Numerical Continuation Methods for Dynamical

Systems Path following and boundary value

problems Understanding Complex Systems 1st
Edition Bernd Krauskopf pdf download
https://ebookgate.com/product/numerical-continuation-methods-for-
dynamical-systems-path-following-and-boundary-value-problems-
understanding-complex-systems-1st-edition-bernd-krauskopf/

Get Instant Ebook Downloads – Browse at https://ebookgate.com

Instant digital products (PDF, ePub, MOBI) available
Download now and explore formats that suit you...

Dynamical Inverse Problems of Distributed Systems

Vyacheslav I. Maksimov

https://ebookgate.com/product/dynamical-inverse-problems-of-
distributed-systems-vyacheslav-i-maksimov/

ebookgate.com

Dynamical Systems Method and Applications Theoretical

Developments and Numerical Examples 1st Edition Alexander
G. Ramm
https://ebookgate.com/product/dynamical-systems-method-and-
applications-theoretical-developments-and-numerical-examples-1st-
edition-alexander-g-ramm/
ebookgate.com

Green s Functions and Boundary Value Problems Third

Edition Ivar Stakgold

https://ebookgate.com/product/green-s-functions-and-boundary-value-
problems-third-edition-ivar-stakgold/

ebookgate.com

Differential Equations with Boundary Value Problems 2nd

Edition John Polking

https://ebookgate.com/product/differential-equations-with-boundary-
value-problems-2nd-edition-john-polking/

ebookgate.com
Dynamical Systems 2010th Edition Werner Krabs

https://ebookgate.com/product/dynamical-systems-2010th-edition-werner-
krabs/

ebookgate.com

Formal Methods Applied to Complex Systems 1st Edition

Jean-Louis Boulanger

https://ebookgate.com/product/formal-methods-applied-to-complex-
systems-1st-edition-jean-louis-boulanger/

ebookgate.com

Applied Differential Equations with Boundary Value

Problems 1st Edition Vladimir A. Dobrushkin

https://ebookgate.com/product/applied-differential-equations-with-
boundary-value-problems-1st-edition-vladimir-a-dobrushkin/

ebookgate.com

Case Studies in System of Systems Enterprise Systems and

Complex Systems Engineering 1st Edition Alex Gorod

https://ebookgate.com/product/case-studies-in-system-of-systems-
enterprise-systems-and-complex-systems-engineering-1st-edition-alex-
gorod/
ebookgate.com

Software Enabled Control Information Technology for

Dynamical Systems 1st Edition Tariq Samad

https://ebookgate.com/product/software-enabled-control-information-
technology-for-dynamical-systems-1st-edition-tariq-samad/

ebookgate.com
Springer Complexity
Springer Complexity is an interdisciplinary program publishing the best research
and academic-level teaching on both fundamental and applied aspects of complex
systems – cutting across all traditional disciplines of the natural and life sciences,
engineering, economics, medicine, neuroscience, social and computer science.
Complex Systems are systems that comprise many interacting parts with the abil-
ity to generate a new quality of macroscopic collective behavior the manifestations
of which are the spontaneous formation of distinctive temporal, spatial or functional
structures. Models of such systems can be successfully mapped onto quite diverse
“real-life” situations like the climate, the coherent emission of light from lasers,
chemical reaction-diffusion systems, biological cellular networks, the dynamics of
stock markets and of the internet, earthquake statistics and prediction, freeway traf-
fic, the human brain, or the formation of opinions in social systems, to name just
some of the popular applications.
Although their scope and methodologies overlap somewhat, one can distinguish
the following main concepts and tools: self-organization, nonlinear dynamics, syn-
ergetics, turbulence, dynamical systems, catastrophes, instabilities, stochastic pro-
cesses, chaos, graphs and networks, cellular automata, adaptive systems, genetic al-
gorithms and computational intelligence.
The two major book publication platforms of the Springer Complexity program
are the monograph series “Understanding Complex Systems” focusing on the vari-
ous applications of complexity, and the “Springer Series in Synergetics”, which is
devoted to the quantitative theoretical and methodological foundations. In addition
to the books in these two core series, the program also incorporates individual titles
ranging from textbooks to major reference works.
Understanding Complex Systems
Founding Editor: J.A. Scott Kelso

Future scientific and technological developments in many fields will necessarily depend upon com-
ing to grips with complex systems. Such systems are complex in both their composition – typically
many different kinds of components interacting simultaneously and nonlinearly with each other
and their environments on multiple levels – and in the rich diversity of behavior of which they are
capable.
The Springer Series in Understanding Complex Systems series (UCS) promotes new strategies
and paradigms for understanding and realizing applications of complex systems research in a wide
variety of fields and endeavors. UCS is explicitly transdisciplinary. It has three main goals: First,
to elaborate the concepts, methods and tools of complex systems at all levels of description and in
all scientific fields, especially newly emerging areas within the life, social, behavioral, economic,
neuro- and cognitive sciences (and derivatives thereof); second, to encourage novel applications
of these ideas in various fields of engineering and computation such as robotics, nano-technology
and informatics; third, to provide a single forum within which commonalities and differences in the
workings of complex systems may be discerned, hence leading to deeper insight and understanding.
UCS will publish monographs, lecture notes and selected edited contributions aimed at commu-
nicating new findings to a large multidisciplinary audience.

Editorial and Programme Advisory Board

Péter Érdi
Center for Complex Systems Studies, Kalamazoo College, USA
and Hungarian Academy of Sciences, Budapest, Hungary

Karl Friston
Institute of Cognitive Neuroscience, University College London,
London, UK

Hermann Haken
Center of Synergetics, University of Stuttgart, Stuttgart, Germany

Janusz Kacprzyk
System Research, Polish Academy of Sciences, Warsaw, Poland

Jürgen Kurths
Nonlinear Dynamics Group, University of Potsdam,
Potsdam, Germany

Linda Reichl
Center for Complex Quantum Systems, University of Texas,
Austin, USA

Peter Schuster
Theoretical Chemistry and Structural Biology, University of Vienna, Vienna, Austria

Frank Schweitzer
System Design, ETH Zurich, Zurich, Switzerland

Didier Sornette
Entrepreneurial Risk, ETH Zurich, Zurich, Switzerland
Bernd Krauskopf . Hinke M. Osinga . Jorge
Galán-Vioque (Eds.)

Numerical Continuation
Methods for Dynamical
Systems
Path following and boundary value problems

With 200 Figures

Dedicated to Eusebius J. Doedel for his 60th birthday

ABC
Dr. Bernd Krauskopf Dr. Hinke M. Osinga
Dept of Engineering Mathematics Dept of Engineering Mathematics
University of Bristol University of Bristol
Bristol Bristol
BS8 1TR BS8 1TR
United Kingdom United Kingdom

Dr. Jorge Galán-Vioque

Applied Mathematics II
University of Sevilla
Escuela Superior de Ingenieros
Camino de los Descubrimientos, s/n,
41092 Sevilla
Spain

A C.I.P. Catalogue record for this book is available from the Library of Congress
ISSN 1860-0832
ISBN 978-1-4020-6355-8 (HB)
ISBN 978-1-4020-6356-5 (e-book)

Published by Springer,
P.O. Box 17, 3300 AA Dordrecht, The Netherlands
In association with
Canopus Publishing Limited,
27 Queen Square, Bristol BS1 4ND, UK
www.springer.com and www.canopusbooks.com
All Rights Reserved
© Canopus Publishing Limited 2007
No part of this work may be reproduced, stored in a retrieval system, or transmitted in
any form or by any means, electronic, mechanical, photocopying, microfilming, record-
ing or otherwise, without written permission from the Publisher, with the exception of
any material supplied specifically for the purpose of being entered and executed on a
computer system, for exclusive use by the purchaser of the work.
A Continuing Influence in Dynamics

A well-established method for studying a given dynamical system is to identify

the compact invariant objects, such as equilibria, periodic orbits and invariant
tori, and to consider the local behavior around them. This local information
then needs to be assembled in a consistent way, frequently with the help of
geometric and topological arguments, to obtain a unified global picture of the
system. The aim is to find qualitative (and often also quantitative) repre-
sentations of the different types of behavior that the system may exhibit in
dependence of key parameters. The main result of such an effort is a bifur-
cation diagram, that is, information on the division of parameter space into
regions of topologically different behavior together with representative phase
portraits. The list of theoretical tools one may employ is long, well developed
and dates back at least to the 19th century. However, even when one consid-
ers seemingly simple systems, theoretical tools need to be supplemented with
numerical calculations.
Mainly for technological reasons, numerical methods have a shorter history
than theoretical tools. The first and commonly used tool is numerical time
integration, which allows one to explore the dynamics by solving a (possibly
large) number of initial value problems. This approach is very practical for
the representation of chaotic attractors, and especially their ‘fingerprints’ in
a suitable Poincaré section. However, when it comes to the study of how the
behavior changes as a function of parameters the tool of choice is numerical
continuation — one also speaks of path following or homotopy methods. The
basic idea is to compute an implicitly defined curve of a suitable system of
equations that defines the dynamical object under consideration. In its basic
form, numerical continuation implements the stability and bifurcation theory
of equilibria of differential equations. More global objects, such as periodic
and homoclinic orbits, and their bifurcations can be computed by setting up
defining equations in the form of boundary value problems.
Path following in combination with boundary value problem solvers has
emerged as a continuing and strong influence in the development of dynamical
systems theory and its application in many diverse fields of science. It is widely
acknowledged that the software package Auto — developed by Eusebius J.
Doedel about thirty years ago and further expanded and developed ever since
— plays a central role in the brief history of numerical continuation. When
we were thinking how best to present the origin and development of Auto,
VI

a copy of the first edition of the Auto86 manual, with its authentic ochre
Caltech cover, came in very handy. We quote from the preface, dated May
1986:
The Auto package was first written in 1979. It was based on a
related program written in 1976 while the author was working with H.B
Keller at the California Institute of Technology. A first publication
referring to the package by its current name appeared in [22].
Applications often revealed some inadequacy in the algorithms and
resulted in changes. The applications also pointed to additional capa-
bilities that would be useful to have integrated in the package, and effort
was spent on making it easy to use. This explains the delay in publica-
tion of an extensive account of the algorithm implemented. Indeed, the
difference in effort between a theoretical analysis of a new method and
its implementation and integration appears to be considerable. We are
confident, however, that the methods and software presented here will
be of some use in the numerical exploration of nonlinear phenomena
in ordinary differential equations.
This quote not only highlights the intricate interplay at the very earliest stage
between the development of the software and applications, but it also contains
a major understatement: Auto has not just been “of some use”, but it has
been used by many hundreds of researchers from all around the world! To
give a rough idea of its impact in the general scientific community, ISI Web of
Knowledge reveals that the different versions of the Auto manual, which was
never published other than as a Caltech preprint, has more than 700 citations.
Similarly, the seminal reference [22] in the quote, the paper E.J. Doedel, Auto:
A program for the automatic bifurcation analysis of autonomous systems,
Cong. Num. 30 (Proc. 10th Manitoba Conf. Num. Math. and Comp.), 1981,
265–284, has more than 400 citations.
This book has been compiled on the occasion of Sebius Doedel’s 60th
birthday with the aim to illustrate the power and versatility of numerical
continuation techniques. As is demonstrated in the chapters of this book,
many recent developments build on the ideas of Sebius Doedel as implemented
in the package Auto, whose core of path following routine and collocation
boundary value problem solver is essentially still the same as when it was
released in 1986. It lies in the nature of the subject and the versatility of
Sebius Doedel’s work that we had to make a choice about which topics to
include. The emphasis of this book is on continuation methods for different
types of systems and dynamical objects, and on examples of how numerical
bifurcation analysis can be used in concrete applications. While recognizing
that there are other topics that could have been included, we believe that this
choice is in the spirit of the original motivation for the development of Auto
as expressed in the above quote. In this way, we hope to give an impression
of the continuing influence and future potential of these powerful numerical
methods for the bifurcation analysis of different types of dynamical systems.
 A Continuing Influence in Dynamics VII

The book opens with an extended foreword by Herb Keller, who is widely
recognized as the founding father of numerical continuation. Chapter 1 is an
edited part of lecture notes that Sebius Doedel has been using in his own
courses. It introduces the basic concepts of numerical bifurcation analysis and
forms a basis for the remainder of the book. The other eleven chapters by
leading experts focus on selected topics that have been influenced strongly by
Sebius Doedel’s work. In fact, at least half of the chapters discuss research
in which he has been involved as a co-author. Chapter 2 by Willy Govaerts
and Yuri Kuznetsov surveys recent developments of interactive continuation
tools. Chapter 3 by Mike Henderson is concerned with higher-dimensional
continuation, and Chap. 4 by Bernd Krauskopf and Hinke Osinga discusses
the computation of invariant manifolds with a continuation approach. The
next three chapters are devoted to applications. In Chap. 5 Don Aronson and
Hans Othmer consider the dynamics of a SQUID consisting of two Josephson
junctions. Chapter 6 by Sebastian Wieczorek discusses global bifurcations
in laser systems, and Chap. 7 by Emilio Freire and Alejandro Rodrı́guez-
Luis demonstrates the use of numerical bifurcation analysis for the study of
electronic circuits. The remaining chapters deal with continuation for spe-
cial types of dynamical systems. Chapter 8 by John Guckenheimer and Drew
LaMar is concerned with slow-fast systems, and Chap. 9 by Jorge Galán-
Vioque and André Vanderbauwhede with symmetric Hamiltonian systems.
Spatially extended systems are the topic of Chap. 10 by Wolf-Jürgen Beyn
and Vera Thümmler and of Chap. 11 by Alan Champneys and Björn Sand-
stede. Finally, in Chap. 12 Dirk Roose and Róbert Szalai survey numerical
continuation techniques for systems with delay.
We are very grateful for the enthusiastic support from all who were in-
volved in this book project. First of all, we thank all authors for their con-
tributions and for making every effort to stay within the limits of a tight
production schedule. We also thank Tom Spicer of Canopus Publishing Ltd
for his support of this project from its conception to the final production of
the book. Last, but not least, we would like to thank Sebius Doedel for his
support over many years of collaboration, and for agreeing to the publication
of Chap. 1 without knowing exactly what we were up to. Happy birthday,
Sebius!

Bernd Krauskopf, Hinke Osinga and Jorge Galán-Vioque

Bristol and Sevilla, March 2007.
Foreword

Herbert B Keller

California Institute of Technology, Pasadena, USA, and University of California,

San Diego, USA.

Sebius (diminutive for Eusebius) Doedel obtained his Ph.D. in Applied Math-
ematics from the University of British Columbia in 1976. His advisor was my
friend and ex-colleague Jim Varah. As a consequence, I was able to employ
Sebius as a Research Fellow in Applied Mathematics at Caltech in 1975. Over
the next 26 years, he spent 13 of them at Caltech. However, he was also much
appreciated at Concordia University, where he was employed in 1979 and
rapidly rose to Professor of Computer Science, winning many awards which
fortunately included several years on leave with pay!
We cycled together occasionally, even in Holland, where Sebius was born.
I remember one ride in particular, when on a rather warm day we went east
from Pasadena to Claremont, about 28 miles each way. Somewhere along the
way, I became quite thirsty and so we stopped to get a cool drink. We sat
outside, relaxed and, I thought, enjoyed our drinks. I started to wax poetic
about how nice it was enjoying the outdoors and lovely California weather
when Sebius said: “Herb, do you know where we are?” I said: “sure, near
Claremont.” He replied: “This is Pomona, the drive-by shooting capital of the
world — I can’t wait to get out of here.” I have never again enjoyed going
past that part of our ride.
Early during his first appointment at Caltech, Sebius became interested
in bifurcation phenomena, two-point boundary value problems and numerical
path-following or continuation methods, perhaps as a result of sitting in on my
course in these areas. In 1976, I was writing the paper [17] in which pseudoar-
clength continuation was introduced and Sebius was willing to do some calcu-
lations to illustrate how these methods worked. Of course, he did a wonderful
job producing all of the results in §7 of that paper, but more importantly, as
a result, he essentially started working on Auto at that time. The first publi-
cation on Auto appeared in 1981 [7]. It has evolved dramatically since then,
culminating in Auto2000 [22], a fully parallel code in C++ with great graph-
ics (available for free via http://sourceforge.net/projects/auto2000/)
that was produced mainly by Randy Paffenroth, working at Caltech under
Sebius’s direction.
X Herbert B Keller

Auto is without a doubt the most powerful and efficient tool for determin-
ing the bifurcation structure of nonlinear parameter-dependent systems of al-
gebraic and ordinary differential equations. Influenced by his colleagues at the
University of British Columbia, who developed COLSYS [5], Sebius has em-
ployed orthogonal collocation approximations and mesh refinement to obtain
extremely high accuracy. The code is able to determine heteroclinic, homo-
clinic and periodic orbits, both stable and unstable, by means of a two-point
boundary value problem formulation. Using a brilliant elimination procedure,
the relatively sparse Jacobian is reduced to a low-dimensional dense matrix
from which the Floquet multipliers are computed. The bifurcation structure
at singular points is readily determined in this way.
The development of Auto is but one of the main projects that Sebius
has undertaken. In the course of this work powerful theoretical results have
been produced, many with colleagues and his students, in the general areas
of bifurcation theory, dynamical systems, periodic orbits, delay differential
equations, collocation methods for nonlinear elliptic PDEs, coupled oscillator
theory, control of bifurcation phenomena, continuation theory of manifolds,
and numerous additional topics. However, there is no doubt that the Auto
software has had a tremendous impact on many applied mathematics areas
and is, indeed, one of the leading tools in scientific computing. The code has
been incorporated into many other large software systems that solve nonlinear
problems involving continuation and bifurcation phenomena.
Essentially, all of the authors contributing to this volume have been coau-
thors with Sebius on papers related to the work presented here. However,
I would like to point out a few of my favorite contributions made through
these collaborations, not all of which have been fully appreciated yet. A bril-
liant contribution is contained in a paper by Wolf-Jürgen Beyn and Sebius
Doedel [6], in which it is shown that a continuous nonlinear boundary value
problem and the corresponding discretized problem have the same number of
solutions for all sufficiently fine meshes.
Sebius introduced a very powerful technique to keep computed families of
periodic solutions of autonomous differential equations in phase. The idea is
simply to minimize the ‘distance’ between neighboring solutions with respect
to a change in phase. That is, if u(t, λ) is the solution over the normalized
period 0 ≤ t ≤ 1 at parameter value λ, then the neighboring solution u(t, λ+δ)
with phase shift θ lies at distance
Z 1
D2 (θ) = || u(t + θ, λ + δ) − u(t, λ) ||2 dt.
0

We seek to minimize this distance with respect to the phase shift θ. Standard
calculus of variations near zero leads to the integral condition
Z 1
u̇∗ (t, λ + δ) u(t, λ) dt = 0.
0
 Foreword XI

This is a generalization of the standard Poincaré phase or transversality condi-

tion that is applied only at one point on the orbit. However, the above global
condition is much more robust in calculations, as has been shown in many
examples [9, 10, 12] (the Poincaré condition remains preferable for analytical
proofs). I am not sure when this global condition first appeared in the liter-
ature, but we have referred to it in [16] as having been introduced by Sebius
in 1981 [7], which also happens to be his first publication on Auto.
Many of Sebius’s publications have to do with periodic solutions of dy-
namical systems. These arise in a great variety of applications starting with
chemical reactors [23], then on to systems of oscillators [2, 21], heteroclinic
orbits [9] in which the above phase condition is crucial, resonances in excitable
systems [1] such as forced Fitzhugh-Nagumo systems, current biased and cou-
pled Josephson junctions [3, 4], delay differential equations [11, 13, 14], mod-
ified Van der Pol oscillators [8], conservative and Hamiltonian systems [20],
cardiac pacemakers [19], the circular restricted three-body problem and the
figure-eight orbit of Chenciner and Montgomery [12, 15], and many more. A
large number of these contributions are in the bio-physics area and, thus, it
turns out that Sebius may be a closet biologist.
More recently, Sebius has returned with others to the important problem
of computing higher-dimensional manifolds, either stable or unstable [18].
This brief account of some of Sebius’s publications and obvious collabo-
rations does not do justice to the impact he has had in the field of scientific
computation. He has had numerous students, extremely well trained, and
now making their own contributions. Furthermore, he has worked with many
outstanding scientists and has invariably enhanced their ability to do signifi-
cant scientific computations so much that it would be difficult to measure his
tremendous influence in our field. Hopefully, he will continue as he reaches
maturity.

H. B. Keller
Caltech / UCSD
November, 2006
XII Herbert B Keller

References
1. J. C. Alexander, E. J. Doedel, and H. G. Othmer. On the resonance structure
in a forced excitable system. SIAM J. Appl. Math. 50(5):1373–1418, 1990.
2. D. G. Aronson, E. J. Doedel, and H. G. Othmer. An analytical and numerical
study of the bifurcations in a system of linearly-coupled oscillators. Physica D
25(1-3):20–104, 1987.
3. D. G. Aronson, E. J. Doedel, and H. G. Othmer. The dynamics of coupled
current-biased Josephson junctions. II. Internat. J. Bifur. Chaos Appl. Sci.
Engrg., 1(1): 51–66, 1991.
4. D. G. Aronson, E. J. Doedel, and D. H. Terman. A codimension-two point
associated with coupled Josephson junctions. Nonlinearity 10(5):1231–1255,
1997.
5. U. Ascher, J. Christiansen, and R. Russell. A collocation solver for mixed order
systems of boundary value problems. Math. Comp., 33(146):659–679, 1979.
6. W.-J. Beyn and E. J. Doedel. Stability and multiplicity of solutions to dis-
cretizations of nonlinear ordinary differential equations. SIAM J. Sci. Comp.,
2(1):107–120, 1981.
7. E. J. Doedel. Auto: a program for the automatic bifurcation analysis of au-
tonomous systems. Congr. Numer., 30:265–384, 1981.
8. E. J. Doedel, E. Freire, E. Gamero, and A. J Rodríguez-Luis. An analytical
and numerical study of a modified Van der Pol oscillator. J. Sound Vibration,
256(4):755–771, 2002.
9. E. J. Doedel and M. Friedman. Numerical computation of heteroclinic orbits.
Continuation techniques and bifurcation problems. J. Comput. Appl. Math.,
26(1-2):155–170, 1989.
10. E. J. Doedel, H. B. Keller, and J.-P. Kernevez. Numerical analysis and control
of bifurcation problems II: Bifurcation in infinite dimensions. Internat. J. Bifur.
Chaos Appl. Sci. Engrg., 1(4): 745–772, 1991.
11. E. J. Doedel and P. C. Leung. Numerical techniques for bifurcation problems
in delay equations. Congr. Numer., 34:225–237, 1982.
12. E. J. Doedel, R. C. Paffenroth, H. B. Keller, D. J. Dichmann, J. Galán-Vioque,
and A. Vanderbauwhede. Computation of periodic solutions of conservative
systems with applications to the 3-body problem. Internat. J. Bifur. Chaos
Appl. Sci. Engrg., 13(6): 1353–1381, 2003.
13. K. Engelborghs and E. J. Doedel. Convergence of a boundary value difference
equation for computing periodic solutions of neutral delay differential equations.
J. Differ. Equations Appl., 7(6):927–940, 2001.
14. K. Engelborghs and E. J. Doedel. Stability of piecewise polynomial collocation
for computing periodic solutions of delay differential equations. Numer. Math.,
91(4):627–648, 2002.
15. J. Galán-Vioque, F. J. Muñoz-Almaraz, E. Freire, E. J. Doedel, and A. Vander-
bauwhede. Stability and bifurcations of the figure-8 solution of the three-body
problem. Phys. Rev. Lett. 88(24):241101, 2002.
16. A. D. Jepson and H. B. Keller. Steady state and periodic solution paths: their
bifurcations and computations. In T. Kupper and H. D. Mittleman, editors,
Numerical Methods for Bifurcation Problems, pages 219–246. (Birkäuser Verlag,
1984).
 Foreword XIII

17. H. B. Keller. Numerical solution of bifurcation and nonlinear eigenvalue prob-

lems. In P. H. Rabinowitz, editor, Applications of Bifurcation Theory, pages
359–384. (Academic Press, New York, 1978).
18. B. Krauskopf, H. M. Osinga, E. J. Doedel, M. E. Henderson, J. Guckenheimer,
A. Vladimirsky, M. Dellnitz, and O. Junge. A survey of methods for comput-
ing (un)stable manifolds of vector fields. Internat. J. Bifur. Chaos Appl. Sci.
Engrg., 15(3):763–791, 2005.
19. T. Krogh-Madsen, L. Glass, E. J. Doedel, and M. R. Guevara. Apparent dis-
continuities in the phase-resetting response of cardiac pacemakers. J. Theor.
Biol., 230(4):499–519, 2004.
20. F. J. Muñoz-Almaraz, E. Freire, J. Galán, and E. J. Doedel. Continuation of
periodic orbits in conservative and Hamiltonian systems. Physica D, 181(1-
2):1–38, 2003.
21. H. G. Othmer, D. G. Aronson, and E. J. Doedel. Resonance and bistability in
coupled oscillators. Phys. Lett. A, 113(7):349–354, 1986.
22. R. C. Paffenroth and E. J. Doedel. The Auto2000 command line user interface.
In Ninth International Python Conference (Long Beach, California, USA 2001),
pages 233-241, 2001. Available via http://cmvl.cs.concordia.ca/.
23. A. B. Poore, E. J. Doedel, and J. E. Cermak. Dynamics of the Iwan-Blevins
Wake Oscillator Model. Internat. J. Non-Linear Mechanics, 21(4):291–302,
1986.
Contents

A Continuing Influence in Dynamics

................................................................ V
Foreword
Herbert B Keller . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . IX
1 Lecture Notes on Numerical Analysis of Nonlinear
Equations
Eusebius J Doedel . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
2 Interactive Continuation Tools
Willy Govaerts, Yuri A Kuznetsov . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51
3 Higher-Dimensional Continuation
Michael E Henderson . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 77
4 Computing Invariant Manifolds via the Continuation of
Orbit Segments
Bernd Krauskopf, Hinke M Osinga . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 117
5 The Dynamics of SQUIDs and Coupled Pendula
Donald G Aronson, Hans G Othmer . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 155
6 Global Bifurcation Analysis in Laser Systems
Sebastian M Wieczorek . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 177
7 Numerical Bifurcation Analysis of Electronic Circuits
Emilio Freire, Alejandro J Rodrı́guez-Luis . . . . . . . . . . . . . . . . . . . . . . . . . . . 221
8 Periodic Orbit Continuation in Multiple Time Scale
Systems
John Guckenheimer, M Drew LaMar . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 253
XVI Contents

9 Continuation of Periodic Orbits in Symmetric Hamiltonian

Systems
Jorge Galán-Vioque, André Vanderbauwhede . . . . . . . . . . . . . . . . . . . . . . . . . 269
10 Phase Conditions, Symmetries and PDE Continuation
Wolf-Jürgen Beyn, Vera Thümmler . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 301
11 Numerical Computation of Coherent Structures
Alan R Champneys, Björn Sandstede . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 331
12 Continuation and Bifurcation Analysis of Delay Differential
Equations
Dirk Roose, Robert Szalai . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 359
List of Contributors

Don Aronson Montreal, QC H3G 1M8

School of Mathematics and Canada
Institute for Mathematics [email protected]
and its Applications
University of Minnesota Emilio Freire
Minneapolis, MN 55455 Departamento de Matemática
USA Aplicada II
[email protected] Escuela Superior de Ingenieros
Universidad de Sevilla
Wolf-Jürgen Beyn Camino de los Descubrimientos s/n
Department of Mathematics 41092 Sevilla
University of Bielefeld Spain
P.O. Box 100131 [email protected]
33501 Bielefeld Jorge Galán-Vioque
Germany Departamento de Matemática
[email protected] Aplicada II
Escuela Superior de Ingenieros
Alan R Champneys Universidad de Sevilla
Department of Engineering Camino de los Descubrimientos s/n
Mathematics 41092 Sevilla
University of Bristol Spain
Queen’s Building [email protected]
Bristol BS8 1TR
United Kingdom Willy Govaerts
[email protected] Deptartment of Applied Mathemat-
ics and Computer Science
Eusebius J Doedel Ghent University
Department of Computer Science Krijgslaan 281 -S9
Concordia University B-9000 Gent
1455 De Maisonneuve Blvd. West Belgium
EV 3285 [email protected]
XVIII List of Contributors

Michael E Henderson University of Minnesota

IBM Watson Research Center Minneapolis, MN 55455
P.O. Box 218 USA
33-215 [email protected]
Yorktown Heights, NY 10598
USA Alejandro J Rodrı́guez-Luis
[email protected] Departamento de Matemática
Aplicada II
Herbert B Keller Escuela Superior de Ingenieros
Applied and Computational Universidad de Sevilla
Mathematics Camino de los Descubrimientos s/n
California Institute of Technology 41092 Sevilla
MC 217-50 Spain
Pasadena, CA 91125 [email protected]
USA
[email protected] Dirk Roose
Department of Computer Science
Bernd Krauskopf KU Leuven
Department of Engineering Celestijnenlaan 200A
Mathematics B-3001 Heverlee - Leuven
University of Bristol Belgium
Queen’s Building [email protected]
Bristol BS8 1TR
Björn Sandstede
United Kingdom
Department of Mathematics
[email protected]
University of Surrey
Guildford, GU2 7XH
Yuri A Kuznetsov United Kingdom
Department of Mathematics [email protected]
Budapestlaan 6
3584 CD Utrecht Robert Szalai
The Netherlands Department of Engineering
[email protected] Mathematics
University of Bristol
Hinke M Osinga Queen’s Building
Department of Engineering Bristol BS8 1TR
Mathematics United Kingdom
University of Bristol [email protected]
Queen’s Building
Bristol BS8 1TR Vera Thümmler
United Kingdom Department of Mathematics
[email protected] University of Bielefeld
P.O. Box 100131
Hans Othmer 33501 Bielefeld
School of Mathematics and Germany
Digital Technology Center [email protected]
 List of Contributors XIX

André Vanderbauwhede Sebastian M Wieczorek

Department of Pure Mathematics Mathematics Research Institute
and Computer Algebra University of Exeter
Ghent University Harrison Building
Krijgslaan 281 North Park Road
B-9000 Gent Exeter EX4 4QF
Belgium United Kingdom
[email protected] [email protected]
1
Lecture Notes on Numerical Analysis of
Nonlinear Equations

Eusebius J Doedel

Department of Computer Science, Concordia University, Montreal, Canada

Numerical integrators can provide valuable insight into the transient behavior
of a dynamical system. However, when the interest is in stationary and peri-
odic solutions, their stability, and their transition to more complex behavior,
then numerical continuation and bifurcation techniques are very powerful and
efficient.
The objective of these notes is to make the reader familiar with the ideas
behind some basic numerical continuation and bifurcation techniques. This
will be useful, and is at times necessary, for the effective use of the software
Auto and other packages, such as XppAut [17], Content [24], Matcont
[21], and DDE-Biftool [16], which incorporate the same or closely related
algorithms.
These lecture notes are an edited subset of material from graduate courses
given by the author at the universities of Utah and Minnesota [9] and at
Concordia University, and from short courses given at various institutions,
including the Université Pierre et Marie Curie (Paris VI), the Centre de
Recherches Mathématiques of the Université de Montréal, the Technische
Universität Hamburg-Harburg, and the Benemérita Universidad Autónoma
de Puebla.

1.1 The Implicit Function Theorem

Before starting our discussion of numerical continuation of solutions to nonlin-

ear equations, it is important first to discuss under what conditions a solution
will actually persist when problem parameters are changed. Therefore, we
begin with an overview of the basic theory. The Implicit Function Theorem
(IFT) is central to our analysis and we discuss some examples. The discus-
sion in this section follows the viewpoint of Keller in graduate lectures at the
California Institute of Technology, a subset of which was published in [23].
2 Eusebius J Doedel

1.1.1 Basic Theory

Let B denote a Banach space, that is, a complete, normed vector space. In
the presentation below it will be implicitly assumed that B is Rn , although
the results apply more generally. For x0 ∈ B, we denote by Sρ (x0 ) the closed
ball of radius ρ centered at x0 , that is,

Sρ (x0 ) = {x ∈ B | || x − x0 ||≤ ρ}.

Existence and uniqueness of solutions is obtained by using two theorems.

Theorem 1 (Contraction Theorem). Consider a continuous function F :
B → B on a Banach space B and suppose that for some x0 ∈ B, ρ > 0, and
some K0 with 0 ≤ K0 < 1, we have

|| F (u) − F (v) || ≤ K0 || u − v ||, for all u, v ∈ Sρ (x0 ),

|| F (x0 ) − x0 || ≤ (1 − K0 ) ρ.

Then the equation

x = F (x), x ∈ B,
has one and only one solution x∗ ∈ Sρ (x0 ), and x∗ is the limit of the sequence

xk+1 = F (xk ), k = 0, 1, 2, . . . .

Proof. Let x1 = F (x0 ). Then

|| x1 − x0 ||=|| F (x0 ) − x0 ||≤ (1 − K0 ) ρ ≤ ρ.

Thus, x1 ∈ Sρ (x0 ). Now assume inductively that x0 , x1 , · · · , xn ∈ Sρ (x0 ).

Then for k ≤ n we have

|| xk+1 − xk || = || F (xk ) − F (xk−1 ) || ≤ K0 || xk − xk−1 ||

= ··· ≤ K0k || x1 − x0 ||
k
≤ K0 (1 − K0 ) ρ.

Thus,

|| xn+1 − x0 || ≤ || xn+1 − xn || + || xn − xn−1 || + · · · + || x1 − x0 ||

≤ (K0n + K0n−1 + · · · + 1) (1 − K0 ) ρ
= (1 − K0n+1 ) ρ
≤ ρ.

Hence xn+1 ∈ Sρ (x0 ), and by induction xk ∈ Sρ (x0 ) for all k. We now show
that {xk } is a Cauchy sequence:
 1 Numerical Analysis of Nonlinear Equations 3

|| xk+n − xk || ≤ || xk+n − xk+n−1 || + · · · + || xk+1 − xk ||

≤ (K0n−1 + K0n−2 + · · · + 1) K0k (1 − K0 ) ρ
= (1 − K0n ) K0k ρ
≤ K0k ρ.
1
For given ε > 0, choose k such that K0k ρ < 2 ε. Then

|| xk+ℓ − xk+m ||≤|| xk+ℓ − xk || + || xk+m − xk ||≤ 2K0k ρ < ε,

independently of ℓ and m. Hence, {xk } is a Cauchy sequence and, therefore,

converges to a unique limit lim xk = x∗ , where x∗ ∈ Sρ (x0 ). Since we assumed
that F is continuous, we have

x∗ = lim xk = lim F (xk−1 ) = F (lim xk−1 ) = F (lim xk ) = F (x∗ ).

This proves the existence of x∗ . We get uniqueness as follows. Suppose there

are two solutions, say, x, y ∈ Sρ (x0 ) with x = F (x) and y = F (y). Then

|| x − y ||=|| F (x) − F (y) ||≤ K0 || x − y || .

Since K0 < 1, this is a contradiction. ⊔

⊓

The second theorem ensures the parameter-dependent existence of a solution.

Theorem 2 (Implicit Function Theorem). Let G : B × Rm → B satisfy:
• G(u0 , λ0 ) = 0 for u0 ∈ B and λ0 ∈ Rm ;
• Gu (u0 , λ0 ) is nonsingular with bounded inverse,

|| Gu (u0 , λ0 )−1 ||≤ M

for some M > 0;

• G and Gu are Lipschitz continuous, that is, for all u, v ∈ Sρ (u0 ), and for
all λ, µ ∈ Sρ (λ0 ) the following inequalities hold for some KL > 0:

|| G(u, λ) − G(v, µ) || ≤ KL (|| u − v || + || λ − µ ||),

|| Gu (u, λ) − Gu (v, µ) || ≤ KL (|| u − v || + || λ − µ ||).

Then there exists δ, with 0 < δ ≤ ρ, and a unique function u(λ) that is
continuous on Sδ (λ0 ), with u(λ0 ) = u0 , such that

G(u(λ), λ) = 0, for all λ ∈ Sδ (λ0 ).

If G(u, λ0 ) = 0 and if Gu (u0 , λ0 ) is invertible with bounded inverse, then

u0 is called an isolated solution of G(u, λ0 ) = 0. Hence, the IFT states that
isolation (plus Lipschitz continuity assumptions) implies the existence of a
locally unique solution family (or solution branch) u = u(λ), with u(λ0 ) =
u0 .
4 Eusebius J Doedel

Proof. We use the notation G0u = Gu (u0 , λ0 ). Then we rewrite the problem
as

G(u, λ) = 0 ⇔ G0u u = G0u u − G(u, λ)

−1
⇔ u = G0u [G0u u − G(u, λ)] .
| {z }
≡F(u,λ)

Hence, G(u, λ) = 0 if and only if u is a fixed point of F(·, λ). (Note that the
corresponding fixed point iteration is, in fact, the Chord Method for solving
G(u, λ) = 0.) We must verify the conditions of the Contraction Theorem.
Pick u, v ∈ Sρ1 (u0 ), and any fixed λ ∈ Sρ1 (λ0 ), where ρ1 is to be chosen
later. Then

−1  0
F(u, λ) − F(v, λ) = G0u Gu [u − v] − [G(u, λ) − G(v, λ)] . (1.1)

By the Fundamental Theorem of Calculus, we have

Z 1
d
G(u, λ) − G(v, λ) = G(tu + (1 − t)v, λ) dt
0 dt
Z 1
= Gu (tu + (1 − t)v, λ) dt [u − v]
0
= Ĝu (u, v, λ) [u − v],

where in the last step we used the Mean Value Theorem to get Ĝ. Then (1.1)
becomes

|| F(u, λ) − F(v, λ) ||
≤ M || G0u − Ĝu (u, v, λ) || || u − v ||
Z 1
= M || Gu (u0 , λ0 ) − Gu (tu + (1 − t)v, λ) dt || || u − v ||
0
Z 1
≤M || Gu (u0 , λ0 ) − Gu (tu + (1 − t)v, λ) || dt || u − v ||
0
Z 1
≤M KL (|| u0 − (tu + (1 − t)v) || + || λ0 − λ ||) dt || u − v ||
0 | {z }
∈Sρ1 (u0 )

≤ M KL 2ρ1 || u − v || .
| {z }
≡K0

Therefore, if we take
1
ρ1 < ,
2M KL
then K0 < 1. The second condition of the Contraction Theorem is also satis-
fied, namely,
 1 Numerical Analysis of Nonlinear Equations 5

|| F(u0 , λ) − u0 ||
= || F(u0 , λ) − F(u0 , λ0 ) ||

−1
−1
= || G0u [G0u u0 − G(u0 , λ)] − G0u [G0u u0 − G(u0 , λ0 )] ||

−1
= || G0u [G(u0 , λ0 ) − G(u0 , λ)] ||
≤ M KL || λ − λ0 ||
≤ M KL ρ,

where ρ (with 0 < ρ ≤ ρ1 ) is to be chosen. We want the above to be less than

or equal to (1 − K0 )ρ1 , so we choose

(1 − K0 ) ρ1
ρ≤ .
M KL

Hence, for each λ ∈ Sρ (λ0 ) we have a unique solution u(λ). We now show that
u(λ) is continuous in λ. Let λ1 , λ2 ∈ Sρ (λ0 ), with corresponding solutions
u(λ1 ) and u(λ2 ). Then

|| u(λ1 ) − u(λ2 ) ||
= || F(u(λ1 ), λ1 ) − F(u(λ2 ), λ2 ) ||
≤ || F(u(λ1 ), λ1 ) − F(u(λ2 ), λ1 ) || + || F(u(λ2 ), λ1 ) − F(u(λ1 ), λ2 ) ||
≤ K0 || u(λ1 ) − u(λ2 ) || +

−1
−1
|| G0u [G0u u(λ2 ) − G(u(λ2 ), λ1 )] − G0u [G0u u(λ2 ) − G(u(λ2 ), λ2 )] ||
≤ K0 || u(λ1 ) − u(λ2 ) || +M KL || λ1 − λ2 || .
|{z}
<1

Hence,
M KL
|| u(λ1 ) − u(λ2 ) ||≤ || λ1 − λ2 ||,
1 − K0
which concludes the proof of the IFT. ⊔
⊓
So far, under mild assumptions, we have shown that there exists a locally
unique solution family u(λ). If we impose the condition that F(u, λ) is con-
tinuously differentiable in λ, then we can show that u(λ) is also continuously
differentiable. To this end, the Banach Lemma is very useful.
Lemma 1 (Banach Lemma). Let L : B → B be a linear operator with
|| L ||< 1. Then (I + L)−1 exists and
1
|| (I + L)−1 ||≤ .
1− || L ||

Proof. Suppose I + L is not invertible. Then there exists y ∈ B, y 6= 0, such

that
(I + L) y = 0.
6 Eusebius J Doedel

Thus, y = −Ly and

|| y ||=|| Ly ||≤|| L || || y ||<|| y ||,
which is a contradiction. Therefore, (I +L)−1 exists. We can bound the inverse
as follows:

(I + L)(I + L)−1 = I
⇔ (I + L)−1 = I − L(I + L)−1
⇔|| (I + L)−1 || ≤ 1+ || L || || (I + L)−1 ||
1
⇔|| (I + L)−1 || ≤ .
1− || L ||
This proves the Banach Lemma. ⊔
⊓
The Banach Lemma can be used to show the following.
Lemma 2. Under the conditions of the IFT, there exists M1 > 0 and δ > 0
such that Gu (u, λ)−1 exists and || Gu (u, λ)−1 ||≤ M1 in Sδ (u0 ) × Sδ (λ0 ).
Proof. Using again the notation G0u = Gu (u0 , λ0 ), we have

Gu (u, λ) = G0u + Gu (u, λ) − G0u

= G0u [I + (G0u )−1 (Gu (u, λ) − G0u )].
| {z }
≡L

Similar to how we verified the second condition of the Contraction Theorem

in the proof of the IFT, we can show that

|| L ||≤ M KL (|| u − u0 || + || λ − λ0 ||) ≤ M KL 2δ.

As for the IFT, we choose

1
δ< ,
2M KL
and conclude that, therefore, (I + L)−1 exists and
1
|| (I + L)−1 ||≤ .
1 − M KL 2δ
Hence, Gu (u, λ)−1 exists and
M
|| Gu (u, λ)−1 ||=|| (G0u )−1 (I + L)−1 ||≤ ≡ M1 ,
1 − M KL 2δ
as required. ⊔
⊓
We are now ready to prove differentiability of the solution branch.
Theorem 3. In addition to the assumptions of the IFT, assume that the
derivative Gλ (u, λ) is continuous in Sρ (u0 ) × Sρ (λ0 ). Then the solution
branch u(λ) has a continuous derivative uλ (λ) on Sδ (u0 ) × Sδ (λ0 ).
 1 Numerical Analysis of Nonlinear Equations 7

Proof. Using the definition of (Fréchet) derivative, we are given that there
exists Gu (u, λ) such that G(u, λ)−G(v, λ) = Gu (u, λ) (u−v)+R1 (u, v, λ),
where R1 (u, v, λ) is such that

|| R1 (u, v, λ) ||
→ 0 as || u − v ||→ 0. (1.2)
|| u − v ||

Similarly, there exists Gλ (u, λ) such that G(u, λ) − G(u, µ) = Gλ (u, λ) (λ −

µ) + R2 (u, λ, µ), where R2 (u, λ, µ) satisfies

|| R2 (u, λ, µ) ||
→ 0 as || λ − µ ||→ 0. (1.3)
|| λ − µ ||

We must show that there exists uλ (λ) such that

u(λ) − u(µ) = uλ (λ) (λ − µ) + r(λ, µ),

with
|| r(λ, µ) ||
→ 0 as || λ − µ ||→ 0.
|| λ − µ ||
Now

0 = G(u(λ), λ) − G(u(µ), µ)
= G(u(λ), λ) − G(u(µ), λ) + G(u(µ), λ) − G(u(µ), µ)
= Gu (u(λ), λ) (u(λ) − u(µ)) + R1 (u(λ), u(µ), λ)
+Gλ (u(µ), λ) (λ − µ) + R2 (u(µ), λ, µ).

Lemma 2 guarantees the existence of Gu (u(λ), λ)−1 , and we find

u(λ) − u(µ) = −Gu (u(λ), λ)−1 [Gλ (u(µ), λ) (λ − µ) − (R1 + R2 )]

= −Gu (u(λ), λ)−1 [Gλ (u(λ), λ) (λ − µ) − r],

where
r = [Gλ (u(λ), λ) − Gλ (u(µ), λ)] (λ − µ) + R1 + R2 .
Let us, for the moment, ignore the harmless factor Gu (u(λ), λ)−1 and consider
each term of r. Since u and Gλ are continuous, we have

|| [Gλ (u(λ), λ) − Gλ (u(µ), λ)] (λ − µ) ||

→ 0 as || λ − µ ||→ 0.
|| λ − µ ||

Also, the existence of Gλ implies (1.3)

|| R2 (u(λ), λ, µ) ||
→ 0 as || λ − µ ||→ 0.
|| λ − µ ||

Using (1.2), we have

8 Eusebius J Doedel

|| R1 (u(λ), u(µ), λ) || || R1 (u(λ), u(µ), λ) || || u(λ) − u(µ) ||

= → 0,
|| λ − µ || || u(λ) − u(µ) || || λ − µ ||

as || λ − µ || → 0 because the second factor is bounded due to continuity of

u(λ) (see the end of the proof of the IFT). Thus,

uλ (λ) = −Gu (u(λ), λ)−1 Gλ (u(λ), λ).

To prove that uλ is continuous it suffices to show that Gu (u(λ), λ)−1 is

continuous. Indeed,

|| Gu (u(λ), λ)−1 − Gu (u(µ), µ)−1 ||

= || Gu (u(λ), λ)−1 [Gu (u(µ), µ) − Gu (u(λ), λ)] Gu (u(µ), µ)−1 ||
≤ M12 KL (|| u(µ) − u(λ) || + || µ − λ ||),

which concludes the proof of Theorem 3 ⊔

⊓

Remark 1. In fact, if Gλ is Lipschitz continuous then uλ is Lipschitz contin-

uous (we already assume that Gu is Lipschitz continuous). More generally,
it can be shown that uλ is C k if G is C k , that is, u inherits the degree of
continuity of G.

We now give some examples where the IFT is used to show that a given
solution persists, at least locally, when a problem parameter is changed. We
also identify some cases where the conditions of the IFT are not satisfied.

1.1.2 A Predator-Prey Model

Our first example is that of a predator-prey model defined as

 ′
u1 = 3u1 (1 − u1 ) − u1 u2 − λ(1 − e−5u1 ),
(1.4)
u′2 = −u2 + 3u1 u2 .

We can think of u1 as ‘fish’ and u2 as ‘sharks’, while the term λ(1−e−5u1 ) rep-
resents ‘fishing’, with ‘fishing-quota’ λ. When λ = 0 the stationary solutions
are 
3u1 (1 − u1 ) − u1 u2 = 0
⇒ (u1 , u2 ) = (0, 0), (1, 0), ( 13 , 2).
−u2 + 3u1 u2 =0
The Jacobian matrix is
 
3 − 6u1 − u2 − 5λe−5u1 −u1
J= = J(u1 , u2 ; λ).
u2 −1 + 3u1

Hence, we have
 1 Numerical Analysis of Nonlinear Equations 9

u1 3

5
8

4
9
1
1

2 lambda

u2

Fig. 1.1. Stationary solution branches of the predator-prey model (1.4). Solution 2
and solution 4 are branch points, while solution 8 is a Hopf bifurcation point.

 
3 0
J(0, 0; 0) = , eigenvalues 3, −1 (unstable);
0 −1
 
3 −1
J(1, 0; 0) = eigenvalues − 3, 2 (unstable);
,
0 2 

 (−1 − µ)(−µ) + 2 = 0 ⇔
   µ2 + µ + 2 = 0 ⇔
1 −1 − 13 √
J( 3 , 2; 0) = , eigenvalues
6 0 
 µ± = −1±2 −7 ;

Re(µ± ) < 0 (stable).

All three Jacobians at λ = 0 are nonsingular. Thus, by the IFT, all three
stationary points persist for (small) λ 6= 0. In this problem we can explicitly
find all solutions (see Fig. 1.1):
I: (u1 , u2 ) = (0, 0).

3u1 (1 − u1 ) 3(1 − 2u1 )

II: u2 = 0 and λ = . (Note that lim λ = lim = 35 .)
1 − e−5u1 u1 →0 u1 →0 5e−5u1
III: u1 = 1
3 and 2
3 − 13 u2 − λ(1 − e−5/3 ) = 0 ⇒ u2 = 2 − 3λ(1 − e−5/3 ).
These solution families intersect at two branch points, one of which is (u1 , u2 , λ) =
(0, 0, 3/5).
The stability of Branch I follows from:
 
3 − 5λ 0
J(0, 0; λ) = , eigenvalues 3 − 5λ, −1.
0 −1
10 Eusebius J Doedel

MAX U1
0.90

0.80
15
0.70 14

0.60 12

0.50 11 13
5
0.40 4
9

0.30
8
0.20

0.10 2
1
0.00

-0.10
0.00 0.20 0.40 0.60 0.80 1.00
0.10 0.30 0.50 0.70 0.90 lambda

Fig. 1.2. Bifurcation diagram of the predator-prey model (1.4). The periodic solu-
tion branch is also shown. For stationary solutions the vertical axis is simply u1 , while
for periodic solutions max(u1 ) is plotted. Solid/dashed lines denote stable/unstable
solutions. Open squares are branch points; the solid square is a Hopf bifurcation.

Hence, the trivial solution is unstable if λ < 3/5, and stable if λ > 3/5, as
indicated in Fig. 1.2. Branch II has no stable positive solutions. At λH ≈
0.67 on Branch III (Solution 8 in Fig. 1.2) the complex eigenvalues cross
the imaginary axis. This crossing is a Hopf bifurcation. Beyond λH there are
periodic solutions whose period T increases as λ increases; see Fig. 1.3 for some
representative periodic orbits. The period becomes infinite at λ = λ∞ ≈ 0.7.
This final orbit is called a heteroclinic cycle.
From Fig. 1.2 we can deduce the solution behavior for increasing λ: Branch
III is followed until λH ; then the behavior becomes oscillatory due to the
periodic solutions of increasing period until λ = λ∞ ; finally, the dynamics
collapses to the trivial solution (Branch I).

1.1.3 The Gelfand-Bratu Problem

The IFT is not only useful in the context of solution branches of equilibria.
The periodic orbits in Sect. 1.1.2 are also computed using the IFT principle.
This section gives an example of a solution branch of a two-point boundary
value problem. The Gelfand-Bratu problem [12] is defined as
 ′′
u (x) + λeu(x) = 0, ∀x ∈ [0, 1],
(1.5)
u(0) = u(1) = 0.
 1 Numerical Analysis of Nonlinear Equations 11

U2
0.70

0.60

0.50

0.40

0.30

0.20

0.10

0.00
0.10 0.20 0.30 0.40 0.50 0.60 0.70
U1

Fig. 1.3. Some periodic solutions of the predator-prey model (1.4). The final orbits
are very close to a heteroclinic cycle.

MAX U
15.0

5
12.5

10.0

7.5

5.0

2.5
3
2
1
0.0
0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0
LAMBDA

Fig. 1.4. Bifurcation diagram of the Gelfand-Bratu equation (1.5). Note that there
are two solutions for 0 < λ < λC , where λC ≈ 3.51. There is one solution for λ = λC
and for λ ≤ 0, and are no solutions for λ > λC .
12 Eusebius J Doedel

U
15.0

12.5
5

10.0

7.5

5.0

4
3
2.5
1 2

0.0
0.00 0.20 0.40 0.60 0.80 1.00
0.10 0.30 0.50 0.70 0.90 X

Fig. 1.5. Some solutions to the Gelfand-Bratu equation (1.5).

If λ = 0 then u(x) ≡ 0 is a solution. We show here that this solution is

isolated, so that there is a continuation u = ũ(λ), for |λ| small. Consider

u′′ (x) − λeu(x) = 0,
⇒ u = u(x; q, λ).
u(0) = 0, u′ (0) = q,

We want to solve u(1; q, λ) = 0, for |λ| small. Here F(0, 0) = 0.

| {z }
≡F(q,λ)

We must show (IFT) that Fq (0, 0) ≡ uq (1; 0, 0) 6= 0:


u′′q (x) − λ0 eu0 (x) uq = 0,
where u0 ≡ 0.
uq (0) = 0, u′q (0) = 1,

Now uq (x; 0, 0) satisfies


u′′q = 0,
uq (0) = 0, u′q (0) = 1.

Hence, uq (x; 0, 0) = x, so that uq (1; 0, 0) = 1 6= 0.

1.1.4 A Nonlinear Eigenvalue Problem

The equations for column buckling (from nonlinear elasticity theory) [31] are
given by
 1 Numerical Analysis of Nonlinear Equations 13

u′′1 (x) + µu1 (x) = 0,
u′3 (x) + 12 u′1 (x)2 + µβ = 0,
for x ∈ [0, 1], with

u1 (0) = u1 (1) = 0,
u3 (0) = −u3 (1) = λ (λ > 0),

where µ is a stress (to be determined) and β is another physical constant; take

β = 1. Note that the boundary conditions are ‘overspecified’ (to determine
µ). We rewrite the equations as the first order system
 ′
 u1 = u2 , u1 (0) = u1 (1) = 0,
u′2 = −µu1 , (1.6)
u3 = − 12 u22 − µ, u3 (0) = −u3 (1) = λ.
 ′

Note that u1 ≡ u2 ≡ 0 implies u′3 = −µ, so that u3 (x) = λ − µx, with

u3 (0) = λ and u3 (1) = λ − µ = −λ. Thus, we must have µ = 2λ, so that
u3 (x) = λ − 2λx = λ(1 − 2x). Hence,

u1 ≡ u2 ≡ 0, u3 (x) = λ(1 − 2x), µ = 2λ,

is a solution for all λ. Are these solutions isolated? In the formal set-up,
consider 
u′1 = u2 , u1 (0) = 0, 
u′2 = −µu1 , u2 (0) = p, ⇒ u = u(x, p, µ; λ).
u′3 = − 12 u22 − µ, u3 (0) = λ.


We must have

u1 (1, p, µ; λ) = 0,
∼ F(p, µ; λ) = 0,
u3 (1, p, µ; λ) + λ = 0,

with F : R2 × R → R2 . So the question is: Is (Fp | Fµ )(λ) nonsingular along

the basic solution branch?
To answer the above question quickly, we omit explicit construction of F.
We linearize (1.6) about u1 ,u2 , u3 , µ, and λ, with respect to u1 , u2 , u3 , and
µ, acting on v1 , v2 , v3 , and µ, to obtain the linearized homogeneous equations
 ′
 v1 = v2 , v1 (0) = v1 (1) = 0,
v2′ = −µv1 − µu1 ,
 ′
v3 = −u2 v2 − µ, v3 (0) = −v3 (1) = 0.

In particular, the linearized homogeneous equations about u1 ≡ u2 ≡ 0,

u3 (x) = λ(1 − 2x), and µ = 2λ are

v1′ = v2 , v1 (0) = v1 (1) = 0, 
v2′ = −2λv1 , ⇒ µ = 0, v3 ≡ 0.
v3′ =

−µ, v3 (0) = −v3 (1) = 0.
14 Eusebius J Doedel

Now, if 2λ 6= k 2 π 2 , k = 1, 2, 3, . . . , then

v1′′ + 2λv1 = 0,
v1 (0) = v1 (1) = 0,

has the unique solution v1 ≡ 0 and, hence, also v2 ≡ 0. Thus, if λ 6= 12 k 2 π 2

then the basic solution branch is locally unique. However, if λ = 12 k 2 π 2 then
the linearization is singular, and there may be bifurcations. (In fact, there are
buckled states.)

1.1.5 The Pendulum Equation

The equation of a damped pendulum subject to a constant torque is given by

m R φ′′ (t) + ε m φ′ (t) +m g sin φ(t) = |{z}

I ,
| {z }
damping torque

that is,
ε ′ g I
φ′′ (t) + φ (t) + sin φ(t) = .
R R mR
dφ dφ ds
Scaling time as s = c t we have φ′ = dt = ds dt = c φ̇ and, similarly, φ′′ = c2 φ̈,
we obtain
εc g I
c2 φ̈(s) + φ̇(s) + sin φ(s) = ,
R R mR
ε g I
φ̈ + φ̇ + sin φ = .
Rc Rc 2 m R c2
p
Choose c such that Rgc2 = 1, that is, c = g/R, and set ε̃ = ε/(R c) and
I˜ = I/(m R c2 ). Then the equation becomes φ̈ + ε̃ φ̇ + sin φ = I,
˜ or, dropping
the ˜, and using ′ ,
φ′′ + ε φ′ + sin φ = I. (1.7)
We shall consider special solutions, called rotations, that satisfy φ(t + T ) =
φ(t) + 2π, for all t or, equivalently,
≡0
z}|{
φ(T ) = φ(0) +2π (= 2π),
φ′ (T ) = φ′ (0),

where T is the period.

 1 Numerical Analysis of Nonlinear Equations 15

The Undamped Pendulum

First consider the undamped unforced pendulum

φ′′ + sin φ = 0,

that is, (1.7) with ε = I = 0. Suppose the initial data for a rotation are
φ(0) = 0, and φ′ (0) = p > 0. We have φ(T ) = 2π, and φ′ (T ) = φ′ (0) = p.
Integration gives
R t ′ ′′ Rt
0
φ φ dt + 0 φ′ sin φ dt = 0
t
1 ′2 t
⇔ 2φ 0 − cos φ = 0
0
1 ′ 2 1 2
⇔ 2 φ (t) − cos φ(t) = p −1
2
1 ′ 2 1
⇔ φ (t) + 1 − cos φ(t) = p2 .
|2 {z } | {z } 2
kinetic energy potential energy

Thus,
dφ
p
φ′ (t) = dt = p2 − 2 + 2 cos φ(t)
dt 1
⇔ dφ =
p
2
p − 2 + 2 cos φ
Z 2π
R 2π dt 1
⇔ 0 dφ dφ = p dφ
2
p − 2 + 2 cos φ
0
Z 2π
1
⇔ T = p dφ.
0 p2 − 2 + 2 cos φ
We see that
T →0 as p → ∞,
and Z 2π
1
T → √ dφ = ∞ as p → 2.
0 2 + 2 cos φ
In fact, rotations exists for all p > 2.

The Forced Damped Pendulum

We now consider the forced damped pendulum (1.7),

φ′′ + ε φ′ + sin φ = I,

with φ(0) = 0 (which sets the phase) and φ′ (0) = p. We write the solution
as φ = φ(t; p, I, ε). Do there exist rotations, i.e., does there exist T such that
φ(T ; p, I, ε) = 2π and φ′ (T ; p, I, ε) = p?
16 Eusebius J Doedel

Theorem 4. Let φ0 be a rotation of the undamped unforced pendulum:

φ′′0 + sin φ0 = 0,

φ0 (0) = 0, φ′0 (0) = p0 ,

φ0 (T0 ) = 2π, φ′0 (T0 ) = p0 .
Then there exist (smooth) functions T = T (p, ε) and I = I(p, ε), with
T (p0 , 0) = T0 and I(p0 , 0) = 0, such that φ(t; p, I(p, ε), ε) is a rotation of
period T (p, ε) of the damped forced pendulum

φ′′0 + ε φ′ + sin φ0 = I,

φ(0; p, I(p, ε), ε) = 0, φ′ (0; p, I(p, ε), ε) = p,

′
φ(T (p, ε); p, I(p, ε), ε) = 2π, φ (T (p, ε); p, I(p, ε), ε) = p.
for all (p, ε) sufficiently close to (p0 , 0).

Proof. The Jacobian matrix with respect to T and I, of the algebraic system

φ(T ; p, I, ε) − 2π = 0,
φ′ (T ; p, I, ε) − p = 0,

evaluated at p = p0 , T = T0 , and I = ε = 0, is
 ′ 0 
φ0 φI
J0 = ′ (T0 ).
φ′′0 φ0I

We must show that detJ0 6= 0. We have

φ′′0 + sin φ0 = 0 ⇒ φ′′0 (T0 ) = − sin (φ0 (T0 )) = − sin (2π) = 0,

φ′0 (T0 ) = p0 6= 0.
′
Thus, detJ0 6= 0 if φ0I (T0 ) 6= 0. Here, φI satisfies

φ′′I + ε φ′I + φI cos φ = 1, φI (0) = φ′I (0) = 0.

In particular,
′′ ′
φ0I + φ0I cos φ0 = 1, φ0I (0) = φ0I (0) = 0.

From ′′
φ0I φ′0 + φ0I cos φ0 φ′0 = φ′0 ,
and
φ′′0 + sin φ0 = 0 ⇒ φ′′′ ′
0 + cos φ0 φ0 = 0,

we have ′′
φ0I φ′0 − φ0I φ′′′ ′
0 = φ0 .

Using integration, we find

 1 Numerical Analysis of Nonlinear Equations 17
Z T0 Z T0 Z T0
′′
φ0I φ′0 − φ0I φ′′′
0 = φ′0 = 2π,
0 0 0
T0
T0
Z T0 Z T0
′ ′ ′
φ0I φ′0 − φI0 φ′′0 − φ0I φ′′0 + φ0I φ′′0 = 2π,
0 0 |{z} 0
− sin φ0 0
′ ′
φ0I (T0 ) φ′0 (T0 ) − φ0I (0) φ′0 (0) = 2π.
| {z } | {z }
p0 0

Hence,
′ 2π
φ0I (T0 ) = 6= 0.
p0
⊔
⊓

A more general analysis of this type for coupled pendula can be found in [1]
(see also Chap. 5).

1.2 Continuation of Solutions

As mentioned, the IFT plays an important role in the design of algorithms

for computing families of solutions to nonlinear equations. Such continuation
methods are applied in a parameter-dependent setting. Hence, we consider
the equation
G(u, λ) = 0, u, G(·, ·) ∈ Rn , λ ∈ R.
Let x ≡ (u, λ). Then the equation can be written as

G(x) = 0, G : Rn+1 → Rn .

1.2.1 Regular Solutions

A solution x0 of G(x) = 0 is regular [22] if the n (rows) by n + 1 (columns)

matrix G0x ≡ Gx (x0 ) has maximal rank, i.e., if Rank(G0x ) = n.
In the parameter formulation G(u, λ) = 0, we have


 (i) G0u is nonsingular,

 or 

Rank(G0x ) = Rank(G0u | G0λ ) = n ⇔  dim N (G0u ) = 1,



 (ii) and
 0
Gλ 6∈ R(G0u ).


Here, N (G0u ) denotes the null space of G0u , and R(G0u ) denotes the range of
G0u , i.e., the linear space spanned by the n columns of G0u .
18 Eusebius J Doedel
u

x0
u0

λ
λ
0

Fig. 1.6. A solution branch of G(u, λ) = 0; note the two folds.

Theorem 5. Let x0 ≡ (u0 , λ0 ) be a regular solution of G(x) = 0. Then, near

x0 , there exists a unique one-dimensional continuum of solutions x(s), called
a solution family or a solution branch, with x(0) = x0 .
Proof. Since Rank(G0x ) = Rank(G0u | G0λ ) = n, either G0u is nonsingular and
by the IFT we have u = u(λ) near x0 , or else we can interchange columns
in the Jacobian G0x to see that the solution can locally be parametrized by
one of the components of u. Thus, a unique solution family passes through a
regular solution. ⊔
⊓
Remark 2. We remark here that the second case in the above proof is that of
a simple fold ; see also Fig. 1.6.

1.2.2 Parameter Continuation

In the parameter-dependent setting we assume that the continuation param-

eter is λ. Suppose we have a solution (u0 , λ0 ) of

G(u, λ) = 0,

as well as the direction vector u̇0 = du/dλ, and we want to compute the
solution u1 at λ1 = λ0 + ∆λ; this is illustrated in Fig. 1.7.
To compute the solution u1 we use Newton’s method
(
(ν) (ν) (ν)
Gu (u1 , λ1 ) ∆u1 = −G(u1 , λ1 ),
(ν+1) (ν) (ν) (1.8)
u1 = u1 + ∆u1 , ν = 0, 1, 2, . . . .

As initial approximation, we use

(0)
u1 = u0 + ∆λu̇0 .

If Gu (u1 , λ1 ) is nonsingular and ∆λ is sufficiently small, then the convergence

theory for Newton’s method guarantees that this iteration will converge.
 1 Numerical Analysis of Nonlinear Equations 19

"u"

u
1 du
u (= dλ
at λ0 )

u0
(0)
u1

∆λ

λ
λ0 λ1

Fig. 1.7. Graphical interpretation of parameter continuation.

After convergence, the new direction vector u̇1 can be computed by solving

Gu (u1 , λ1 ) u̇1 = −Gλ (u1 , λ1 ).

This equation follows from differentiating G(u(λ), λ) = 0 with respect to λ

at λ = λ1 . Note that, in practice, the calculation of u̇1 can be done without
another LU -factorization of Gu (u1 , λ1 ). Thus, the extra work to find u̇1 is
negligible.
As an example, consider again the Gelfand-Bratu problem of Sect. 1.1.3
given by  ′′
u (x) + λeu(x) = 0, ∀x ∈ [0, 1],
u(0) = u(1) = 0.
If λ = 0 then u(x) ≡ 0 is an isolated solution; see Sect. 1.1.3. We discretize
this problem by introducing a mesh,

0 = x0 < x1 < · · · < xN = 1,

xj − xj−1 = h, 1 ≤ j ≤ N, h = 1/N.

The discrete equations are:

uj+1 − 2uj + uj−1
+ λeuj = 0, j = 1, . . . , N − 1,
h2
with u0 = uN = 0. (More accurate discretization is discussed in Sect. 1.3.1.)
Let  
u1
 u2 
u≡  · .


uN −1
Then we can write the above as G(u, λ) = 0, where G : Rn × R → Rn , with
n = N − 1.
20 Eusebius J Doedel

"u"

u1
u0
( u0, λ 0 )
u0
∆s
λ0

λ
λ0 λ1

Fig. 1.8. Graphical interpretation of pseudo-arclength continuation.

For the parameter continuation we suppose that we know λ0 , u0 , and u̇0 .

(0)
Then we set λ1 = λ0 + ∆λ and apply Newton’s method (1.8) with u1 =
u0 + ∆λ u̇0 . After convergence find u̇1 from

Gu (u1 , λ1 ) u̇1 = −Gλ (u1 , λ1 ),

and repeat the above procedure to find u2 , u3 , and so on. Here,

 2 1

− h2 + λeu1 h2


1
h2 − h22 + λeu2 h12 

Gu (u, λ) = 
 . . . .

 . . . 
1 2 uN −1
h2 − h2 + λe

Hence, we must solve a tridiagonal system for each Newton iteration. The
solution branch has a fold where the parameter-continuation method fails; see
Figs. 1.4 and 1.5.

1.2.3 Keller’s Pseudo-Arclength Continuation

In order to allow for continuation of a solution branch past a fold, Auto

[8, 11, 12] uses Keller’s Pseudo-Arclength Continuation [22]. Suppose we have
a solution (u0 , λ0 ) of G(u, λ) = 0, as well as the direction vector (u̇0 , λ̇0 ) of the
solution branch. Pseudo-arclength continuation solves the following equations
for (u1 , λ1 ): 
G(u1 , λ1 ) = 0,
∗ (1.9)
(u1 − u0 ) u̇0 + (λ1 − λ0 ) λ̇0 − ∆s = 0.
Figure 1.8 shows a graphical interpretation of this continuation method. New-
ton’s method for pseudo-arclength continuation becomes
 1 Numerical Analysis of Nonlinear Equations 21

"X-space"

x1

x0
x
0
∆s

Fig. 1.9. Parameter-independent pseudo-arclength continuation.

! (ν)
! (ν) (ν)
!
(G1u )(ν) (G1λ )(ν) ∆u1 G(u1 , λ1 )
(ν)
=− (ν) (ν)
,
u̇∗0 λ̇0 ∆λ1 (u1 − u0 )∗ u̇0 + (λ1 − λ0 ) λ̇0 − ∆s
with the new direction vector defined as
! ! !
G1u G1λ u̇1 0
= ,
u̇∗0 λ̇0 λ̇1 1

Note that
• In practice (u̇1 , λ̇1 ) can be computed with one extra back-substitution;
• The orientation of the branch is preserved if ∆s is sufficiently small;
• The direction vector must be rescaled, so that indeed || u̇1 ||2 +λ̇21 = 1.
Theorem 6. The Jacobian of the pseudo-arclength system is nonsingular at
a regular solution point.

Proof. Let x = (u, λ) ∈ Rn+1 . Then pseudo-arclength continuation can be

written as
G(x1 ) = 0,
∗
(x1 − x0 ) ẋ0 − ∆s = 0, (|| ẋ0 ||= 1).
Figure 1.9 shows
 0a graphical interpretation. The matrix in Newton’s method
Gx
at ∆s = 0 is . At a regular solution we have N (G0x ) = Span{ẋ0 }. We
ẋ∗0 
G0x
must show that is nonsingular at a regular solution. Suppose, on the
ẋ∗
 0 0
Gx
contrary, that is singular. Then there exists some vector z 6= 0 with
ẋ∗0

G0x z = 0 and ẋ∗0 z = 0.

Thus, z = cẋ0 , for some constant c. But then
0 = ẋ∗0 z = cẋ∗0 ẋ0 = c || ẋ0 ||2 = c,
so that z = 0, which is a contradiction. ⊔
⊓
22 Eusebius J Doedel

Consider pseudo-arclength continuation for the discretized Gelfand-Bratu

problem of Sect. 1.1.3. Then the matrix
! !
Gx Gu Gλ
=
ẋ∗ u̇∗ λ̇

in Newton’s method is a ‘bordered tridiagonal’ matrix of the form

 
• • •
• • • •
 

 • • • •

 • • • •

 • • • •.

 • • • • 


 • • • • 
 • • •
• • • • • • • • •

We now show how to solve such linear systems efficiently.

1.2.4 The Bordering Algorithm

The linear systems in Newton’s method for pseudo-arclength continuation are

of the form     
A c x f
= .
b∗ d y h
The special structure of this extended system can be exploited; a general
presentation of the numerical linear algebra aspects of extended systems can
be found in [20, 24]. If A is a sparse matrix whose LU -decomposition can be
found relatively cheaply (e.g., if A is tridiagonal), then the following bordered
LU -decomposition [22] will be efficient:
    
A c L 0 U γ
= .
b∗ d β∗ 1 0∗ δ

After decomposing A = LU (which may require pivoting) we compute γ, β,

and δ from

Lγ = c,
U ∗ β = b,

δ = d − β ∗ γ.

The linear system can then be written as

 1 Numerical Analysis of Nonlinear Equations 23
     
L 0 U γ x f
= ,
β∗ 1 0∗ δ y h
| {z }
 
f̂
≡ 
ĥ

and we can compute the solution (x, y) through the following steps:

Lf̂ = f ,

ĥ = h − β ∗ f̂ ,

y = ĥ/δ,

U x = f̂ − yγ.

Theorem 7.The bordering

 algorithm outlined above works if A and the full
A c
matrix A ≡ are nonsingular.
b∗ d
In the proof of Theorem 7 we make use of the Bordering Lemma [22]
 
A c
Lemma 3 (Bordering Lemma). Let A ≡ . Then
b∗ d
(a) A nonsingular ⇒ A nonsingular if and only if d 6= b∗ A−1 c;

∗ c 6∈ R(A),
(b) dim N (A) = dim N (A ) = 1 ⇒ A nonsingular if
b 6∈ R(A∗ );
(c) If dim N (A) ≥ 2 then A is singular.

Proof. (a) (A nonsingular)

  
A 0 I A−1 c
In this case A = , where e = d − b∗ A−1 c. Clearly,
b∗ 1 0∗ e
A is nonsingular if and only if e 6= 0.

(b) (dim N (A) = 1)

Suppose A is singular in this case. Then there exist z ∈ Rn and ξ ∈ R,
not both zero, such that
      
A c z Az + ξc 0
A= = = .
b∗ d ξ b∗ z + ξd 0

We see that c ∈ R(A) if ξ 6= 0, which contradicts the assumptions. On

the other hand, if ξ = 0 and z 6= 0 then

N (A) = Span{z} and b ∈ N (A)⊥ .

24 Eusebius J Doedel

Since, in general, N (A)⊥ = R(A∗ ), it follows that b ∈ R(A∗ ), which also

contradicts the assumptions.

(c) (dim N (A) ≥ 2)

This case follows from a rank argument. ⊔
⊓
Proof (Theorem 7). The crucial step in the bordering algorithm is the com-
putation of z = ĥ/δ. Namely, we must have δ 6= 0. Since δ is determined in
the bordered LU -decomposition, we have
δ = d − β∗ γ = d − (U ∗−1 b)∗ (L−1 c)
= d − b U L c = d − b∗ (LU )−1 c
∗ −1 −1

= d − b∗ A−1 c,
which is nonzero by Conclusion (a) of the Bordering Lemma. ⊔
⊓
Remark 3. In pseudo-arclength continuation we have
! !
A c Gu Gλ
A= =
b∗ d u̇∗0 λ̇0 ,
that is, A = Gu , which is singular at a fold. Therefore, the bordering algorithm
will fail when it is used exactly at a fold. In practice, the method may still work.
We consider another approach, used in Auto, when discussing collocation
methods in Sect. 1.3.1.

1.3 Boundary Value Problems

Consider the first-order system of ordinary differential equations
u′ (t) − f (u(t), µ, λ) = 0, t ∈ [0, 1],
where
u(·), f (·) ∈ Rn , λ ∈ R, µ ∈ Rnµ ,
subject to boundary conditions
b(u(0), u(1), µ, λ) = 0, b(·) ∈ Rnb ,
and integral constraints
Z 1
q(u(s), µ, λ) ds = 0, q(·) ∈ Rnq .
0

We want to solve this boundary value problem (BVP) for u(·) and µ. In order
for this problem to be formally well posed we require that
nµ = nb + nq − n ≥ 0.
We can think of λ as the continuation parameter in which the solution (u, µ)
may be continued. A simple case is nq = 0, nb = n, for which nµ = 0.
 1 Numerical Analysis of Nonlinear Equations 25

1.3.1 Orthogonal Collocation

Auto solves boundary value problems using the method of orthogonal collo-
cation with piecewise polynomials [2, 7]. This method is very accurate, and
allows adaptive mesh-selection. The set-up is as follows.
First, we introduce a mesh

{0 = t0 < t1 < · · · < tN = 1},

with
hj = tj − tj−1 , (1 ≤ j ≤ N ).
Define the space of (vector-valued) piecewise polynomials Phm as
n o
Phm = ph ∈ C[0, 1] | ph |[tj−1 ,tj ] ∈ P m ,

where P m is the space of (vector-valued) polynomials of degree ≤ m. The

orthogonal collocation method with piecewise polynomials [3] consists of find-
ing ph ∈ Phm and µ ∈ Rnµ , such that the following collocation equations are
satisfied:

p′h (zj,i ) = f (ph (zj,i ), µ, λ), j = 1, . . . , N, i = 1, . . . , m,

and such that ph satisfies the boundary and integral conditions. The collo-
cation points zj,i in each subinterval [tj−1 , tj ] are the (scaled) roots of the
mth-degree orthogonal polynomial (Gauss points); see Fig. 1.10 for a graph-
ical interpretation. Since each local polynomial is determined by (m + 1)n,
coefficients, the total number of degrees of freedom (considering λ as fixed) is
(m + 1)nN + nµ . This is matched by the total number of equations:

collocation: mnN,
continuity: (N − 1)n,
constraints: nb + nq (= n + nµ ).

If the solution u(t) of the BVP is sufficiently smooth then the order of accuracy
of the orthogonal collocation method is m, i.e.,

|| ph − u ||∞ = O(hm ).

At the main meshpoints tj we have superconvergence:

maxj | ph (tj ) − u(tj ) |= O(h2m ).

The scalar variables µ are also superconvergent [7].

26 Eusebius J Doedel

0 1
t t t t
0 1 2 N

t j-1 tj
z j,1 z j,2 z j,3

t j-1 tj
t j-2/3 t j-1/3
lj,3(t) lj,1(t)

Fig. 1.10. The mesh {0 = t0 < t1 < · · · < tN = 1}. Collocation points and
‘extended-mesh points’ are shown for the case m = 3, in the jth mesh interval. Also
shown are two of the four local Lagrange basis polynomials.

1.3.2 Implementation in Auto

The implementation in Auto [12] is done via the introduction of Lagrange

basis polynomials for each subinterval [tj−1 , tj ]. Define

{ℓj,i (t)}, j = 1, . . . , N, i = 0, 1, . . . , m,

by
m
Y t − tj− k
ℓj,i (t) = m
,
tj− mi − tj− k
k=0,k6=i m

where
i
tj− mi = tj −
hj .
m
The local polynomials can then be written as
m
X
pj (t) = ℓj,i (t)uj− mi .
i=0

With the above choice of basis

uj approximates u(tj ) and uj− mi approximates u(tj− mi ),

where u(t) is the solution of the continuous problem.

Then the collocation equations are

p′j (zj,i ) = f (pj (zj,i ), µ, λ), i = 1, . . . , m, j = 1, . . . , N,

 1 Numerical Analysis of Nonlinear Equations 27

u0 u1 u2 u1 u2 uN T λ
3 3
• • • • • • • • • •
• • • • • • • • • •
• • • • • • • • • •
• • • • • • • • • •
• • • • • • • • • •
• • • • • • • • • •
• • • • • • • • • •
• • • • • • • • • •
• • • • • • • • • •
• • • • • • • • • •
• • • • • • • • • •
• • • • • • • • • •
• • • • • • • • • •
• • • • • • • • • •
• • • • • • • • • •
• • • • • • • • • •
• • • • • • • • • •
• • • • • • • • • •
• • • • • •
• • • • • •
• • • • • • • • • • • • • • • • • • • • • •
• • • • • • • • • • • • • • • • • • • • • •

Fig. 1.11. Structure of the Jacobian for the case of n = 2 differential equations
with the number of mesh intervals N = 3, the number of collocation points per
mesh interval m = 3, the number of boundary conditions nb = 2, and the number
of integral constraints nq = 1. The last row corresponds to the pseudo-arclength
equation, which is not included in the nq = 1 count. From E.J. Doedel, H.B. Keller,
J.P. Kernévez, Numerical analysis and control of bifurcation problems (II): Bifurca-
tion in infinite dimensions, Internat. J. Bifur. Chaos Appl. Sci. Engrg. 1(4) (1991)
745–772 c 1991 World Scientific Publishing; reproduced with permission.

the discrete boundary conditions are

bi (u0 , uN , µ, λ) = 0, i = 1, . . . , nb ,

and the integrals constraints can be discretized as

N X
X m
ωj,i qk (uj− mi , µ, λ) = 0, k = 1, . . . , nq ,
j=1 i=0

where the ωj,i are the Lagrange quadrature coefficients.

The pseudo-arclength equation is
Z 1
(u(t) − u0 (t))∗ u̇0 (t) dt + (µ − µ0 )∗ µ̇0 + (λ − λ0 ) λ̇0 − ∆s = 0,
0

where (u0 , µ0 , λ0 ), is the previously computed point on the solution branch,

and (u̇0 , µ̇0 , λ̇0 ), is the normalized direction of the branch at that point. The
 Another Random Document on
Scribd Without Any Related Topics
sanctioned by the gospel, with the precarious notions of heaven
derived from the same authority—to have less fear of death than
deists, by them called infidels. But I have known many deists die, and
certainly I have never seen any believer in Christianity meet death
with more firmness than those who did not believe. Some of the last-
mentioned class have died of painful complaints with the most
admirable equanimity. But to illustrate the injustice of these reverend
gentlemen in their undue assumption of the superior fearlessness of
death of those persons who believe in the gospel over those who do
not so believe, I will here quote a passage from the travels of a
Christian missionary, which describes the equanimity with which the
Chinese meet death. From the work of the French missionary, Mr.
Huc, it will be seen that death is contemplated with far less fearful
apprehension in China than in Christendom.

Observations of the Missionary, Mr. Huc, to whom reference

has just been made.

2053. “In no other country than China, perhaps, could

men be heard exchanging compliments on the subject of a
coffin. People (in Christendom) are most shy of mentioning
the lugubrious objects destined to contain the mortal
remains of a relation or friend; and when death does enter
the house, the coffin is got in in secresy and silence, in
order to spare the feelings of the mourning family. But it is
quite otherwise in China; there a coffin is simply an article
of the first necessity to the dead, and of luxury and fancy
to the living. In the great towns you see them displayed in
the shops with all sorts of tasteful decorations, painted,
and varnished, and polished, and trimmed up to attract the
eyes of passengers, and give them the fancy to buy
themselves one.
2054. “People in easy circumstances, who have money to
spare for their pleasures, scarcely ever fail to provide
themselves beforehand with a coffin to suit their own taste,
and which they consider becoming; and, until the moment
arrives for lying down in it, it is kept in the house; not as
an article of immediate necessity, but as one that cannot
fail to be consoling and pleasant to the eye in a nicely-
furnished apartment.
2055. “For well-brought up children, it is a favourite
method of expressing the fervour of their filial piety toward
the authors of their being, a sweet and tender consolation
for the heart of a son, to be able to purchase a beautiful
coffin for an aged father or mother, and come in state to
present the gift at the moment when they least expect
such an agreeable surprise. If one is not sufficiently
favoured by fortune to be able to afford the purchase of a
coffin in advance, care is always taken before ‘saluting the
world,’ as the Chinese say, a sick person shall at least have
the satisfaction of casting a glance at his last abode; and if
he is surrounded by at all affectionate relations, they never
fail to buy him a coffin and place it by the side of his bed.
2056. “In the country, this is not always so easy, for
coffins are not kept quite ready, and, besides, peasants
have not such luxurious habits as townspeople. The only
way, then, is to send for the carpenter of the place, who
takes the measure of the sick person, not forgetting to
observe to him that it must be made a little longer than
would seem necessary, because one always stretches out a
little when dead. A bargain is then made concerning the
length and breadth, and especially the cost; wood is
brought, and the workmen set about their task in the yard,
close to the chamber of the dying person, who is
entertained with the music of the saw and the other tools,
while death is at work with him, preparing him to occupy
the snug abode when it is ready.
2057. “All this is done with the most perfect coolness,
and without the slightest emotion, real or affected. We
have ourselves witnessed such scenes more than once, and
it has always been one of the things that most surprised us
in the manners of this extraordinary country. A short time
after our arrival in the mission of the north, we were
walking one day in the country with a Chinese seminarist,
who had the patience to reply to all our long and tedious
questions about the men and things of the Celestial
Empire. While we were keeping up the dialogue as well as
we could, in a mixture of Latin and Chinese, using a word
of one or the other as we found occasion, we saw coming
toward us a rather numerous crowd, who advanced in an
orderly manner along a narrow path. It might have been
called a procession.
2058. “Our first impulse was to turn aside, and get into
some safe corner behind a large hill; for, not having as yet
much experience in the manners and customs of the
Chinese, we had some hesitation in producing ourselves,
for fear of being recognised and thrown into prison;
possibly even condemned and strangled. The crowd had
now come up with us, and we stood aside to let it pass. It
was composed of a great number of villagers, who looked
at us with smiling faces, and had the appearance of being
uncommonly pleased. After them came a litter, on which
was borne an empty coffin, and then another litter, upon
which lay extended a dying man, wrapped in blankets. His
face was haggard and livid, and his expiring eyes were
fixed upon the coffin that preceded him. When every one
had passed, we hastened to ask the meaning of this
strange procession. ‘It is some sick man,’ said the
seminarist, ‘who has been taken ill in a neighbouring
village, and whom they are bringing home to his family.
The Chinese do not like to die away from their own house.’
‘That is very natural; but what is the coffin for?’ ‘For the
sick man, who probably has not many days to live. They
seem to have made every thing ready for his funeral.’ I
remarked by the side of the coffin a piece of white linen.
‘That, they mean to use for the mourning.’
 2059. “These words threw us in the most profound
astonishment, and we saw then that we had come into a
new world—into the midst of a people whose ideas and
feelings differed widely from those of Europeans. These
men quietly setting about to prepare for the funeral of a
still living friend and relation—this coffin placed purposely
under the eyes of the dying man, doubtless with the
purpose of doing what was agreeable to him; all this
plunged us into a strange reverie, and the walk was
continued in silence. The astonishing calmness with which
the Chinese see the approach of death does not fail when
the last moment arrives. They expire with the most
incomparable tranquillity, without any of the emotions, the
agitations, the agonies that usually render the moment of
death so terrific.”
2060. It is remarkable that Mr. Huc cites an “entire want of religious
feeling, as among the causes of this indifference to death.”[57] But it
may be inquired whether that can be a proper kind of religious feeling
which interferes with equanimity at the prospect of our spiritual birth.
I can easily believe Mr. Huc to be correct, if his entire want of
religious feeling means the absence of all fear of an eternal broiling,
like that of Dives.
2061. It is the absence of that sort of religion which this sectarian
would teach; that which consigns the great majority of mankind to
perpetual misery on account of their disbelief in Romanism.
2062. Widely different is the effect of the religion I have espoused.
It has made a prodigious change in my feelings. I look forward to
death with hope, rather than fear. (See page 32 (108) of this work for
the different effect of Romanism on my mind.)

Conclusion of Strictures on Mr. Mahan’s Religious Errors.

2063. It will be observed that under the general head of Mr.

Mahan’s errors, I have treated of not only those which he has
advocated, but such as he sanctions by his general endorsement of
Scripture. However, I here take my leave of Mr. Mahan and his errors.
 CONCLUSION.
2064. In a work by the English bishop Warburton, is to be found
the following allegation:
“The doctrine of a future state of rewards and punishments is not
to be found in, nor did it make part of, the Mosaic dispensation.”

The Pentateuch inconsistently represented as the basis of

belief in human immortality, as will appear by the
prefixed quotation, as well as the inability of any of its
devotees to point out any part which conflicts with the
right reverend author’s opinion, above mentioned.—
Injustice of representing disbelievers in the Bible as not
having as good grounds for belief in immortality as those
who rest their belief on a work which, by its silence,
tends to discountenance the hope of a future life.—Those
who uphold the Bible against Spiritualism, the real
antagonists of the only satisfactory evidence ever given
to man of a future habitation in the spirit world.

2065. Religious people all very justly insist on the immense

importance of a belief in a future state of existence, in which we are
to enjoy a degree of happiness in proportion to our good conduct in
this life, or of misery in proportion as we do evil.
2066. Throughout Christendom I believe such a belief is necessary
to render a person competent as a witness, or to hold any office. As
it is assumed generally by Christians that there is no other proof of
immortality than that alleged to be afforded by Scripture,
unbelievers in Scripture are assumed to be unbelievers in a future
state, and the most unfavourable insinuations are made respecting
such persons, (1310.) Bishop McIlvaine, in his “Evidences,” has
charitably represented that, as a class, such men are peculiarly
vicious, and in their domestic relations immoral; not recollecting how
far Christian prelates have been found wanting in the present, as
well as in past times. But notwithstanding that the disbelief in a
future state is held to be so universally pernicious in its influence,
self-called orthodox Christians deem it impious not to bow before the
Bible as the holy word of God, or to be wanting in respect for his
inspired missionary, Moses. But admitting that the books found by
Hilkiah (1940) were written, under divine inspiration, by Moses, how
does it happen that those books and their author stand in such high
estimation, when they actually give no evidence of immortality, but
rather tend to prove that God, in communicating his will to Moses,
thought it more important to give to the Jews directions for the
decoration of a tabernacle, than to impart to them the invaluable
knowledge of their eternal existence? Moreover, God is represented
as holding in especial favour, those who did not think it of as much
importance to inquire into the truth of immortality, as to obtain
decorations for pharisaical worship. Why is Moses, a materialist, to
be venerated, who so grossly trifled with his opportunities? There is
either a flagitious misrepresentation, or he, of all men, had the best
opportunity of learning this all-important truth; and therefore, dying
ignorant of it, is proportionally more culpable. Wherefore is the
uncandid and unfounded pretence resorted to, that attacking the Old
Testament is attacking the basis of the hope of heaven, when, on
the contrary, that record has really the opposite tendency—that of
enfeebling the evidence of immortality?
2067. We see the heathen Cyrus, on his death-bed, (1980,)
striving to impress upon his children a belief in immortality, while
Moses, dying a worldly-minded, blood-thirsty materialist, employs his
last moments in giving inhuman directions for the merciless
massacre of every conquered pagan. Yet, while Bishop McIlvaine, as
respects his contemporaries, represents materialism as irreconcilable
with virtue, Moses is to be venerated and the books attributed to his
authorship idolatrously worshipped as the word of God! But if the
advocacy of any thing which tends to lessen the hope of immortality
be culpable, is it not culpable, against the truths of Spiritualism to
hold up a book which authorizes disbelief in immortality, and that,
too, upon such questionable authority as an obscure priest and
fanatical and barbarous autocrat? (1937.) As respects the gospel, it
has, I hope, been shown that, with all the exertions of the good
parson Harbaugh to exhibit the heaven of Scripture in its most
favorable aspect, (777 to 805,) Spiritualism has immensely the
advantage in the description given of the spirit world, (409 to 469,)
sanctioned by the replies given to my queries under test conditions,
(552).
2068. Another ground of pre-eminence is the perfect immunity
from any association with such a priesthood as that described by the
Right Rev. Bishop Hopkins. Media, replacing the priests, will owe
their office to nature, not to any aristocracy, monarch, or theocrat,
(1307).
2069. Private disinterested Media will always outnumber and
control any of the same class who may attempt to acquire unfair
ascendancy, even if, in the nature of the case, it were possible that
such a wrong could be contemplated.
2070. I trust I have shown that the actual morals of Christendom
are irreconcilable with the precepts of the gospel, which denounce
wealth and enjoin submission to wrong, these morals being also
inconsistent with the materialism of the books of Moses. Under these
circumstances, let the reader turn to the evidence which has
converted me from a prepossessed skeptic to a devout believer in
spiritual communication. Let that glorious portraiture of the spirit-
world be considered which has been opened to the view of mortals
through the high spirits who have accredited me as one of their
servants. I will not go over the evidence, nor recapitulate the
arguments which I have already so fully urged upon the attention of
my readers. I implore them to read with candour, and think earnestly
of the facts and reasoning submitted in this volume.
P. S.—Explanation respecting Jesus Christ.
 My spirit sister alleges that Christ never uttered the
language recorded as his, and upon which I have
commented. This being admitted, I wish that nothing
which I have said may be considered as bearing
personally on a Being who is so much the object of
devotion with many of my dearest connections, relations,
and friends. I wish that it should be considered that it is
only upon the doctrines imputed to Christ, that I have
intended to animadvert, not on Jesus himself. Surely I am
justified as treating that language as coming from Jesus
Christ, which is ascribed to him in the only history of his
teachings which has come down to us, and which, under
the name of the holy canonical gospel, is considered by
more than three-fourths of all the Christians in existence
as inspired by God!
 APPENDIX.
Under this head I place some articles which could not be
embodied in this work; but which it may nevertheless be expedient
to place within the reach of certain readers. Among the articles
referred to is the theory of electricity, which I first published in 1848,
and which has been approved by the spirit of Franklin, and, in
obedience to his advice, inserted in this volume. Unfortunately, there
are but few persons sufficiently acquainted with the phenomena
which form its basis, and the technical language employed by
professed electricians, to find it agreeable to study the subject in
question; but they may qualify themselves to do so by studying the
elementary works on this branch of science.

LETTER TO THE EPISCOPAL CLERGY.

Although the subjoined letter has been published in various
channels, as well as in a separate pamphlet, I deem it proper to
record it in this work, as, otherwise, many who may see the one
might not see the other. It will be perceived that the substance of
my second letter has been already incorporated in the preceding
pages, (714 to 776.) Of course it is to the first letter that I now
allude, and which I intend to record here.

Letter from Dr. Hare to the Clergy of the Protestant

Episcopal Church, offering to lay before them the New
Evidence of Immortality. (Submitted to the late
Convention, Philadelphia, May 15, 1855.)

Reverend and Dear Sirs: Having, from my youth, been on

friendly terms with the clergy of the Episcopal Church,
within the pale of which I was born and christened;
having, in fact, had among the clergy of that church some
excellent friends and relatives,—it has been a source of
regret that I have not been able to see doctrines deeply
affecting the happiness of mankind in the same light. I
am, however, fully sensible of the kindness and courtesy
with which I have been treated by clergymen in general,
and especially by those of the church above designated. I
have always been under the belief that in no part of the
globe, nor at any period of human history, has a
priesthood existed as moral, as sincere, and truly pious as
those of my country; and among that priesthood, I
believe, none have stood higher in these qualifications
than such as are of the Episcopal Church.
It is happy for me that of late I have, in one respect,
found myself more in accordance with the Christian
clergy: I allude here to the awakening of perfect
confidence in the immortality of the soul. There was on
this subject, heretofore, this difference between my
sentiments and those of my clerical friends, that while I
hoped for a future state, I was no less skeptical respecting
the evidence of witnesses who lived some thousand years
ago than of those who have, in modern times, alleged
themselves to have witnessed supernatural
manifestations. I required in the former case, no less than
in the latter, intuitive proof; or the consistent testimony of
independent observers having sufficient sense,
knowledge, and integrity to make reliable witnesses.
Happily, in the case of Spiritualism, both of these tests
have been afforded to me, so that I now believe in a
future state no less firmly than the orthodox Christian.
Like St. Paul, in the case of Christianity, I entered upon
the investigation of Spiritualism with a view to refutation;
but the very instruments which I contrived to accomplish
that object produced the opposite effect.
If human testimony is not to be taken when advanced
by contemporaries known to be conscientious, truthful,
and well informed, how is it to be relied on with respect to
those of whom we know nothing available beside what
their own writings mention?
I am prepared to submit a communication respecting
the spirit world from my father, sanctioned by a
convocation of spirits, whose approbation was manifested
by means which no mortal could pervert.
The practical influence on my mind has been to make
me far more happy, to remove all fear of death, and to
render me more watchful as to my deportment in life. I
know that my sainted parents, and other relatives and
friends, my children who died in infancy, are around me,
witnessing every act and exercising a limited power over
my safety and my health.
Mourning for the dead now seems to be groundless,
and at all events can be indulged only upon selfish
considerations. But who would grieve deeply at a transient
separation, even for years, from friends made happier by
the change, when sure of a happy reunion ultimately?
No evidence of any important truth in science can be
shown to be more unexceptionable than that which I have
received of this glorious fact, that heaven is really “at
hand,” and that our relatives, friends, and acquaintances
who are worthy of happiness, while describing themselves
as ineffably happy, are still progressing to higher felicity;
and while hovering aloft in our midst, are taking interest in
our welfare with an augmented zeal or affection, so that,
by these means, they may be a solace to us, in despite of
death.
 As the reverend clergy of the Episcopal Church are
about to meet in Philadelphia, I deem it my duty to afford
them an opportunity of hearing the evidence on which I
rely; and which, with due effort, they can have subjected
to their own intuition.
Should the clergy deem it expedient to listen to my
exposition, I shall be ready to answer any queries which
may be made.
I am aware that there may be considerations which may
justify the clergy in declining to hear me. I have never, in
my own case, deemed it wise to seek abstract right at the
expense of practical evil. I would not urge persons in
certain official stations to become converts to Spiritualism,
lest it should, by consequent unpopularity, interfere with
their usefulness, as in the case of Judge Edmonds; and a
like objection must arise as to the conversion of
clergymen, so far as to bring their convictions in
competition with their professional vocation. Orthodox
Christians are generally educated to believe not only that
the revelation on which they rely is true, but that no other
can be justifiable. Hence they are evidently displeased
that spiritualists should allege themselves to have come
by other means to that belief in immortality which is
admitted on all sides to be the greatest comfort under the
afflictions to which temporal life is liable.
There is, moreover, this discordancy in doctrine:
Agreeably to Scripture, man is placed here for probation,
and is liable to be eternally punished if he prove
delinquent. According to Spiritualism, man is placed here
for progression, and when he goes to the next world, still
will have the opportunity to progress, however wicked he
may be when he departs this life.
It is conceived by spiritualists that if, as the orthodox
allege, God be omnipotent, he can make his creatures to
suit his will; if he be omniscient, he must know what they
are when made; and if he be prescient, he can foresee
what they will be, and consequently cannot have the
smallest conceivable motive for exposing them to
probation.
I foresee that it may not be deemed expedient to take
any notice of this letter; but whatever may be the result in
this way, does not interfere with the propriety of my
putting it in your power to avail yourselves of my offer;
since I have a sanction from a higher source, the spirit of
the immortal Washington, the proofs of whose communion
with me I am prepared to submit to any respectable
inquirers.
I am aware that this language would, a few years ago,
have made me attach the idea of insanity to the author;
but this cannot, nevertheless, in the slightest degree, be
deducible from it now, from the notorious fact that the
same monomania is never entertained by any two
persons, and in my hallucination, if it be such, there are a
multitude of participators. That is to say, there are a
multitude of persons of every grade who believe that they
have communicated with their spirit friends, as I have with
mine; and who, like me, have believed themselves to have
held communion with the spirits of some of the most
distinguished men who have departed this life. A faith in
the miracles of the New Testament may as well be
adduced as insanity as belief in spiritual manifestations
under these circumstances.
The fact that manifestations have been made and
truthfully described has been admitted by the Catholic
Church, but are ascribed to Satanic agency.
Let the doctrines of Spiritualism, and those of the
church in question, be compared, in order to determine
which owes most to Satan.
 The existence of a devil being admitted, was there ever
a more fertile source of diabolical intolerance than the
idea that a peculiar belief being necessary to save men’s
souls from hell fire, any temporal evil to which mortals
might be subjected to coerce belief, would be as justifiable
as the forcible extirpation of an incipient cancer from the
body of a child unwilling to submit to the operation? If
ever there was a devil’s agency, it may be seen in the
auto-da-fé, the Inquisition, and the massacre of St.
Bartholomew’s day.
Of the same devilish character was the execution of
Servetus by Calvin, or the persecution of the Quakers and
witches by the Puritans.
Respectfully, your well-wisher, Robert Hare.

A LETTER FROM DR. HARE.

Addressed to the President of the Association for the Advancement
of Science, at their meeting, August 18, 1855.

Preliminary Remarks.

Having addressed the subjoined letter to the American Association

for the Advancement of Science, the standing committee resolved
that the subject did not fall sufficiently within the objects of the
Association to allow even of its being read to the meeting. In the
instance of my letter to the Episcopal clergy, it was stated that its
acknowledgment by them was not expected; and after this
impression was verified, I admitted that, in their not replying, the
interest of the church was best consulted. In the present case I
admit that the harmony of the Association was perhaps best
consulted in not recognising that the objects of the Association
involved the duty of allowing certain facts to be stated before it,
which are at war with the received doctrines of science, no less than
with those of revelation.
During the Dark Ages, the so-called word of God (but really the
words of ignorant propagandists) had taken such hold of the proper
domain of science, that it was heresy to assert the rotundity of this
planet, or that the sun did not revolve about it diurnally. But at this
time science has established itself upon the domain claimed for
religious truth, so that between the positive science of the atheist
Comte and the dogmatic opinions of the orthodox savan, there is no
room for the germ of Spiritualism to shoot up.
It seems to me that, in due courtesy and liberality, the standing
committee might have had my letter read to the meeting, and have
let the members judge whether it should be acknowledged. But I
place it on record in this volume, and leave the propriety of their
having neglected to acknowledge it to be estimated hereafter in this
world, as it has already been in the world of spirits, as respects its
influence on the estimation of the parties.
I am aware, however, that every man in society is more or less a
peon, and that there is no small analogy between the situation of
many holding worldly pre-eminence and that of the poor apothecary
of Shakspeare. Conscience and reason are ever under the control of
expediency. Those who live in society must be governed by the
hearts and heads of others as well as their own: unless they are
quite sure that the cause of truth will suffer by their silence, they
should not speak to give others umbrage. I always considered it my
duty not to do any thing which would injure the institution of which I
was a member for nearly thirty years. Doubtless, the sachems of the
Association did what they thought best, as probably I should have
done, had I been situated as they were, and holding their opinions.
In the letter as actually sent to the Association, I introduced the
arguments founded on Dr. Bell’s observations, (111, 287, 864,) also
the facts and reasoning submitted in the Supplemental Preface. I
shall, of course, leave the reader to recur to those passages, and
introduce here only the other portions of my letter:
To the President of the Association for the Advancement of
Science:
Dear Sir: Being engaged in putting a work to press, I
am sorry to be unable to be present at the meeting of the
Association on the 15th inst.
When I was at the last meeting, I stated an experiment
made with the greatest care and precision, which proved
the existence of a power independent of any possible or
conceivable mortal agency; and I had on that occasion an
opportunity of experiencing the fate of the Dutch
ambassador who first made the king of Ava acquainted
with the fact that bodies of water can be frozen so as to
enable people to walk on a solidified aqueous surface. It
was a disease of the mind in either case.
But let no one apply to his soul the self-complacent
unction that it was my hallucination, not bigoted
ignorance, that originated that diagnosis. Since that time,
the fact of movements being made intelligibly, without any
perceptible or assignable mortal aid, has been verified
hundreds of times by others; while under my own intuition
it has been reiterated many times, the experiment which I
adduced having been repeated with every imaginable
precaution and instructive variation. * * * * *
About two years have elapsed, since I erroneously
sanctioned Farraday’s explanation of the manifestations, in
ascribing them to involuntary muscular action. This arose
from my being, no less than that philosopher, so utterly
incredulous and intolerant of any idea of spiritual agency
in any of the phenomena of nature, excepting those which
I ascribed to God, that I did not take the possible agency
of spirits into view; but having been obliged to admit the
facts conceded by Dr. Bell, (287,) and having received
interesting and intellectual communications of which he
has not been informed, I cannot, with him, stop half-way:
 nor could he, were he to have the interesting
communications with his spirit friends which I have had
with mine.
There has been a time when religion repressed science;
and it would seem that at the present era science is to
revenge itself by repressing religious truth, by sanctioning
indirectly the alleged manifestations of antiquity, while
deriding those of the present time; believing on miracles
told by no one knows who, yet denying the allegations of
eye-witnesses known to be truthful; while straining at
spiritual gnats, swallowing scriptural camels.
With high esteem, your well-wisher,
Robert Hare.

FARRADAY’S SPECULATION.
Speculation touching Electric Conduction and the nature of Matter.
By Farraday.
Having expressed my objections to Farraday’s inferences
respecting matter, &c., I feel that justice requires that I should
submit the article which drew forth my strictures. At this time, when
electricity and matter are scrutinized with a view to understand the
analogous, but different, entities of the spirit world, the ideas of an
experimental investigator so eminently successful, must be of
interest to readers:
Light and electricity are two great and searching
investigators of the molecular structure of bodies, and it
was, while considering the probable nature of conduction
and insulation in bodies not decomposable by the
electricity to which they were subject, and the relation of
electricity to space contemplated as void of that which, by
the anatomist, is called matter, that considerations,
something like those which follow, were presented to my
mind.
 If the view of the constitution of matter, already
referred to, be assumed to be correct, and I may be
allowed to speak of the particles of matter and of the
space between them (in water, or in the vapour of water,
for instance) as two different things, then space must be
taken as the only continuous part, for the particles are
considered as separated by space from each other. Space
will permeate all masses of matter in every direction like a
net, except that, in place of meshes, it will form cells,
isolating each atom from its neighbours, and itself only
being continuous.
Then take the case of a piece of shellac, a non-
conductor, and it would appear at once, from such a view
of its constitution, that space is an insulator, for if it were
a conductor, the shellac could not insulate, whatever
might be the relation as to conducting power of its
material atoms; the space would be like a fine metallic
web penetrating it in every direction, just as we may
imagine of a heap of siliceous sand having all its pores
filled with water, or as we may consider of a stick of black
wax, which, though it contains an infinity of particles of
conducting charcoal diffused through every part of it,
cannot conduct, because a non-conducting body (a resin)
intervenes and separates them one from another like the
supposed space in the lac.
Next take the case of a metal, platinum or potassium,
constituted, according to the atomic theory, in the same
manner. The metal is a conductor; but how can this be,
except space be a conductor, for it is the only continuous
part of the metal, and the atoms not only do not touch,
(by the theory,) but, as we shall see presently, must be
assumed to be a considerable way apart. Space,
therefore, must be a conductor, or else the metals could
not conduct, but would be in the situation of the black
sealing-wax referred to a little while ago.
 But if space be a conductor, how then can shellac,
sulphur, &c. insulate? for space permeates them in every
direction. Or, if space be an insulator, how can a metal or
other similar body conduct?
It would seem, therefore, that in accepting the ordinary
atomic theory, space may be proved to be a non-
conductor in non-conducting bodies, and a conductor in
conducting bodies; but the reasoning ends in this, a
subversion of that theory altogether, for if space be an
insulator, it cannot exist in conducting bodies, and if it be
a conductor, it cannot exist in insulating bodies. Any
ground of reasoning which tends to such conclusions must
in itself be false.
In connection with such conclusions, we may consider
shortly what are the probabilities that present themselves
to the mind, if the extension of the atomic theory which
chemists have imagined be applied in conjunction with the
conducting powers of metals. If the specific gravity of the
metals be divided by the atomic numbers, it gives us the
number of atoms, upon the hypothesis, in equal bulks of
the metals. In the following table the first column of
figures expresses nearly the numbers of atoms in, and the
second column of figures the conducting power of, equal
volumes of the metals named:
Atoms. Conducting power.
1·00 gold 6·00
1·00 silver 4·66
1·12 lead 0·52
1·30 tin 1·00
2·20 platinum 1·04
2·27 zinc 1·80
2·87 copper 6·33
2·90 iron 1·00

So here iron, which contains the greatest number of

atoms in a given bulk, is the worst conductor excepting
one. Gold, which contains the fewest, is nearly the best
conductor; not that these conditions are in inverse
proportions, for copper, which contains nearly as many
atoms as iron, conducts better still than gold, and with
above six times the power of iron. Lead, which contains
more atoms than gold, has only about one-twelfth of its
conducting power; lead, which is much heavier than tin
and much lighter than platina, has only half the
conducting power of either of these metals. And all this
happens among substances which we are bound to
consider at present as elementary or simple. Whichever
way we consider the particles of matter and the space
between them, and examine the assumed constitution of
matter by this table, the results are full of perplexity.
Now let us take the case of potassium, a compact
metallic substance with excellent conducting powers—its
oxide or hydrate a non-conductor; it will supply us with
some facts having very important bearings on the
assumed atomic construction of matter.
When potassium is oxidized, an atom of it combines
with an atom of oxygen to form an atom of potassa, and
an atom of potassa combines with an atom of water,
consisting of two atoms of oxygen and hydrogen, to form
an atom of hydrate of potassa, so that an atom of hydrate
of potassa contains four elementary atoms. The specific
gravity of potassium is 0·865, and its atomic weight 40·;
the specific gravity of cast hydrate of potassa, in such a
state of purity as I could obtain it, I found to be nearly 2;
its atomic weight, 57. From these, which may be taken as
facts, the following strange conclusions flow: A piece of
potassium contains less potassium than an equal piece of
the potash formed by it and oxygen. We may cast into
potassium oxygen, atom for atom, and then again both
oxygen and hydrogen in a twofold number of atoms, and
with all these additions the matter shall become less and
less, until it is not two-thirds of its original volume. If a
given bulk of potassium contains 45 atoms, the same bulk
of hydrate of potassa contains 70 atoms nearly of the
metal potassium, and, besides that, 210 atoms more of
oxygen and hydrogen. In dealing with assumptions, I
must assume a little more for the sake of making any kind
of statement; let me therefore assume that in the hydrate
of potassa the atoms are all of one size and nearly
touching each other, and that in a cubic inch of that
substance there are 2800 elementary atoms of potassium,
oxygen, and hydrogen; take 2100 atoms of oxygen and
hydrogen, and the 700 atoms of potassium remaining will
swell into more than a cubic inch and a half; and if we
diminish the number until only those containable in a
cubic inch remain, we shall have 430, or thereabout. So a
space which can contain 2800 atoms, and among them
700 of potassium itself, is found to be entirely filled by 430
atoms of potassium, as they exist in the ordinary state of
that metal. Surely, then, under the suppositions of the
atomic theory, the atoms of potassium must be very far
apart in the metal, i. e. there must be much more of
space than of matter in that body; yet it is an excellent
conductor; and so space must be a conductor, but then
what becomes of shellac, sulphur, and all the insulators?
for space must also, by the theory, exist in them.
Again, the volume which will contain 430 atoms of
potassium, and nothing else while in the state of metal,
will, when that potassium is converted into nitre, contain
very nearly the same number of atoms of potassium, i. e.
416, and also then seven times as many, or 2912 atoms,
of nitrogen and oxygen beside. In carbonate of potassa,
the space which will contain only the 430 atoms of
potassium as metal, being entirely filled by it, will, after
the conversion, contain 256 atoms more of potassium,
making 686 atoms of that metal, and in addition 2744
atoms of oxygen and carbon.
These and similar considerations might be extended
through compounds of sodium and other bodies, with
results equally striking, and indeed more so, when the
relations of one substance, as oxygen and sulphur, with
different bodies are brought into comparison.
I am not ignorant that the mind is most powerfully
drawn by the phenomena of crystallization, chemistry, and
physics generally to the acknowledgment of centres of
force. I feel myself constrained, for the present,
hypothetically to admit them, and cannot do without
them; but I feel great difficulty in the conception of atoms
of matter which in solids, fluids, and vapours are
supposed to be more or less apart from each other, with
intervening space not occupied by atoms, and perceive
great contradictions in the conclusions which flow from
such a view.
If we must assume at all, as indeed in a branch of
knowledge like the present we can hardly help it, then the
safest course appears to be to assume as little as
possible; and in that respect the atoms of Boscovich
appear to me to have a great advantage over the more
usual notion. His atoms, if I understand aright, are mere
centres of forces or powers, not particles of matter in
which the powers themselves reside. If in the ordinary
view of atoms, we call the particle of matter away from
the powers a, and the system of powers or forces in and
around it m, then in Boscovich’s theory a disappears, and
is a mere mathematical point, while in the usual notion it
is a little, unchangeable, impenetrable piece of matter, and
m is an atmosphere of force grouped around it.
In many of the hypothetical uses made of atoms, as in
crystallography, chemistry, magnetism, &c., this difference
in the assumption makes little or no alteration in the
results; but in other cases, as of electric conductors, the
nature of light, the manner in which bodies combine to
produce compounds, the effect of forces, as heat or
electricity, upon matter, the difference will be very great.
Thus, referring back to potassium, in which as a metal
the atoms must, as we have seen, be, according to the
usual view, very far apart from each other, how can we for
a moment imagine that its conducting property belongs to
it any otherwise than as a consequence of the properties
of the space, or, as I have called it above, the m? So also
its other properties in regard to light, or magnetism, or
solidity, or hardness, or specific gravity, must belong to it,
in consequence of the properties or forces of the m, not
those of the a, which, without the forces, is conceived of
as having no powers. But then, surely, the m is the matter
of the potassium, for where is there the least ground
(except in a gratuitous assumption) for imagining a
difference in kind between the nature of that space
midway between the centres of two contiguous atoms,
and any other spot between these centres? A difference in
degree or even in the nature of the power consistent with
the laws of continuity I can admit, but the difference
between a supposed little hard particle and the powers
around it, I cannot imagine.
To my mind, therefore, the a or nucleus vanishes, and
the substance consist of the powers or m; and indeed
what notion can we form of the nucleus independent of its
powers? All our perception and knowledge of the atom,
and even our fancy, is limited to ideas of its powers; what
thought remains on which to hang the imagination of an a
independent of the acknowledged forces? As mind just
entering on the subject may consider it difficult to think of
the powers of matter independent of a separate
something to be called the matter, but it is certainly far
more difficult, and indeed impossible to think of or
imagine that matter independent of the powers. Now, the
powers we know and recognise in every phenomena of
the creation, the abstract matter in one; why, then,
assume the existence of that of which we are ignorant,
which we cannot conceive, and for which there is no
philosophical necessity?
Before concluding these speculations, I will refer to a
few of the important differences between the assumption
of atoms consisting merely of centres of force like those of
Boscovich, and that other assumption of molecules of
something specially material, having powers attached in
and around them.
With the latter atoms a mass of matter consists of
atoms and intervening space; with the former atoms
matter is everywhere present, and there is no intervening
space unoccupied by it. In gases the atoms touch each
other just as truly as in solids. In this respect the atoms of
water touch each other, whether that substance be in the
form of ice, water, or steam; no mere intervening space is
present. Doubtless, the centres of force vary in their
distance one from another, but that which is truly the
matter of one atom touches the matter of its neighbours.
Hence matter will be continuous throughout, and in
considering we have not to suppose a distinction between
its atoms and any intervening space. The powers around
the centres give these centres the properties of atoms of
matter; and these powers again, when many centres by
their conjoint forces are grouped into a mass, give to
every part of that mass the properties of matter. In such a
view all the contradiction resulting from the consideration
of electric insulation and conduction disappears.
The atoms may be conceived of as highly elastic,
instead of being supposed excessively hard and
Welcome to Our Bookstore - The Ultimate Destination for Book Lovers
Are you passionate about books and eager to explore new worlds of
knowledge? At our website, we offer a vast collection of books that
cater to every interest and age group. From classic literature to
specialized publications, self-help books, and children’s stories, we
have it all! Each book is a gateway to new adventures, helping you
expand your knowledge and nourish your soul
Experience Convenient and Enjoyable Book Shopping Our website is more
than just an online bookstore—it’s a bridge connecting readers to the
timeless values of culture and wisdom. With a sleek and user-friendly
interface and a smart search system, you can find your favorite books
quickly and easily. Enjoy special promotions, fast home delivery, and
a seamless shopping experience that saves you time and enhances your
love for reading.
Let us accompany you on the journey of exploring knowledge and
personal growth!

ebookgate.com