited spot. We admired the
assemblage of rocks, of which the almost regular gradations
precipitate and diversify the fall of the waters; with admiration we
viewed all these masses, of various figures, forming an hundred
different basons, which contain the limpid sheets of water, coloured
and shaded by trees of immense height, some of which have their
roots in the very reservoirs themselves. Let it suffice that some men
exist, whose bold pencil can trace the image of these inimitable
beauties: this cascade deserves to be drawn by the greatest painter.
Our situation Mean while our situation grew worse every
grows worse moment of our stay here, and during all the time
every day. which we spent without advancing homeward. The
number of those who were ill of the scurvy, and their complaints
encreased. The crew of the Etoile was in a still worse condition than
ours. Every day I sent boats out to sea, in order to know what kind of
weather there was. The wind was constantly at south, blowing
almost a storm with a dreadful sea. Under these circumstances it
was impossible to get under sail, especially as this could not be done
without getting a spring upon an anchor that was to be slipped all at
once; and in that case it would have been impossible in the offing to
hoist in the boats that must have remained to weigh the anchor,
which we could not afford to leave behind us. These obstacles
determined me to go on the 23d to view a passage between
Hammer island and the main land. I found one, through which we
could go out with a south wind, hoisting in our boats in the channel.
This passage had indeed great inconveniences, and happily we
were not obliged to make use of it. It rained without We leave
intermission all the night between the 23d and 24th. Port Praslin.
At day-break the weather became fair and calm. We immediately
weighed our small bower, fastened a warp to some trees, bent a
hawser to a stream-anchor, and hove a-peek on the off-anchor.
During the whole day we waited for the moment of setting sail; we
already despaired of it, and the approach of night would have
obliged us to moor again, when at half past five o’clock a breeze
sprung up from the bottom of the harbour. We immediately slipt our
shore-fast, veered out the hawser of the stream-anchor, from which
the Etoile was to set sail after us, and in half an hour’s time we were
got under sail. The boats towed us into the middle of the passage,
where there was wind enough to enable us to proceed without their
assistance. We immediately sent them to the Etoile to bring her out.
Being got two leagues out to sea, we lay-to in order to wait for her,
holding in our long-boat and small boats. At eight o’clock we began
to see the Etoile which was come out of port; but the calm did not
permit her to join us till two hours after midnight. Our barge returned
at the same time, and we hoisted her in.
During night we had squalls and rain. The fair weather returned at
day break. The wind was at S. W. and we steered from E. by S. to N.
N. E. turning to northward with the land. It would not have been
prudent to endeavour to pass to windward of it: we suspected that
this land was New Britain, and all the appearances confirmed us in it.
Indeed the lands which we had discovered more to the westward
came very close to this, and in the midst of what one might have
taken for a passage, we saw separate hummocks, which doubtless
joined to the other lands, by means of some low grounds. Such is
the picture Dampier gives of the great bay, which he calls St.
George’s Bay, and we have been at anchor at the N. E. point of it, as
we verified on the first days after our leaving the port. Dampier was
more successful than we were. He took shelter near an inhabited
district, which procured him refreshments, and whereof the
productions gave him room to conceive great hopes concerning this
country; and we, who were as indigent as he was, fell in with a
desart, which, instead of supplying all our wants, has only afforded
us wood and water.
When I left Port Praslin, I corrected my longitude by that which we
obtained from the calculation of the solar eclipse, which we observed
there; my difference was about 3°, which I was to the eastward. The
thermometer during the stay which we made there, was constantly at
22° or 23°; but the heat was greater than it seemed to shew. I
attribute the cause of this to the want of air, which is common here;
this bason being closed in on all sides, and especially on the side of
the reigning winds.
CHAP. VI.
Run from Port Praslin to the Moluccas; stay at Boero.
We put to sea again after a stay of eight days, during which time, as
we have before observed, the weather had been constantly bad, and
the wind almost always southerly. The 25th it returned to S. E.
veering round to E. and we followed the direction of the coast at
about three leagues distance. It rounded insensibly, and we soon
discovered in the offing a succession of islands, one after the other.
We passed between them and the main, and I gave them the names
of the principal officers. We now no longer doubted that we were
coasting New Britain. This land is very high, and seemed to be
intersected with fine bays, in which we perceived fires, and other
marks of habitations.
Distribution of The third day after our departure I caused our
cloaths to the field-tents to be cut up, and distributed trowsers to
sailors. the two ships companies. We had already, on
several occasions, made the like distributions of cloathing of all
kinds. Without that, how would it have been possible that these poor
fellows should be clad during so long a voyage, on which they were
several times obliged to pass alternately from cold to hot, and to
endure frequent deluges of rain? I had, upon the Extreme want
whole, nothing more to give them, all was of victuals.
exhausted, and I was even forced to cut off another ounce of the
daily allowance of bread. Of the little provisions that remained, part
was spoiled, and in any other situation all our salt provisions would
have been thrown over-board; but we were under the necessity of
eating the bad as well as the good, for it was impossible to tell when
our situation would mend. Thus it was our case to suffer at once by
what was past, which had weakened us; by our present situation, of
which the melancholy circumstances were every instant repeated
before us; and lastly, by what was to come, the indeterminate
duration of which was the greatest of all our calamities. My personal
sufferings encreased by those of others. However, I must declare
that not one suffered himself to be dejected, and that our patience
under sufferings has been superior to the most critical situations.
The officers set the example, and the seamen never ceased dancing
in the evenings, as well in the time of scarcity, as in that of the
greatest plenty. Nor has it been necessary to double their pay[122].
Description of We had New Britain constantly in sight till the 3d
the inhabitants of August, during which time we had little wind,
of New Guinea. frequent rain, the currents against us, and the ships
went worse than ever. The coast trenched more and more to the
westward, and on the 29th in the morning, we found ourselves
nearer it than we had yet been: this approach procured us a visit
from some periaguas; two came within hail of the frigate, and five
others went to the Etoile. They carried each of them five or fix black
men, with frizled woolly hair, and some of them had powdered it
white. They had pretty long beards, and white ornaments round their
arms, in form of bracelets. Their nudities were but indifferently
covered with the leaves of trees. They are tall, and appeared active
and robust. They shewed us a kind of bread, and invited us by signs
to go ashore. We desired them to come on board; but our invitations,
and even the gift of some pieces of stuff which we threw over-board,
did not inspire them with confidence sufficient to make them venture
along-side. They took up what was thrown into the water, and by way
of thanks one of them with a sling flung a stone, which did not quite
reach on board; we would not return them evil for evil, so they
retired, striking all together on their canoes, and setting up loud
shouts. They without doubt carried their hostilities farther on board
the Etoile, for we saw our people fire several muskets, which put
them to flight. Their periaguas are long, narrow, and with out-riggers;
they all have their heads and sterns more or less ornamented with
sculptures, painted red, which does honour to their skill.
The next day there came a much greater number of them, who
made no difficulty of coming along-side the ship. One of their
conductors, who seemed to be the chief, carried a staff about two or
three feet long, painted red, with a knob at each end, which, in
approaching us, he raised with both hands over his head, and
continued some time in that attitude. All these negroes seemed to be
dressed out in their best, some had their woolly hair painted red,
others had plumes on their heads, certain seeds in their ears by way
of ear-rings, or large white round plates hanging to their necks; some
had rings passed through the cartilage of the nose; but an ornament
pretty common to them all was bracelets, made of the mouth of a
large shell, sawed asunder. We were desirous of forming an
intercourse, in order to engage them to bring us some refreshments,
but their treachery soon convinced us that we could not succeed in
that attempt. They strove to seize what was offered them, and would
give nothing in exchange. We could scarce get a few roots of yams
from them; therefore we left off giving them, and they retired. Two
canoes rowed towards the frigate at the beginning of night, but a
rocket being fired for some signal, they fled precipitately.
They attack the Upon the whole, it seemed that the visits they
Etoile. made us these two last days had been with no other
view than to reconnoitre us, and to concert a plan of attack; for the
31st, at day-break, we saw a swarm of periaguas coming off shore, a
part of them passed athwart us without stopping, and all directed
their course for the Etoile, which they had no doubt observed to be
the smallest vessel of the two, and to keep astern. The negroes
made their attacks with stones and arrows, but the action was short,
for one platoon disconcerted their scheme, many threw themselves
into the sea, and some periaguas were abandoned: from this time
we did not see any more of them.
Description of The coast of New Britain now ran W. by N. and W.
the northern and in this part it became considerably lower. It was
part of New no longer that high coast adorned with several rows
Britain.
of mountains; the northern point which we
discovered was very low land, and covered with trees from space to
space. The five first days of the month of August were rainy, the
weather thick and unsettled, and the wind squally. We discovered
the coast only by piecemeal, in the clear intervals, without being able
to distinguish the particulars of it: however, we saw enough of it to be
convinced that the tides continued to carry us a part of the moderate
run we made each day. I then steered N. W. and N. W. by W. to
avoid a cluster of islands that ly off the northern extremity of New
Britain. The 4th in the afternoon we discovered two 1768.
islands, which I take to be those that Dampier calls August.
Matthias Island and Stormy or Squally Island. Matthias Island is high
and mountainous, and extends to N. W. about eight or nine leagues.
The other is not above three or four leagues long, and between the
two lies a small isle. An island which we thought we perceived the
5th, at two o’clock in the morning, to the westward, caused us again
to stand to the northward. We were not mistaken; for at ten o’clock
the fog, which till then had been thick, being dissipated, we saw that
island, which is small and low, bearing S. E. by S. The tides then
ceased to set to the southward and eastward which seemed to arise
from our having got beyond the northern point of New Britain, which
the Dutch have called Cape Salomaswer. We were then in no more
than 00° 41′ south lat. We had sounded almost every day without
finding bottom.
Isle of We steered west till the 7th, with a pretty fresh
Anchorets. gale and fair weather, without seeing land. The 7th
in the evening, the sky being very hazy, and appearing at sun-set to
be a horizon of land from W. to W. S. W. I determined to steer S. W.
by S. for the night; at daylight we steered west again. In the morning
we saw a low land, about five or six leagues a-head of us. We
steered W. by S. and W. S. W. to pass to the southward of it, and we
ranged along it at about a league and a half distance. It was a flat
island, about three leagues long, covered with trees, and divided into
several parts, connected together by breakers and sand-banks.
There are upon this island a great quantity of cocoa-nut trees, and
the sea-shore is covered with a great number of habitations, from
which it may be supposed to be extremely populous. The huts were
high, almost square, and well covered. They seemed to us larger
and handsomer than the huts built with reeds generally are, and we
thought we again beheld the houses of Taiti. We discovered a great
number of periaguas employed in fishing all round the island; none
of them seemed to be disturbed at seeing us pass, from which we
judged that these people, who were not curious, were contented with
their fate. We called this island the Isle of Hermits, or Anchorets.
Three leagues to the westward of this, we saw another low island
from the mast-head.
Archipelago; by The night was very dark, and some fixed clouds to
us called the the southward made us suppose there was land;
Echiquier. and, in fact, at day-light we discovered two small
isles, bearing S. S. E. ¾ E. at eight or nine leagues distance. We
had not yet lost sight of them, at half past eight o’clock, when we
discovered another low island, bearing W. S. W. and a little after, an
infinite number of little islands extending to W. N. W. and S. W. of
this last, which might be about two leagues long; all the others,
properly speaking, are nothing but a chain of little flat isles, or keys,
covered with wood; which, indeed, was a very disagreeable
discovery to us. There was, however, an island separated from the
others, and more to the southward, which seemed to us more
considerable. We shaped our course between that and the
Archipelago of isles, which I called the Chess-board, (l’Echiquier)
and which I wanted to leave to the northward. We were not yet near
getting clear of it, This chain discovered, ever since the morning,
extended much farther to the south-westward, than we were at that
time able to determine.
Danger which We endeavoured, as I have observed before, to
we run there. double it to the southward; but in the beginning of
the night, we were still engaged with it, without knowing precisely
how far it extended. The weather being continually squally, had
never shewn us at once, all that we had to fear; to add to our
embarrassment, it became calm in the beginning of the night, and
the calm scarce ended at the return of day. We passed the night
under continual apprehensions of being cast ashore by the currents.
I ordered two anchors to be got clear, and the cables bitted with a
range along the deck, which was almost an unnecessary precaution;
for we sounded several times without finding bottom. This is one of
the greatest dangers of these coasts; for you have not the resource
of anchoring at twice the ship’s length from the ledges, by which they
are bounded. The weather fortunately continued without squalls; and
about midnight a gentle breeze sprung up from the northward, which
enabled us to get a little to the south-eastward. The wind freshened
in proportion as the sun ascended, and carried us from these low
islands; which, I believe, are uninhabited; at least, during the time we
were carried near enough to discern them, we distinguished neither
fires, nor huts, nor periaguas. The Etoile had been, during the night,
in still greater danger than us; for she was a very long time without
steerage-way, and the tide drew her insensibly towards the shore,
when the wind sprung up to her relief. At two o’clock, in the
afternoon, we doubled the westermost of the islands, and steered W.
S. W.
We get sight of The 11th, at noon, being in 2° 17′ south latitude,
New Guinea. we perceived, to the southward, a high coast, which
seemed to us to be that of New Guinea. Some hours after, we saw it
more distinctly. The land is high and mountainous, and in this part
extends to the W. N. W. The 12th, at noon, we were about ten
leagues from the nearest land; it was impossible to observe the
coast minutely at that distance there: it appeared to us only a large
bay, about 2° 25′ south latitude; in the bottom of which, the land was
so low, that we only saw it from the mast-head. We also judged from
the celerity with which we doubled the land, that the currents were
become favourable to us; but in order to determine with any
exactness, the difference they occasioned in our estimated run, it
would have been necessary to sail at a less distance from the coast.
We continued ranging along it, at ten or twelve leagues distance; its
direction was constantly W. N. W. and its height immense. We
remarked particularly two very high peaks, neighbours to each other,
which surpassed all the other mountains in height. We called them
the Two Cyclops. We had occasion to remark, that the tides set to
the N. W. The next day we actually found ourselves further off from
the coast of New Guinea; which here tended away west. The 14th, at
break of day, we discovered two islands and a little isle or key, which
seemed to be between them, but more to the southward. Their
corrected bearings are E. S. E. and W. N. W. They are at about two
leagues distance from each other, of a middling height, and not
above a league and a half in extent each.
Direction of the We advanced but little each day. Since our arrival
winds and on the coast of New Guinea, we had pretty regularly
currents. a light breeze from east to N. E. which began about
two or three o’clock in the afternoon, and lasted till about midnight;
this breeze was succeeded with a longer or shorter interval of calm,
which was followed by the land-breeze, varying from S. W. to S. S.
W. and that terminated also towards noon, in two or three hours
calm. The 15th, in the morning, we again saw the westmost of the
two islands we had seen the preceding evening. We discovered at
the same time other land, which seemed to us to be islands,
extending from S. E. to W. S. W. very low, over which, in a distant
point of view, we perceived the high mountains of the continent. The
highest, which we set at eight o’clock in the morning, bearing S. S.
E. by compass, detached from the others, we called the Giant of
Moulineau, and we gave the name of la Nymphe Alice to the
westmost of the low islands, to the N. W. of Moulineau. At ten in the
morning we fell into a race of a tide, where the current seemed to
carry us with violence to N. and N. N. E. It was so violent, that till
noon it prevented our steering; and as it carried us much into the
offing, it became impossible for us to fix a positive judgment of its
true direction. The water, in the first tide-line, was covered with the
trunks of drift trees, sundry fruits and rock-weeds; it was at the same
time so agitated, that we dreaded being on a bank; but sounding, we
had no bottom at 100 fathom. This race of a tide seems to indicate
either a great river in the continent, or a passage which would here
divide New Guinea; a passage whose entrance would be almost
north and south. According to two distances, between the sun and
moon, observed with an octant, by the chevalier du Bouchage and
M. Verron, our longitude, the 15th at noon, was 136° 16′ 30″ east of
Paris. My reckoning continued from the determined Observations
longitude of Port Praslin; differed from it 2° 47′. We compared with
observed the same day 1° 17′ south latitude. the reckoning.
The 16th and 17th it was almost calm; the little wind that did blow,
was variable. The 16th, we did not see the land till seven in the
morning; and then only from the mast-head, extremely high and
rugged. We lost all that day in waiting for the Etoile, who, overcome
by the current, could not keep her course; and the 17th, as she was
very far from us, I was obliged to bear down to join her; but this we
did not accomplish, till the approach of night, which proved very
stormy, with a deluge of rain and frightful thunder. The six following
days were all as unpropitious to us; we had rain and calms; and the
little wind that did blow was right a-head. It is impossible to form an
idea of this, without being in the situation we were then in. The 17th,
in the afternoon, we had seen from S. by W. ½ W. to S. W. ½ W. by
compass, at about sixteen leagues distance, a high coast, which we
did not lose sight of till night came on. The 18th, at nine in the
morning, we discovered a high island, bearing S. W. by W. distance
about twelve leagues: we saw it again the next day; and at noon it
bore from S. S. W. to S. W. at the distance of 15 or 20 leagues.
During these three last days, the currents gave us ten leagues
northing: we could not determine what they had helped us in
longitude.
We cross the The 20th we crossed the line, for the second time
Equator. the voyage. The currents continued to set us from
the land; and we saw nothing of it the 20th or 21st, although we had
kept on those tacks by which we approached it most. It became,
however, necessary to make the coast, and to range along it, near
enough, so as not to commit any dangerous error, which might make
us miss the passage into the Indian Sea, and carry us into one of the
gulphs of Gilolo. The 22d, at break of day, we had sight of a higher
coast than any part of New Guinea that we had yet seen. We
steered for it, and at noon we set it, when it bore from S. by E. ½ E.
to S. W. where it did not seem to terminate. We passed the line for
the third time. The land ran W. N. W. and we Cross the line
approached it, being determined not to quit it any again.
more till we arrived at its extremity, which geographers call Cape
Mabo. In the night we doubled a point, on the other side of which the
land, still very high, trenched away W. by S. and W. S. W. The 23d at
noon, we saw an extent of coast, of about twenty leagues; the
westmost part of which bore from us S. W. thirteen or fourteen
leagues. We were much nearer two low islands, covered with wood,
distant from each other about four leagues. We Unsuccessful
stood within about half a league; and whilst we attempt on
waited for the Etoile, who was a great distance from shore.
us, I sent the chevalier de Suzannet, with two of our boats armed, to
the northermost of the two islands. We thought we saw some
habitations there, and were in hopes of getting some refreshments.
A bank, which lies the length of the island, and extends even pretty
far to the eastward, obliged the boats to take a large circuit to double
it. The chevalier de Suzannet found neither dwellings, inhabitants,
nor refreshments. What had seemed to us at a distance to form a
village, was nothing but a heap of rocks, undermined and hollowed
into caverns by the sea. The trees that covered the island, bore no
fruits proper to be eaten by man. We buried an inscription here. The
boats did not return on board till ten o’clock at night, when the Etoile
had joined us. The constant sight of the land shewed us that the
currents set here to the N. W.
Continuation of After hoisting in our boats, we strove to keep the
New Guinea. shore on board, as well as the winds, which were
constantly at S. and S. S. W. would permit us. We were obliged to
make several boards, with an intent to pass to windward of a large
island, which we had seen at sun-set, bearing W. and W. by N. The
dawn of day surprised us, still to leeward of this island. Its eastern
side, which may be about five leagues long, runs nearly N. and S.
and off the south point lies a low island of small extent. Between it
and the coast of New Guinea, which runs here nearly S. W. by W.
there appeared a large passage, the entrance of which, of about
eight leagues, lay N. E. and S. W. The wind blew out of it, and the
tide set to the N. W. it was not possible to gain in turning to windward
against wind and sea; but I strove to do it till nine in the morning. I
saw with concern that it was fruitless, and resolved to bear away, in
order to range the northern side of the island, abandoning with regret
a passage, which I thought a fine one, to extricate me out of this
everlasting chain of islands.
Hidden danger. We had two successive alarms this morning. The
first time they called from aloft, that they saw a long range of
breakers a-head, and we immediately got the other tacks on board.
These breakers, at length, more attentively examined, turned out to
be the ripling of a violent tide, and we returned to our former course.
An hour after, several persons called from the forecastle, that they
saw the bottom under us; the affair was pressing; but the alarm was
fortunately as short as it had been sudden. We should even have
thought it false, if the Etoile, who was in our wake, had not perceived
the same shoal for near two minutes. It appeared to them a coral-
bank. Almost north and south of this bank, which may have still less
water in some places, there is a sandy creek, in which are built some
huts, surrounded with cocoa-trees. This mark may so much the
better serve for a direction, as hitherto we had not seen any traces of
habitations on this coast. At one o’clock in the afternoon, we doubled
the N. E. point of the great island; which from thence extended W.
and W. by S. near 20 leagues. We were obliged to hug our wind to
coast it; and it was not long ere we perceived other islands, bearing
W. and W. by N. We saw one at sun-set, which bore even N. E. by
N. to which there joined a ledge, which seemed to extend as far as
N. by W. thus were we once more hemmed in.
Loss of the This day we lost our first master, called Denys,
master of the who died of the scurvy. He was a native of St.
ship. Malo’s, and aged about fifty years; most of them
spent in the king’s service. The sentiments of honour, and extensive
knowledge, that distinguished him in his important charge, caused
him to be universally regretted among us. Forty-five other persons
were afflicted with the scurvy; lemonade and wine only suspended
its fatal progress.
Difficult course. We spent the night upon our tacks; and the 25th,
at day-light, found ourselves surrounded with land. Three passages
presented themselves to us; one opened to the S. W. the second to
W. S. W. and the third almost east and west. The wind was fair for
none but the east; and I did not approve of it, as I did not doubt that it
would carry us into the midst of the isles of Papua. It was necessary
to avoid falling any farther to the northward; for fear, as I have before
observed, we should be imbayed in one of the gulphs, on the east
side of Gilolo. The essential means for getting out of these critical
parts, was therefore to get into a southern latitude; for on the other
side of the S. W. passage we observed to the southward an open
sea, to the utmost extent of our view, therefore I resolved to ply to
windward, in order to gain that outlet. All these islands, which
inclosed us, are very steep, of a moderate height, and covered with
trees. We did not perceive the least appearance of their being
inhabited.
Fourth passage At eleven o’clock in the afternoon, we sounded 45
of the line. fathom, a sandy bottom; this was one resource. At
noon we observed in 00° 5′ N. latitude, having crossed the line a
fourth time. At six in the evening we were so far to windward, as to
be able to fetch the W. S. W. passage, having gained about three
leagues by working the whole day. The night was more favourable,
thanks to the moon-shine, which enabled us to turn to windward
between the rocks and islands. The current, which had been against
us whilst we were passing by the two first passages, likewise
became favourable for us as soon as we opened the S. W. passage.
Description of The channel through which we at last passed out
the channel this night, may be about three leagues broad. It is
through which bounded to the westward by a cluster of pretty high
we pass.
islands and keys. Its eastern side, which at first sight
we took for the westmost point of the great island, is also nothing but
a heap of small islands and rocks, which, at a distance, seemed to
form only one body; and the separations between these islands
shew at first the appearance of fine bays; this is what we discovered
in each tack, that we made towards that shore. It was not till half past
four o’clock in the morning, that we were able to double the
southmost of the little islands of the new passage, which we called
the French Passage. We deepened our water in the midst of this
Archipelago of Islands, in advancing to the southward. Our
soundings were from 55 to 75 and 80 fathom, grey sand, ooze, and
rotten shells. When we were entirely out of the channel, we sounded
and found no bottom. We then steered S. W.
Pass the line a The 26th, at break of day, we discovered an
fifth time. island, bearing S. S. W. and a little after another
bearing W. N. W. At noon we saw no more of the labyrinth of islands
we had left, and the meridian altitude gave us 00° 23′ south latitude.
This was the fifth time of our passing the line. We continued close on
a wind, with the larboard tacks on board, and in the afternoon we
had sight of a small island in the S. E. The next day, at sun-rise, we
saw it somewhat elevated, bearing N. E. about nine or ten leagues
distance, seeming to extend N. E. and S. W. about two leagues. A
large hummock, very steep, and of a remarkable height, which we
named Big Thomas, (Gros Thomas) shewed itself at ten in the
forenoon. At its southern point there is a small island, and there are
two at the northern one. The currents ceased setting us to the
northward; we had, on the contrary, a difference to the southward.
This circumstance, together with our observed latitude, which made
us to the southward of Cape Mabo, totally convinced me that we
were at length entered into the Archipelago of the Moluccas.
Discussion Let me now ask, which this Cape Mabo is, and
concerning where it is situated? Some make it the Cape, which,
Cape Mabo. to the northward, terminates the western part of
New Guinea. Dampier and Woods Rogers place it the former, in one
of the gulphs of Gilolo in 30′ S. lat. The second, eight leagues at
farthest from this great island. But all this part is an extensive
Archipelago of little isles; which, on account of their number, were
called the Thousand Isles, by admiral Roggewein, who passed
through them in 1722. Then in what manner does this Cape Mabo,
which is in the neighbourhood of Gilolo, belong to New Guinea?
Where shall we place it, if (as there is so much reason to believe) all
New Guinea itself is a heap of great islands? the various channels
between which are as yet unknown. It must certainly belong to the
westmost of these considerable isles.
Entrance into On the 27th, in the afternoon, we discovered five
the Archipelago or six islands, bearing from W. S. W. ½ W. to W. N.
of the W. by compass. During night we kept the S. S. E.
Moluccas.
tack; so that we did not see them again the 28th in
the morning. We then perceived five other little isles, which we stood
in for. At noon they bore from S. S. W. 1° W. to S. 10° W. at the
distance of two, three, four, and five leagues. We still saw Big
Thomas bearing N. E. by E. ½ E. about five leagues. We likewise got
sight of another island, bearing W. S. W. seven or eight leagues
distant. During the last twenty-four hours we felt several strong tides,
which seemed to set from the westward. However, the difference
between my reckoning, and the observation at noon, and at the
setting of the bearings, gave us ten or eleven miles to S. W. by S.
and S. S. W. At nine o’clock in the morning I ordered the Etoile to
mount her guns, and sent her cutter to the S. W. isles, in order to see
whether there was any anchorage, and whether these isles had any
interesting productions.
Meeting with a It was almost a calm in the afternoon, and the
negro. boat did not return before nine o’clock in the
evening. She had landed on two isles, where our people had found
no signs of habitation, or cultivation, and not even any kind of fruits.
They were going to return, when, to their great surprise, they saw a
negro, quite by himself, coming towards them in a periagua, with two
outriggers. In one ear he had a golden ring, and his arms were two
lances. He came up to our boat without shewing any marks of fear or
surprize. Our people asked him for something to eat and to drink,
and he offered them water, and a small quantity of a sort of flour,
which seemed to be his ordinary food. Our men gave him a
handkerchief, a looking-glass, and some other trifles of that sort. He
laughed when he received these presents, and did not admire them.
He seemed to know the Europeans, and we thought that he might
possibly be a run-away negro from one of the neighbouring islands
where the Dutch have settlements; or that he had perhaps been sent
out a-fishing. The Dutch call these islands the Five Isles, and send
some people to visit them from time to time. They told us that they
were formerly seven in number, but that two have been sunk by
earthquakes, which happen frequently in these parts. Between these
isles there is a prodigious current, without any anchorage. The trees
and plants are almost all the same here as upon New Britain. Our
people took a turtle here of about two hundred weight.
Sight of Ceram. From this time we continued to meet with violent
tides, which set to the southward, and we kept the course which
came nearest to their direction. We sounded several times without
finding bottom, and till the 30th in the afternoon, we got sight of no
other land than a single isle to the westward, ten or twelve leagues
from us; but then we saw a considerable land bearing south at a
great distance. The current, which was of more service to us than
the wind, brought us nearer to it during night, and on the 31st at day-
break we were about seven or eight leagues from it. This was the
Isle of Ceram. Its coast, which is partly woody and partly cleared,
runs nearly east and west, and we could not see it terminated. This
isle is very high; prodigious mountains rise on it from space to space,
and the numerous fires which we saw on all sides of it, indicate its
being very populous. We passed the day and the next night in
ranging the northern coast of this isle, making our tacks in order to
gain to the westward, and double its westermost point. The current
was favourable to us, but the wind was scant.
Observations I shall here take an opportunity from the contrary
on the winds we had now met with for a long time, to
monsoons in observe, that in the Moluccas, they call the westerly
these parts.
monsoon the northern one, and the easterly
monsoon the southern one; because, during the former, the winds
blow more generally from N. N. W. than from W. and during the
latter, they come most frequently from S. S. E. These winds likewise
prevail in the isles of Papua, and on the coasts of New Guinea; we
got this information by fatal experience, having employed thirty-six
days to make four hundred and fifty leagues in.
1768. The first of September, at the dawn of day, we
September. were at the entrance of a bay, in which we saw
several fires. Soon after we perceived two vessels under sail, built in
form of the Malay boats. We hoisted a Dutch ensign and pendent,
and fired a gun, by which I committed a fault without knowing it. We
have since learnt that the inhabitants of Ceram are at war with the
Dutch, and that they have expelled the latter from almost every part
of their isle. Therefore we made a board into the bay without
success, the boats retreated on shore, and we profited of the fresh
breeze to proceed on our course. The shore at the bottom of the bay
is low and level, surrounded by high mountains; and the bay itself
contains several islands. We were obliged to steer W. N. W. in order
to double a pretty large island, at the point of which you see a little
isle or key, and a sand bank, with some breakers which seem to
extend a league out to sea. This island is called Bonao; it is divided
into two by a very narrow channel. When we had doubled it, we
steered W. by S. till noon.
It blew very fresh from S. S. W. to S. S. E. and we plyed the
remainder of the day between Bonao, Kelang, and Manipa,
endeavouring to make way to the S. W. At ten o’clock in the evening
we discovered the lands of the isle of Boero, by means of the fires
which burnt on it; and as it was my intention to put in there, we
passed the night on our tacks, in order to keep within reach, and if
possible to the windward of it. I knew that the Dutch Project for our
had a weak factory on this isle, which was however safety.
abundant in refreshments. As we were perfectly ignorant of the
situation of affairs in Europe, it was not prudent to venture to learn

Working With Dynamical Systems: A Toolbox For Scientists and Engineers (Series in Computational Biophysics) 1st Edition Len Pismen

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Working With Dynamical Systems: A Toolbox For Scientists and Engineers (Series in Computational Biophysics) 1st Edition Len Pismen

Uploaded by

Copyright:

Available Formats

Working with Dynamical Systems: A

Toolbox for Scientists and Engineers

Patterns and Interfaces in Dissipative Dynamics 2nd

Intelligent Systems for Engineers and Scientists: A

Applied Numerical Methods with Python for Engineers and

Applied Numerical Methods with MATLAB for Engineers and

Machine Learning Toolbox for Social Scientists 1st

Essential Mathematics for Engineers and Scientists 1st

Machine Learning with Neural Networks: An Introduction

Advanced Mathematics for Engineers and Scientists

Molecular Modeling at the Atomic Scale

Coarse-Grained Modeling of Biomolecules

Computational Approaches to Protein Dynamics

Modeling the 3D Conformation of Genomes

Working with Dynamical Systems

For more information about this series, please visit:

and by CRC Press

© 2021 Taylor & Francis Group, LLC

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

ISBN: 9781138591714 (hbk)

Typeset in Computer Modern font

Visit the [companion website/eResources]: [www.routledge.com/9781138591714]

1 Whence Dynamical Systems 1

3.5 Systems with Separated Time Scales . . . . . . . . . . . . . . . . . . 92

4 Chaotic, Forced, and Coupled Oscillators 103

5 Dynamical Systems in Space 163

Online Files 233

Illustration Credits 235

Why This Book?

Whence Dynamical Systems

1.1 Classical Mechanics

Lagrangian formalism. Classical mechanics deals with nondissipative motion of point

Trajectories of a conservative mechanical system lie on the manifold E = const in

Hamiltonian formalism. The Hamiltonian formalism involves, instead of velocities

1.1.2 Systems with One Degree of Freedom

mq̈ = F (q), F = −V (q). (1.12)

Nonlinear oscillator. The simplest nontrivial example is a nonlinear oscillator with a

Figure 1.2. The phase plane of a pendulum

E = E0 = 1/4, is called a homoclinic. It has an inﬁnite period – we’ll have much

1.1.3 Symmetries and Conservation Laws

Integrable systems. A symmetry may be not as obvious as that. It might be possible,

the equations of motion become

1.1.4 Interacting Particles

Elastic pendulum. A simpler device apt to behave chaotically is an elastic pendulum

1.1.5 Dissipative Motion

Driven mechanical systems with friction. Dissipation of me-

mq̈ + kq̇ − F (q) = F0 (q, t), (1.23)

where k > 0 is a friction coeﬃcient.

x = a cos ωt + l sin θ, y = −(a sin ωt + l cos θ), (1.24)

so that the kinetic energy is

q̈ + kq̇ − (1 − q 2 )q = γ cos ωt. (1.28)

1.2 Chemical Kinetics

where I− is a set of reactants and I+ is a set of products. It may be more suggestive

where k± are rate constants of the direct and reverse reactions.

ċi = νi f (C), (1.33)

Chemical reactors. Chemical reactions in a closed system (what is called a “batch”

1.2.2 Adsorption and Catalysis

where A0 is a free catalytic site or an enzyme molecule, P is a nonadsorbing product,

Alternatively, A2 may take part together with A1 in a bimolecular reaction, so

1.2.3 Autocatalysis and Self-Inhibition

1.2.4 Thermal Eﬀects

is a sharply increasing function of temperature reaching saturation only at unrealisti-

Ṫ = νT f (C, T ) + βT (T0 − T ) . (1.52)