ORDINARY DIFFERENTIAL EQUATIONS . JACK K. HALE
KRIEGER PUBLISHING COMPANY MALABAR, FLORIDA
Original Edition 1969 Second Edition 1980 Printed and PUblished by ROBERT E. KRIEGER PVBUSBING COMPANY, INC. KRIEGER DRIVE
M.UABAR, FLORIDA 3~
e e
CopYright 1969 (Original Material) by JOHN WILEY lie SONS,INC. Copyright 1980 (New Material) by . ROBERT E. KRIEGER PUBLISHING COMPANY, INC.
AU rights reserved. No reproduction in tin)' form of this boole, in whole or in Ptlrt (e3Ccept for brief quottltiqn in critical articles or , .. cJiews),rntl)' be mtlde without written GUthoriztlnon from the publisher. Printed in the United States of America
Library of Congress Cataloging in Publication Data Hale, Jack K. Ordinary differential equations. Second edition of original published by Wiley-Interscience, New York, which was issued as v. 21 of Pure and applied mathematics. Bibliography: p. Includes index. 1. Differential equations. I. Title. [QA372.H184 1980] 515.'.352 79-17238 ISBN 0-89874-011-8 10
9
8
7
6
Preface
This book is the outgrowth of a course given for a number of years in the Division of Applied Mathematics at Brown University. Most of the students were in their first and second years of graduate study in applied mathematics, although some were in engineering and pure mathematics. The purpose ofthe book is threefold. First, it is intended to familiarize the reader with some of the problems and techniques in ordinary differential equations, with the emphasis on nonlinear problems. Second, it is hoped that the material is presented in a way that will prepare the reader for intelligent study of the current literature and for research in -differential equations. Third, in order not to lose sight of the applied side of the subject, considerable space has been devoted to specific analytic!J.1 methods which are presently widely used in the applications. Since the emphasis throughout is on nonlinear phenomena, the global theory of two-dimensional systems has been presented immediately after the fundamental theory of existence, uniqueness, and continuous depen-· dence. This also has the advantage of giving the student specific examples and concepts which serve to motivate study of later chapters. Since a satisfactory global theory for general n-dimensional systems is not available, we naturally turn to local problems and, in particular, to the behavior of solutions of differential equations near invariant sets. In the applications it is necessary not only to study the effect of variations of the initial data but also in the vector field. These are discussed in detail in Chapters III and IV in which the invariant set is an equilibrium point. In this way many of the basic and powerful methods in differential equations can be examined at an elementary level. The, analytical methods developed in these chapters are immediately applicable to the most widely used technique in the practical theory of nonlinear oscillations, the method of averaging, which is treated in Chapter V. When the invariant set corresponds to a periodic orbit and only autonomous perturbations in the vector field are permitted, the discussion is similar to that for an equilibrium point and is given in Chapter VI. On the other hand, when the perturbations in the vector field are nonautonomous or the invariant set is a closed curve with equilibrium points, life is not so simple. In Chapter VII an attempt has been made to present this more complicated ix
x
PREFACE
and important subject in such a way that the theory is a natural generalization of the theory in Chapter IV. Chapter VIII is devoted to a general method for determining when a periodic differential equation containing a small parameter has a periodic solution. The reason for devoting a chapter to this subject is that important conclusions are easily obtained for Hamiltonian systems in this framework and the method can be generalized to apply to problems in other fields such as partial differential, integral, and functional differential equations. The abstract generalization is made in Chapter IX with an application to analytic solutions of linear systems with a singularity, but space did not permit applications to other fields. The last chapter is devoted to elementary results and applications of the direct method of Lyapunov to stability theory. Except for Chapter I this topic is independent of the remainder of the book and was placed at the end to preserve continuity of ideas. For the sake of efficiency and to acquaint the student with concrete applications of elementary concepts from functional analysis, I have presented the material with an element of abstraction. Relevant background material appears in Chapter 0 and in the appendix on almost periodic functions, although I assume that the reader has had a course in advanced calculus. A one-semester course at Brown University usually covers the saddlepoint property in Chapter III; the second semester is devoted to selections from the remaining chapters. Throughout the book I have made suggestions for further study and have provided exercises, some of which are difficult. The difficulty usually arises because the exercises are introduced when very little technique has been developed. This procedure was followed to permit the student to develop his own ideas and intuition. Plenty of time should be allowed for the exercises and appropriate hints should be given when the student is prepared to receive them. No attempt has been made to cover all aspects of differential equations. Lack of space, however, forced the omission of certain topics that contribute to the overall objective outlined above; for example, the general subject of bou~dary value prpblems and Green's functions belong in the vocabulary of every serious student of differential equations. This OI~ission is partly justified by the fact that this topic is usually treated in other courses in applied mathematics and, in addition, excellent presentations are available in the literature. Also, specific applications had to be suppressed, but individuals with special interest can .easily make the correlation with the theoretical results herein. I have received invaluable assistance in many conversations ."ith my colleagues and students at Brown University. Special thanks are due to C. Olech for his direct contribution to the presentation of two-dimensional systems, to M. Jacobs for his thought-provoking criticisms of many parts of
xi
PREFACE
the original manuscript, and to W. S. Hall and D. Sweet for their comments. I am indebted to K. Nolan for her endurance in the excellent preparation of the manuscript. I also wish to thank the staff of Interscience for being so efficient and cooperative during the production process.
Jack K. Hale Providence, Rhode Island September, 1969
Preface to Revised Edition For this revised edition, I am indebted to several colleagues for their assistance in the elimination of misprints and the clarification of the presentation. The section on integral manifolds has been enla.rged to include a more detailed discussion of stability. In Chapter VIII, new material is included on Hopf bifurcation, bifurcation with several independent parameters and subharmonic solutions. A new section in Chapter X deals with Wazewski's principle. The Appendix on almost periodic functions has been completely rewirtten using the modern definition of Bochner. lack K. Hale Apri11980
Contents
CHAPTER
O.
Mathematical preliminaries 0.1. Banach spaces and examples 0.2. Linear transformations 0.3. Fixed point theorems
1 1
3 4
CHAPTER I.
General properties of differential equations 1.1. 1.2. 1.3. 1.4. 1.5. 1.6. 1.7. 1.8. 1.9.
Existence Continuation of solutions Uniqueness and continuity properties Continuous dependence and stability Extension of the concept of a differential equation Differential inequalities Autonomous systems-generalities Autonomous systems-limit sets, invariant sets Remarks and suggestions for further study
12
12 16 18
25 28 30 37
46 49
CHAPTER II.
Two dimensional systems ILL II.2. II.3.
J;>lanar two dimensional systems-the PoincareBendixson theory Differential systems on a torus RemarkS and suggestions for further study.
51 51
64 76
CHAPTER III.
Linear systems and linearization IILl.
General linear systems
78 79
xiv
CONTENTS
III.2. III.3. III.4. III.5. III.6. III.7. III.8. III.9. III.I0. III.H.
Stability of linear and perturbed linear systems nth Order scalar equations Linear systems with constant coefficients Two dimensional linear autonomous systems The saddle point property Linear periodic systems Hill's equation Reciprocal systems Canonical systems Remarks and suggestions for further study
83 89 93 101 106 H7 121 131 136 142
OHAPTER IV.
Perturbations of noncritical linear systems IV.1. Nonhomogeneous linear systems IV.2. Weakly nonlinear equatioJl&-,-noncritical case IV.3. The general saddle point property IV.4. More general systems IV.5 The Duffing equation with large damping and large forcing N.6 Remarks and extensions
144 145 154 156 162 168
171
OHAPTERV.
Simple oscillatory phenomena and the method of averaging V.l. Conservative systems V.2. Nonconservative second order equations-limit -oycles V.3. Averaging V.4. The forced van der Pol equation V.5. Duffing's equation with small damping and small harmonic forcing V.6. The subharmonic of order 3 for Duffing's equation V.7. Damped excited pendulum with oscillating support V.8. Exercises V.9. Remarks and suggestions for further study
175 176 184 190 198 199 206 208 210 211
OHAPTER VI.
Beha.vior near a. periodic orbit VLI.
A local coordinate system about an invariant closed curve
213 214
CONTENTS
VI.2. VI.3.
Stability of a periodic orbit Sufficient conditions for orbital stability in two dimensions Autonomous perturbations Remarks and suggestions for further study
VI.4. VI.5.
xv
219 '224 226 227
OHAPTER VII.
Integral manifolds of equations with a small parameter VII.l. VII.2. VII.3. VII.4. VII.5. VII.H. VII.7. VII.S. VII.9.
Methods of determining integral manifolds Statement of results A" nonhomogeneous linear" system The mapping principle Proof of Theorem 2.1 Stability of the·perturbed manifold Applications Exercises Remarks and suggestions for further study
229
231 236 239 245 247 248 250 254 256
OHAPTER VIII.
Periodic systems with a small parameter VIII.I. VIII.2. VIII.3. VIII.4.
A special system of equations Almost linear systems Periodic solutions of perturbed autonomous equations Remarks and suggestions for further study
258 259
275 294 296
OHAPTER IX.
Alternative problems for the solution of functional equations IX.I. IX.2. IX.3. IX.4. IX.5. IX.H.
Equivalent equations A generalization Alternative problems Alternative. problems for periodic solutions The Perron-Lettenmeyer th.e0rem Remarks and suggestions for further study
298 299
302 303 304 307 309
xvi
CONTENTS
x. The direct method of Liapunov CHAPTER
X.I.
X.2. X.3. X.4. X.5. X.6. X.7.
Sufficient conditions for stability and instability in autonomous systems Circuits containing Esaki diodes Sufficient conditions for stability in nonautonomous systems The converse theorems for asymptotic stability Implications of asymptotic stability Wazewski's principle Remarks and suggestions for further study
311 311
320 324 327 331 333 338
APPENDIX
Almost periodic functions References Index
339
352 360
CHAPTER 0 Mathematical Preliminaries
In this chapter we collect a number of basic facts from analysis which play an important role in the theory of differential equations.
0.1. Banach Spaces and Examples
Set intersection is denoted by ( l , set union by u, set inclusion by c and XES denotes x is a member of the· set S. R (or C) will denote the real (or complex) field. An abstract linear vector space (or linear spq,ce) f!( over R (or C) is a collection of elements {x, y, ... } such that for each x, y in f!(, the sum x y is defined, x y E f!(, x y = y x and there is an element 0 in f!( such that x 0 = x for all x E f!(. Also for any numb~r a, b E R (or C), scalar multiplication ax is defined, aXEf!( and l·x=x, (ab)x=a(bx)=b(ax), (a+b)x= ax by for all x, y E f!(. A linear space f!( is a normed linear space if to each x in f!(, there corresponds a real number called the norm of x which satisfies
+
+
+
+
*
+
+
Ixl
Ixl > 0 for x 0, 101 = 0;. Ix + yl ~ Ixl + Iyl (triangle inequality); laxl = lal·lxl for all a in R (or C) and x in X. confusion may arise, we will write 1·1 fl: for the norm function on f!(.
(i) (ii) (iii)
When A sequence {xn} in a normed linear space f!( converges to x in f!( if limn->oo IX n - xl = O. We shall write this as lim Xn = x. A sequence {x n } in f!( 'is a Cauchy sequence if. for every e > 0, there is an N(e) > 0 such that IXn - xml < 13 if n, m ~ N(e). The space f!( is complete if every Cauchy sequence in f!( converges to an element of f!(. A complete normed linear space is a Banach space. The e-neighborhood of an element x of a normed linear space f!( is {y in f!(: Iy - xl < e}. A set S in f!( is open if for every XES, an e-neighborhood of x is also contained in f!(. An element x is a limit point of a set S if each e-neighborhood of x contains points of S. A set S is closed if it contains its limit points. The closure of a set S is the union of S and its limit points. A set S is dense in f!( if the closure of S is f!(. If S is a subset of f!(,
ORDINARY DIFFERENTIAL EQUATIONS
2
A is a subset of R and Va, a E A is a collection of open sets of f£ such that E A Va ::::l S, then the collection it a is called an open covering of S. A set S in :r is compact if every open covering of S contains a finite number of open sets which also cover S. For Banach spaces, this is equivalent to the following; a set S in a Banach space is compact if every sequence {xn}, Xn E S, cont.ains a subsequence which converges to an element of S. A set S in f£ is bounded if there exists an r > 0 such that S c {x E f£; Ixi < r}.
Ua
Example 1.1. Let Rn(Cn) be the space of real (complex) n-dimensional oolumn vectors. For a particular coordinate system, elements x in Rn(Cn) will be written as x = (Xl, ... , xn) where each Xj is in R(C). If x = (Xl, ... , xn), Y = (Yi, . .. , Yn) are in Rn(Cn), then ax + by for a, bin R(C) is defined to be (axI + bYI, ... , aXn + bYn). The space Rn(Cn) is clearly a linear space. It is a Banach space if we choose lxi, X = COI(XI, ... , xn), to be either supilxil, Ii IXil or [Ii IXiI 2 ]Yf. Each of these norms is equivalent in the sense that a sequence converging in one norm converges in any of the other norms. Rn(cn) is complete because convergence implies coordinate wise convergence and R(C) is complete. A set S in Rn(Cn) is compact if and only if it is closed and bounded. EXERCISE 1.1. If E is a finite dimensional linear vector space and 1'1, are two norms on E, prove there are positive constants m, M such that m Ixl ~ IlxJJ ~ M Ixi for all x inE. 11'11
Example 1.2.
Let D be a compact subset of Rm [or cm] and Cf5(D, Rn)
[or Cf5(D, On)] be the linear space of continuous functions which take D into
Rn [br cn]. A se'quence of functions {cpn, n = I, 2, ... } in Cf5(D, Rn) is said to converge uniformly on D if there exists a function cp taking D into Rn such that for every s > 0 there is an N(s) (independent of n) such that Icpn(x) - cp(x)1 < s for all n ;;;;; N(s) and x in D. A sequence {c/in} is said to be uniformly bounded if there exists an M > 0 such that Icpn(x)1 < M for all x in D and all n = I, 2, .... .A sequence {CPn} is said to be equicontinuous if, for every s > 0, there is a ;) > 0 such that n
=
1,2, . ~.,
if Ix - yl < S, x, yin D. A function f in Cf5(D, Rn) is said to be Lipschitzian in D if there is a constant K such that If(x) - f(y)1 ~ Klx - yl for all x, y, in D. The most frequently encountered equicontinuous sequences in Cf5(D, Rn) are sequences {cpn} which are Lipschitzian with a Lipschitz constant independent of n.
j.
LEMMA 1.1. (Ascoli-Arzela). Any uniformly bounded equicontinuous sequence of functions in Cf5(D, Rn) has a subsequence which converges uniformlyon D.
MATHEMATICAL PRELIMINARIES
3
LEMMA 1.2. If a sequence in Cff(D, Rn) converges uniformly on D, then the limit function is in Cff(D, Rn). If we define
It/>I =maxlc/>(x)l, XE
D
then one easily shows this is a norm on Cff(D, Rn) and the above lemmas show that Cff(D, Rn) is a Banach space with this norm. The same remarks apply to Cff(D, On). EXERCISE 1.2. Suppose m = n = 1. Show that Cff (D, R) is a normed linear space with the norm defined by
f
lit/> II = DIt/>(x)1 dx. Give an example to show why this space is not complete. What is the completion of this space?
0.2. Linear Transformations A function taking a set A of some space into a set B of some space will be referred to as a transformation or mapping of A into B. A will be called the domain of the mapping and the set of values of the mapping will be called the range of the mapping. If f is a mapping of A into B, we simply write f: A ~ B and denote the range of f by f(A). If f: A ~ B is one to one and continuous together with its inverse, then we say f is a homeomorphism of A onto B. If fIl, i5Jj are real (or complex) llitnach spaces andf: fIl ~i5Jj is such that f(alxl a2 X2) = atf(xI) a2 f(X2) for all Xl, x2 in fIl and all real (or complex) numbers aI, a2, thenfis called a linear mapping. A linear mappingf of fIl into i5Jj is said to be bounded ifthere is a constant K such that If(x)lw:;:;; Klxlx for all X in fIl.
+
+
LEMMA 2.1. Suppose fIl, i5Jj are Banach spaces. A linear mapping f: fIl ~ i5Jj is bounded if and only if it is continuous. ")
EXERCISE
2.1.
Prove this lemma.
EXERCISE 2.2. Show that each linear mapping of Rn (or On) into Rm (or om) can be represented by an m X n real (or complex) matrix and is therefore necessarily continuous. The norm If I of a continuous linear mapping f: fIl ~i5Jj is defined as
If I = sup{lfxlw: Ixlx = I}. It is easy to show that If I defined in this way satisfies the properties (i)-(iii)
4
ORDINARY DIFFERENTIAL EQUATIONS
in the definition of a norm and also that for all x in fr.
If a linear map taking an n-dimensional linear space into an m-dimensional linear space is defined by an m X n matrix A, we write its norm as IAI. EXERCISE 2.3. If~, I1l/ are Banach spaces, let L(~, 11l/) be the set of bounded linear operators taking ~ into 11l/. Prove the L(~, 11l/) is a Ban.1ch space with the norm defined above.
Example 2_1.
Define I: ~([O, 1], R)_R by 1(4)) =
f: 4>(8) ds. The map
1 is linear and continuous with III = 1. Example 2.2. Define S = {4> in ~([O, 1], R) which have a continuous first derivative}. S is dense in ~([O, 1], Rn). }'or any 4> in S, define N(t) = d4>(t)/dt, 0 ~ t ~ 1. The function 1 is linear but not bounded. In fact, the sequence of functions 4>n(t) = tn, 0 ~ t ~ 1, satisfies II4>n II = 1, but liNn !I = n. Another way to show unboundedness is to prove 1 is not continuous. Consider the functions 4>n(t) = t n - t n+1, 0 ~ t ~ 1, n ~ 1. In ~([O, 1], R), 4>n _0 as n _ 00, but Nn(t) = t n- 1 [n - (n l)t] and 1""'(1) = -1, which does not approach zero as n _ 00. Another very important tool from functional analysis which· can be used, frequently in differential equations is the principle of uniform boundedness. In this book, we have chosen to circumvent this principle by using more elementary proofs except in one instance in Chapter IV.
+
Principle
0/ uniform boundedness
Suppose d is an index set and Ta., ex in d, are bound~d linear maps from a Banach space ~ to a Banach space I1l/ such that for each x in ~, sUPCl,in.otITa.xl < 00. Then sUPClln.otITa.] < 00.
0.3 Fixed Point Theorems A fixed point of a transformation T: ~ _ ~ is a point x in ~ 'such that Tx = x. Theorems concerning the existence of a fixed point ofa transformation are very convenient in differential equations even though not absolutely necessary. Such .theorems should be considered as a tool which aVQids the repetition of standard arguments and permits one to concentrate on tJe essential elements of the problem. . . One standard tool in analysis is successive approximations. The basic elements of the method of successive approximations have been abstracted
'"
MATHEMATICAL PRELIMINARIES
5
. 1 b)y Banach andCacCIoporl. into t he so calId e contractIOn mappmg prmCIple IffF is a· subset of a Banach space fif and T is a transformation taking fF into a Banach space gB (written as T: fF -+!!lJ), then T is alcontractionlon fF if there is a ,\, 0 ~ ,\ < 1, such that ITx - Tyl ~ '\Ix -yl,
x, y E fF.
The constant ,\ is called the contraction constant for T on fF.
I THEOREM 3.l. (Contraction mawinqprinci'}2le of Banach-Caccio'Doli\. If fF is a closed subset of a Banach space fif and T: fF -+fF is a contraction on fF, then T has a unique fixed point x in fF. Also, if Xo in fF is arbitrary, then the sequence {Xn+l = Tx n , n =0, 1, 2, ... } converges to x as n-+ 00 and Ix - xnl ~ ,\n IXI - xol/(1 - ,\), where ,\ < 1 is the contraction constant for T onfF.
,\ < 1 is the contraction constant for T on fF, x=Tx, y=Ty, x, YEfF, then Ix-yl =ITx-TYI ~'\Ix-yl. This implies Ix -.-:.. yl ~ 0 and thus Ix - yl = 0 and x = y. PROOF. . Uniqueness.
If 0
~
Existence. Let Xo be arbitrary, and Xn+l = TXn, n = 0, 1, 2, .... _By hypotheses, each Xn, n =0, 1, ... , is in fF. Also, IXn+1 - Xnl ~ ,\ IXn - Xn-ll ~ ... ~ ,\nIXI-Xol, n =0,1, .... Thus, for m >n,
+ IXm-I- Xm-21 + ... + IXn+I-Xnl ~ [,\m-l + ,\m-2 + :.. + ,\n] IXI - xol = ,\n[1 + ,\ + ... + ,\m-n-l] IXI - xol
IXm -Xnl ~ IXm -xm-ll
,\n[1 _ ,\m-n]
=
,\ 1-
,\n
IXI-xol ~ - - \ IXI-Xol· 1-1\
Thus the sequence {xn} forms a Cauchy sequence and there is an x in fif such that limn-..oo Xn = x. Since fF is closed, x is in fF. Since T is continuous and 1·1 is continuous (the latter because Ixi -Ixn - xl ~ IXnl ~ IXn - xl lxi), it follows that
+
0= lim IXm+1 - TXml = Ilim [xm+1 - TXm]1 = Ix -Txl, ., m-+
00
m ... 00
which implies Tx = x. This gives the existence of a fixed point. To prove the last estimate, take the limit asm -+ 00 in the previous estimate of IXm - xnl. This completes the proof of the theorem. EXERCISE 3.l. . Suppose fif is a Banach space, T:'PI" -+fif is a continuous linear operator with ITI < 1. Show that / - T, / the identity operator, has a bounded inverse; that is, prove that the equation (/ - T)x = y has a unique solution x in fif which depends continuously upon y in fif.
6
ORDINARY DIFFERENTIAL EQUATIONS
Let [!t, qy be Banach spaces and D be an open set in [!t. A function I: J) -+ qy is said to be (Frechet) differentiable at It point. x in D if there is a bounded linear operator A(x) taking [!t·-+qysuch that for every he [!t with x+he D, .. I/(x
+ h) -
I(x) -A(x)hl ~ p(lhl, x),
where p(lhl, x) satisfies p(lhl, x)!lhl-+O as 111,1-+0. The linear operator A(x) is called the derivative 011 at x and A(x)h the differential 011 at x.
-
}1XERCISE 3.2. Supppse [!t, qy are Banach spaces, ;s;I is an open subset of [!t, f: d -+qy, and .... ()1. . l(xo+th}~/(x.o) 11m ... =w Xo fh I-+f)
.
t
(};ldsts for every Xo ed, 11, Eo f!£, where the limit is taken fort real. Suppose (l)(xo) is It continuous linear mapping for all Xo e d and suppose w considered as a mapping from d into L(f!£, qy) is continuou~, whereL([!t, qy) is defined in Exercise 2.3. Prove that I is (Frechet) differentiable with derivative w(xo) at Xo e.9l. EXERCISE. 3.3. Prove that w(xo) of the previous exercise can exist. for every 11" be a continuous linear mapping and not be the Frechet derivative of/at Xo e d. EXERCISE 3.4. Let ~1([0, 1], Rn) denote the space of continuously differentiable functions x: [0,1] -+ Rn, with addition and scalar multiplication defined in the usual way and let Ixl = supostsllx(t)1 suposl~llx(t)1 where x(t) = dx!dt. Prove ~1([0, 1], Rn) is it Banach ~pace. - -
+
EXERCISE 3.5. Let w: ~1([0, 1], Rn) X [0, 1] -+ Rn be the evaluation mapping, w(x, t) = x(t). Prove that w is continuously differentiable and compute its derivative. Do this exercise two ways; using the definition of the derivative and also Exercise 3.2. If I: Rm -+ Rn is differentiable at a point x, then A(x) = o/(x)!ox = (olt(x)!oXj, i = 1, 2, ... ,n, j = 1, 2, ... , m), is the Jacobian matrix of I with respect to x. For later reference, it is convenient to have notation for order relations. We shall say a function I(x) = O(lxl) as Ixl-+O if I/(x)I!lxl is bounded for x in a neigl1borhood of zero and I(x) =o(lxl) as Ixl-+O if I/(x)I!lxl-+O as Ixl-+O.
Suppose fF is a subset of a Banach space [!t, r§ is a subset of a Banach space qy and {Til' ye r§} is a family of operators taking fF -+[!t. Th~.~perator Til is said to be a unilorm contraction on fF if T y' fF -+ fF and there is a A, 0 ~A < 1 such that ITyx - Tllxl ~ Alx -xl
for all y in
r§, x, x
in fF.
7
MATHEMATICAL PRELIMINARIES
In other words, T" is a contraction for each y in t'§ and the contraction con· stant can be chosen independent of y in t'§ . .THEOREM 3.2. If fF is a closed subset of a Banach space !!(, t'§ is a subset of a Banach space dJI, T,,: fF ->;-fF, yin t'§ is a uniform contraction on fF and T" x is continuous in y for each fixed x in fF, then the unique fixed point g(y) of T", yin r§, is continuous in y. Furthermore, if fF, t'§ are the closures of open sets fFo, t'§0 and T"x has continuous first derivatives A(x, y), B(x, y) in y, x, respectively, then g(y) has a continuous fjrst derivative with respect to tI in t'§0. COROLLARY 3.l.Suppose !!( is a Banach space, t'§ is a subset of a Banach space dJI, A,,: !!( ->;-!!( are continuous linear operators for each y in t'§, IA"I ~ < 1 for all yin G and A"x is continuous in y for each x in X. Then the operator I - A" has a bounded inverse which depends continuously upony, I(J-Ayr11 ;£(1-8r 1.
s
Since T,,:
PROOFS.
is a uniform contraction, there is a A, xl for all y in t'§, x, x in fF. Let g(y) be the unique fixed point of T" in fF which exists from Theorem 3.l. Then fF ->;- fF
o ~ A < 1 such that IT"x -
g(y
+ h) -
and Ig(y
g(y)
+ h) -
T"xl ~ AIx -
= T,,+kg(y + h) = T,,+kg(y + h) -
g(Y)1 ~
T"g(y) T,,+kg(y)
AIg(y + h) - g(y)1
+ T,,+kg(y) -
+ IT,,+kg(y) -
T"g(y),
T"g(y)l·
This implies Ig(y
+ h) -
g(y)1 ~ (1 - A)-II T,,+kg(y) - T"g(y)l·
Since Tyx is continuous in y for each fixed x in iF, we see that g(y) is continuouI;!. This proves the first part of the theorem. The proof of Corollary 3.1 is now almost immediate. In fact, we need to show that the equation x - A" x = z has a unique solution for each z in !!( and this solution depends continuously upon y, z. This is equivalent to finding the fixed points of the operator T",.« defined by T II, z x = .4" x + z, x in !!(. Since Ay' is a uniform contraction, (J - Ay 1 exists, is bounded and continuous in y. Also, if (J -Ay)x = y, then Iyl ~ (1 - 8)lxl and the proof of Corollary 3.1 is complete. To prove the last part of the theorem, suppose T"x has continuous first derivatives A(x, y), B(x, y) with respect to y, x, respectively, for y E '(90, X E fFo. Let us use the fact that g(y) = T"g(y) and try to find the equation that the differential z = C(y)h, h in dJI of g will have to satisfy if ghas a derivative C(y). If the chain rule of d·ifferentiation is valid then
r
\o>.~ I
z = B(g(y), y)z
+ A(g(y), y)h
ORDINARY DIFFERENTIAL EQUATIONS
8
where h is an arbitrary element of d,Ij. It is easy to show that T 11 being a uniform contraction implies IB(x,y)1 < 8< 1.forxin [F0, yin C§0. Since IB(x, y)1 < 8 < 1 for x in [F0, yin C§0, an application of Corollary 3.1 implies, for eachy inC§°, h in d,Ij, the existence of a unique solution z(y, h) of (3.1) which is continuous in y, h. From uniqueness, one observes that z(y, lXh + flu) = rx.z(y, h) + (3z(y, u) for all scalars lX, (3 and h, u in d,Ij; that is, z(y, h) is linear in h and may be written as O(y)h, where O(y): d,Ij ~ f£ is a continuous linear operator for each y is also 'continuous in y. To show that C(y) is the derivative of g(y}, let w = g(y + h} - g(y}, B(g(y},y} = B(y}, A (g(y},y) =A(y} an~ observe that wsatisfies the equation
w - B(y}w - A (y}h + f(w, h, y) = 0 where, for any e >.0, there is a" > 0 such that If(w,h,y}1 Ihl < lI, yin r9'0. From Corollary 3.1, .
< e(lwl + Ihl}for
w- [I-B(y}]-lA (y}w-r F(w,h,y} = 0 where IF(w,h,y) I implies
< e{l
c')r 1 (lwl + Ihl} for Ihl <
-
Iwl ~Illhl,.
lI,Y
in
C§0.
But, this
1l=21[I-B(y}]-lA(y}I+l
. Iy - [I-B(y}] -lA(y}hl
~e~
1:} Ihl.
for Ihl < lI. This shows that g(y} is continuously differentiable in y and the derivative is given by C(y} satisfying Equation (3.1). This completes the proof of the tlieorem . .To illustrate the contraction principle, we prove the following important theorem of implicit functions~ In the statement of this result det A for IlIn m X m matrix A denotes the determinant of A. THEOREM 3.3. (of Implicit Functions). .Suppose F: Rm X Rn ~ Rm has continuous first partial derivatives and F(O, 0) = O. If the Jacobian matrix aF(x, y)/ax of F with respect to x satisfiesdet aF(O, O)/ax::j:; 0, .then there exist neighborhoods U, V of 0 in Rm, Rn, respectively, such that for each fixed y in V the equation F(x, y) = 0 has a unique solution x in U. Furthermore, this solution can be given as x = g(y), where g has continuous first derivatives and g(O) = O. PROOF: Let us write F(x., y)
= Ax - N(x, y),
A = aF(O, 0), ax
MATHEMATICAL PRELIMINARIES N(x, y) =
9
of(O,.O) ox x - F(x, y), N(O, 0) = 0.
From the expression for N, we have oN(x, y)
of(O, O)
ox
ox
of(x, y)
ox
The hypothesis of continuity ofoF(x, y)/ox implies oN(x, y)/ox ~O as x ~O, y~O. We th.erefore have the existence of a function k(y, p) which is continuous in y E Rn and p ~.O such that k(O, 0) = and
°
IN(x, y) - N(x, y)1
;a; k(y, p)lx -
xl
for all y in Rn and x, x with lxi, Ixl ;a; p. Since the matrix A is assumed to be nonsingular, finding a solution to F(x, y) = is equivalent to finding a solution of the equation x =A-IN(x, y), where A-I is the inverse of A. This is equivalent to finding a fixed point of the operator Ty: Rm~Rm defined by Tyx = A-IN(x, y). We now show that Ty is a contraction on an appropriate set. There is a constant K (see Exercise 2.2) such that IA-Ixl ;a; Klxl for all x in Rm and therefore
°
ITyXI
= IA-IN(x, y)1 ;a; KIN(x, y)1 = K IN(x, y) ,- N(O, y)
ITyx - TyXI
for lxi, Ixl
;a; Kk(y, p) Ixl + K ;a; Kk(y, p)jx -xl
+ (0, y)1
IN(O, y)1
;a; p and ally. Choos~ e, 8 positive and so small thai Kk(y, p)p
+ KIN(O, y)1 < e
sup{Kk(y, p), Iyl
;a; 8, p ;a;
for
Iyl;a; 8, p;a; e,
e} < 1,
. and let U = {x in Rm: Ixl < e}, V = {y in Rn: Iyl < 8}. It follows that T y is a uniform contraction of U into U for y in V o. Therefore T y has a unique fixed point g(y) in U from Theorem 3.1. It is clear that g(O) = 0. Since Tyx is continuous in y for each x it follows from Theorem 3.2 that g(y) is continuous in y. AlsoTyx has continuous first derivatives with respect to x and y with the derivative with respect to x being given by A-l oN(x, y)/ox. Theorem 3.2 implies therefore the continuous differentiability of g(y) in y and completes the proof of the implicit function theorem. EXERCISE 3.6. State and prove a generalization of Theorem 3.3 for Banach spaces. Hint: Redo the steps in the proof of Theorem 3.3 for Banach spaces making appropriate changes and hypotheses where necessary. T.he contraction mapping principle can be regarded as a fixed point theorem. Other fixed point theorems of a more sophisticated type are very
10
ORDINARY DIFFERENTIAL EQUATIONS
useful in differentia.lequations. We formulate two more which are used in this book. An obvious fixed point theorem in one dimension is the following: any continuous mapping of the closed interval [0, 1] into itself must have a fixed point. The proof is obvious .ifone simply observes that the existence of a fixed point is equivalent to .ying that the graph of the function in 2-space must cross the diagonal of the unit square with vertices at (0, 0), (1, 0), (0, 1), (1, 1). After some thought it seems plausible that a similar result should hold in h~gher dimensions but the proof is difficult. This is the celebrated BROUWER FIxED POINT THEOREM. Any continuous mapping of the closed unit ball in R'" into itself must have a fixed point. If a subset A of R'" is homeomorphic to the closed unit ball in R'" and J is a continuous mapping of A into A, then the Brouwer FiXed Point Theorem implies J has a fixed point in A. Suppose J: R'" - R'" is a continuous mapping. The zeros of the functiori J coincide with the fixed points of the mapping g defined by g(x) = x + J( x). Ifwe ca.n show that there is a set Din R'" which is homeomorphic to the closed unit ball in R'" such that g takes D into D, then the Brouwer fixe4 point theorem implies that g has a fixed point in D andfhas a zero in D. This is a very important application of the Brouwer fixed point theorem. The Brouwer fixed point theorem has been generalized to Banach spaces by Schauder and even more general spaces by Tychonov. We formulate this result for Banach spaces. Recall that a subset JII of a Banach space is compact if any sequence {4>",}, n = 1, 2, ... in JII has a subsequence which converges to an element of JII. A subset JII is convex if for x, yin JII it follow6 that ex + (1 - t)y is in JII for 0 ;;;;:; t ;;;;:; 1; that is, JII contains the" line segment " joining x and y. A mapping J of a Banach space f£ into a Banach space 0/1 is said to be compact if for every bounded set JII in f£ the closure of the set {f(x), x in JII} is compact. If, in addition, J is continuous, it is called completely continuous. SCHAUDER FIXED POINT THEOREM. If JII is a convex, compact subset of a Banach space f£ and J: JII - JII is continuous, then f has a fixed point in JII. COROLLARY, If JII is a closed, bounded, convex subset of a Banach space f£ andf: JII-Jilis completely continuous, thenJhas a fixed point in JII. The proof of the corollary proceeds as follows: Since f(JII) c: JII and JII is closed, the closure of f(JII) belongs to JII and is compact by hypothesis. Furthermore, the convex closure f!I of J(JII) [the smallest closed Convex Set containing f(JII)] belongs to JII since JII is convex: A theorem.1r Mazur states that f!I is compact. Since f!I c: JII, f(f!I) c: J(JII) c: til and the previous result implies the existence of a fixed point in f!I c: JII.
MATHEMATICAL PRELIMINARIES
11
As remarked earlier, the Schauder theorem was extended to more general spaces (locally convex linear topological spaces) by Tychonov. We do .not wish to introduce all of the terminology of locally convex linear topological spaces. In fact, the only such space for which we need the extended form of this theorem is for the space of continuous functions f: R ~Cn for which convergence in the space is equivalent to uniform convergence on compact subsets. A set is bounded in this space if all elements of the set are uniformly bounded continuous functions. The statement of the extended form of this theorem is now exactly the same as before. We will refer to the fixed point theorem in this situation as the Schauiler-Tyclwnov theorem. EXERCISE 3.7. Show the Schauder theorem is false if either the compactness or the convexity of A is eliminated. A very useful compact, convex subset of C(j(1; Rn), 1 a closed bounded interval of RI, is obtained in the following manner. Suppose M, (3 are positive constants and .91 is the subset of C(j(1, Rn) such that c/> in .91 implies Ic/>I ~ (3, Ic/>(t) - c/>O)I ~ Mit - ll, fort, lin 1. The set .91 is obviously convex and closed. Fur.thermore, any sequence {c/>n} in .91 is uniformly bounded and equicontihuous. Lemmas 1.1 and 1.2 imply the existence of a c/> in C(j(1, Rn) such that limn ... oo c/>n = c/>. But .91 is closed so that c/> belongs to .91. This proves compactness of .91. The following books are standard references on analysis and functional analysis.
Dunford, N., and J. T. Schwartz, Linear Operators, Part I: General Theory, Interscience, New York, 1964. Graves, L. M., The Theory of Functions of Real Variables, 2nd Edition, McGraw-Hill, New York, 1956. Hurewicz, W., and H. Walman, Dimension Theory, Princeton University Press, Princeton, N.J., 1941. Liusternik, L. A., and V. J. Sobolev, Elements of Functional Analysis, Ungar, New York, 1965. Rudin, W., Real and Oomplex Analysis, McGraw-Hill, New York, 1966. Yoshida, K., Functional Analysis, Springer-Verlag, Berlin, 1965.
CHAPTER I General Properties of Differential Equations
The· purpose of this chapter is to discuss those properties of differential equations which are not dependent upon the specific form of the vector field. The basic existence theorem of Section 1. shows that a differential equation does define a family of functions and Sections 2 and 3 discuss the dependence of this family upon the initial values and paramet.ers. Section 4 contrasts the concept of stability with the concept of continuous dependence upon initial values. Section 5 is concerned with differential equations with vector fields that are only Lebesgue integrable in t. Section 6 is devoted to different.ial inequalities and their application to the problem of obtaining upper and lower bounds for solutions of differential equations. Sections 7 and 8 deal with some properties of the solution of differential equations which are characteristic of the fact that the vector field is independent of time; namely, the existence of cylinders of orbits near regular points and the concepts of invariant and minimal sets.
1.1. Existence Let t be a real scalar; let D be an open set in Rn+l with an element of D written as (t, x); let f: D -+ Rn, be continuous and let x = dx/dt. A differential equation is a relation of the form
x(t) =f(t, x(t))
(1.1)
or, briefly x =f(t, x).
We say x is alsolutionlof (1.1) on an interval 1 c R if x is a continuously differentiable function defined on 1, (t, x(t)) ED, t E 1 and x satisfies (1.1) on 1. We refer to.f as a vector field on D. Example 1:1.
Let D =
R2,
f(t, x) = x 2 . The function c/>(t)
c an arbitrary real number, c of=. 0, is a solution of c > O;t E ( - 00, -c) if c < 0. (See Fig. 1.1). 12
x=
x 2 for t
E
1
---, t+c
(-c, (0) if
GENERAL PROPERTIES OF DIFFERENTIAL EQUA'l'IONS
13
I
I I I t= -c>O I I I
I I I
I I I I I
I
I
I I I
I
I I I t=-c
Figure 1.1.1
Example 1.2. Let D = R2, f(t, x) = ./;; for x 5?; 0, = 0 for x < O. The function cp(t) = (t - c)2j4, t 5?; c, is a solution of x = .j;, x 5?; O. Notice x = 0 is also a solution (see Fig. 1.2). Suppose (to, xo) E D is given. Anlinitial value problem for equation (1.1)1 consists of finding an interval I containing to and a solution x of (1.1)
•
Figure 1.1.2
14
ORDINARY DIFFERENTIAL EQUATIONS
satisfying x(to) =
Xo .
(1.2)
We write this problem symbolically as [
:i = ht, x), x(to) = Xo, tEl.)
If there exists an interval 1 containing to and an x satisfying (1.2), we refer to this as a solu.tion of (1.1) passing through (to, xo). For the initial value problem, i =x 2 , x(o) = -c, c real, Example 1.1 shows the interval.! may depend upon c and may not be the whole real line. The initial value problem i = .J~, x ~ 0, x(o) = 0, has the solution x = 0 on ( - 00, 00). The function (t -C)2,
x(t)
=
{
t
4
~
~
c
0,
t ~ c.
0,
is also a solution. Therefore there need not be a unique solution of (1.2) for every continuous function f. Our first objective in this chapter is to discuss existence, uniqueness, continuation of solutions, and continuous dependence of solutions on initial data and parameters. First,notice that consideration of vector equations makes it unnecessary to consider nth order equations. fact, if y is a scalar, y(j'j denotes d l y/dti and
In
y(n)
=
F(t, y, y', ... , y(n-l»),
y(I)(tO)±XI+1.0,
then, letting x = (y, equivalent problem
y(l), ... ,
i=f(t,x),
j=0,1, ... ,n-1,
y(n-l»),
x(to)
f = (X2,
= Xo =
... , Xn, F), we obtain the
(XIO, .•• ,
xno).
Also, complex valued differential equations of the real variable tare included in the discussion of (1.1) since one can obtain a real system by taking the real and imaginary parts.
J LEMMA 1.1.1 (1.3)
Problem (1.2) is equivalent to x(t)
=
Xo
+f
t
f(T, X(T)) dT
to
providedf(t, x) is continuous. The proof of this lemma is obvious.
PROOF. Suppose ex, fl are positive numbers chosen so that the closed rectangle B(ex, fl, to ,xo) = B(ex, fl) = {(t, x): t E 1« ,I x - xol ~ fl}, 1 «= l«(to) =
GENERAL PROPERTIES OF DIFFERENTIAL EQUATIONS
15
it: It - tol ~ oc}, belongs to D. Let M = sup{11(t, x)l, (t, x) E B(oc. (3)}. Choose a, ~ so that 0 < a ~ oc, 0 < ~ ~ f3, Ma ~ ~ and define the set .91 = d(a, ~) of functions cp in C(?(la, Rn) which satisfy cp(to) = Xo, Icp(t) - xol ~ ~ for all t in Iii' From Chapter 0 the set .91 is convex, closed and bounded. For any cp in .91 define the function Tcp by the relation Tcp(t}.= Xo
+
f 1(s, t
dJ(8)) ds,
tEla,
to
From Lemma 1.1 finding fixed points of T is equivalent to solving the above initial value problem for (1.1). We now apply the Schauuer fixed point theorem to assert the existence of a fixed point of T in d. Obviously Tcp, cp in .91, is in c(?(la, Rn) and Tcp(to) = Xo. Also, for tEla,
ITcp(t) - xol ~ /(11(8, cp(s))1 dS/ ~ Mit -
tol
~Ma~~
since B(a,
mc: B(oc, (3). Thus T: .91 ITcp(t) -
Tcp(l)1
~/
J:
~d. Also
11(s, cp(s))1 dS/
~ M It -II
for all t, I in la. This implies the set T(d) is an equicontinuous family and tlhlrefore the closure of T(d) is compact. Finally, for any cp, 1"> in .91, it follows from the uniform continuity of 1(t, x) on B(oc, (3) that for any s > 0 there is a 0> 0 such that
ITcp(t) -
T1">(t) I ~
/(/ 1(s, cp(s)) -1 (s, 1">(8)) Ids I ~ sa,
for all t in la provided that Icp(s) - 1">(s) I ~ 0 for aIls in la . But this is precisely the statement that T is a continuous mapping; that is, for any s > 0, there is a 0 >0 such that ITcp - T1">1 ~ sa if Icp -1">1 ~ O. All of the conditions of the Schauder theorem are satisfied and we can assert the exist~nce of a fixed point of d. This completes the proof of the theorem. COROLLARY 1.1. If U is a compact set· of D, U c: V, an open set in D with the closure Vof V in D, then there is an oc > 0 such that, for any initial value (to, xo) E U, there is a solution of (1.1) through (to, xo) which exists at least on the interval to - oc ~ t ~ to oc.
+
PROOF. One merely repeats the proof of Theorem 1.1 with the further restriction on a, ~ to ensure that B(a, ~, to, xo) c: V for all (to, xo) E U.
ORDINARY DIFFERENTIAL EQUATIONS
16
Other proofs of the Peano theorem can be given without using the Schauder theorem. 'Ve illustrate the idea for one special construction, the Euler method of numerical analysis. This method consists of dividing the interval Ia ={t: It -tol ~ oc} into equal segments of length h and then on each of these small segments approximating the "solution" by a straight line. More specifically, for a given h, define the function cph on 1 a by
cph(t) =Xo + f(to, xo)(t - to), cph(t)
=
cph(to + h)
+ f(to + h, cph(to + h))(t -
to
~
to - h),
t
~
to
+ h,
to + h
~
t
~
to + 2h,
and so forth. One may have to choose oc small in order for (t, cph(t)) to be in D. Pictorially, we will have a polygonal function defined which for h small should approximate a solution of (1.1) if it exists. One now chooses a sequence {hk} such' that hk ~O as k ~ 00 and uses the Ascoli-Arzela theorem to show that a subsequence of the sequence cph k converges to a solution of (1.1). EXERCISE 1.1. Supply all details of the proof of the Peano theorem of existence using polygonal line segments. EXERCISE 1.2. State an implicit function theorem whose validity will be implied by the existence Theorem 1.1.
1.2. Continuation of Solutions
*
is a solution of a differential equation on an interval I, we say is a continuation of if is defined on an intervall which properly contains I, coincides with cp on I and satisfies the differential equation on 1. A solution cp is noncontinuable if no such continuation exists; that is, the interval I is the maximal interval 0 existence of the solution .
* *
*
LEMMA 2.1. If D is an open set in Rn+l, f: D ~ Rn is continuous and bounded on D, then any solution cp(t) of (1.1) defined on an interval (a, b) is such that cp(a + 0) and cp(b - 0) exist. If f(b, cp(b - 0)) is or can be defined so that f(t, x) is continuous at (b, cp(b - 0)), then q,(t) is a solution of (1.1) on (a, b]. The same remark applies to the left endpoint a. PROOF. to in (a, b),
We first show that the limits cp(a
cp(t) =:= cp(to) +
f f(s, cp(s)) ds,
+ 0), cp(b -
0) exist. For any
1
a
10
and, for a
~
t2
Icp(t2) - cp(tl)1 ~
f If (s, cp(s))1 ds ~ M(t2 to
h
il),
GEXERAL PROPERTIES OF. DIFFERENTIAL EQUATIONS
17
+
where M is a bound off(t, x) on D. Therefore cp(t2) - cp(t1) -+0 as tl, t2 -+a 0, which implies cp(a 0) exists. A similar argument shows that cp(b - 0) exists. The last conclusion of the lemma is obvious from the integral equation for cp.
+
THEOREM 2.1. If D is an open set in Rn+l, f: D -+ Rn is continuous and cp(t) is a solution of (1.1) on some interval, then there is a continuation of cp to a maximal interval of existence./ Furthermore, if (a, b) is a. maximal interval of existence of a solution x Qf (1.1), then (t, x(t)) tends to the boundary of D as t -+a and t -+b. P~OOF. Suppose x(t) is a solution of (1.1) on an interval I. If I is not a maximal interval of existence, then x can be extended to an interval properly containing I/Therefore: we may assume I is closed on one end, say to the right. We first show that x can be extended to a maximal right interval of existence and, therefor~, may assume that I = [a, b] and that x has no extension over [a, 00). The proof of the extension to the left is very similar. Suppose U is a compact set of D, U c: V, an open set in D with the closure li of V in D. From Corollary 1.1 for any initial value in U there is a solution of (1.1) existing over an interval of length ex depending only on U, V and the bound of f on V. Therefore, if x(t), a ~ t ~ b, belongs to U, then there is an extension of x to an interval [a, b <xl Since U is compact, one can continue this process a finite number of times to conclude there is an extension of x(t) to an interval [a, bu ] such that (b u , x(b u )) does not belong ~o U. . Now choose a .seq.uence V n of open sets in D such that U:'=1 V n = D, lin closed ,bounded, lin c: V n+1 for n = 1, 2, .... For each V n, there is a monotone increasing sequence {b n } constructed as above so that the solution x(t) of (1.1) -on [a, b] has an extension to the interval [a, bn] and (b n , x(b n )) is not in lin. Since the bn are bounded above, let w = limn .... oo bn . It is clear that x has been extended to the interval [a, w) and cannot be extended any further since the sequence (b k , x(b k )) is either unbounded or has a limit point on the boundary of the domain of definition of f. If w = 00, the last assertion in the theorem is trivial. Suppose w is finite,
+
U is a compact set of D and there is a sequence (tk,X(tk)),y in Rn,(w,y) in U, such that ti; -+ W, X(tk) -+ Y as k -+ 00. The fact that f is bounded in a neighborhood of (w,y) implies x is uniformly continuous on [a,w) and x(t)-+y as t -+ w-. Thus, there is an extension of x to the interval [a,w + 0;]. Since w + 0: > w, this is a contradiction and shows there is a tu such that (t,x(t)) is not in U for tu < t <w. Since U is ap. arbitrary compact set, this proves (t,x(t)) tends t? the boundary of D. The proof of the theorem is complete. EXERCISE 2.1. For t; x scalars, give an example of a function f(t, x) which is defined and continuous on an open bounded connected set D and
18
ORDINARY DIFFERENTIAL EQUATIONS
yet not every noncontinuable solution 4> of (1.1) defined on (a, b) has 4>(a + 0), 4>(b - 0) existing. The above continuation theorem can be used in specific examples to verify that a solution is defined on a large time interval. For example, if it is desired to.show that a solution is defined on an interval [to, (0), it is sufficient to proceed as follows. If the function !(t, x) is continuous for t in (t1, (0), h < to , Ixl < ex, and one can by some means ascertain that a (lertain solution x(t) must always satisfy Ix(t)1 ~ fJ < ex for all values of t ~ to for which x(t) is defined, then necessarily x(t) is defined on [to, (0). In fact, choose any T ~ to and ysuch that fJ < y < ex and define the rectangle D1 as D1 ={(t, x): to ~ t ~ T, Ixl ~ y}. Then !(t, x) is bounded on D1 and the continuation theorem implies that the solution x(t) can be continued to the boundary of D 1 . But y > fJ implies that x(t) must reach this' boundary by reaching the face of the rectangle defined by t = T. Therefore x(t) exists for to ~ t ~ T. Since T is arbitrary, this proves the assertion.
1.3. Uniqueness and Continuity Properties
A function !(t, x) defined on a domain D in Rn+1 is said to be\locallll U'P8chitzia1din x if for any closed bounded set U in D there is a k = ku such that I! (t, x) - ! (t, y)1 ~ k Ix - yl for (t, x), (t, y) in U. If! (t, x) has continuous first partial derivatives with respect to x in D, then! (t, x) is locally lipschitzian in x. If !(t, x) is continuous in a domain D, then the fundamental existence theorem implies the existence of at least one solution of (1.1) passing through a given point (to, :1:0) in D. Suppose, in addition, there is only one such solution x(t, to, xo) through a given (to, xo) in D. Forany (to, xo) E D, let (a(to, xo), b(to, xo)) be the maximal interval of existence of x(t, to, xo) and let E c: Rn+2 be defined by E
= {(t, to, xo): a(to, xo) < t < b(to, xo), (to ,xo) ED}.
The trajectory through (to, xo) is the set of points in Rn+1 given by (t, x(t, to ,xo)) for t varying over all possible values for which (t, to, xo) belongs to E. The set E is called the domain of definition ofx(t, to, xo). The basic existence and uniqueness theorem under the hypothesis that !(t, x) is locally lipschitzian in x is usually referred to as the Picard-LindelO! theorem. This result as well as additional information is containedin THEOREM 3.1. If!(t, x) is continuous in D and locally lipschitzian with respect to x in D, then for any (to, xo) in D, there exists a unique solution x(t, to, xo), x(to, to, xo) =xo, of (1.1) passing through (to, xo). Furthermore,
GENERAL PROPERTIES OF DIFFERENTIAL EQUATIONS
19
the domain E in Rn+2 of definition of the function x(t, to, xo) is open and x(t, to, xo) is continuous in E. PROOF. Define 1a = 1a(to) and B(ex, {J, to, xo) as in the proof of Theorem 1.1. For any given closed bounded subset U of D choose positive ex, {J so that B(ex, {J, to, xo) belongs to D for each (to, xo) in U and if
V =u{B(ex, {J, to, xo); (to, xo) in U}, then the closure of V is in D. Let M = sup{lf(t, x)I,(t, x) in V} and .let k be the lipschitz constant of f (t, x) with respect to x on V. Choose ii, p so that 0 < ii ~ ex, 0 < p ~ {J, Mii ~ /1, kii < 1 and let iF = {cp in ~(l.x(O), Rn): cp(O) = 0, Icp(t)1 ~ Pfor t in l.x(O)}. For any cp in iF, define a continuous function Tcp taking 112(0) into Rn by . (3.1)
Tcp(t)
=
f
tHo
j'(s, cp(s - to)
+ xo) ds, tin 112(0) .
to
By Lemma 1.1, the fixed points of T in iF coincide with the sOlutions x(t) = cp(t - to) + Xo of (1.1) which pass through (to, xo) and are such that (t, x(t)) is in B(ii, /1, to, xo). Obviously, Tcp(O) = 0 and it is easy to see that ITcp(t)r ~ Mii ~ /1 for all t in 1(i(0). Therefore, TiF c iF. Also, ITcp(t) - T
p,
20
ORDINARY DIFFERENTIAL EQUATIONS
is a continuous function of (t, g, 7)) for It - gl ~ ti, (g, 7)) in U. There exists an intege~ k such that to + kti > s ~ to + (k - 1 )ti. From uniqueness, we 4ave x(t + to + ti, to, xo) = x(t + to + ti, to + &, x(to + ti, to, xo)) for any t. But the previous remarks imply this function is continuous for It I ~ ti. Therefore, x(g, to, xo) is continuous for Ig -tol ~ 2&, (to, xo) in D as long as (g, to, xo) is in E. An obvious induction argument proves the continuity of x(s, to, xo) at s. The previous argument also implies E is open. This proves the theorem. THEOREM 3.2. If in addition to the hypotheses of Theorem 3.1 the function 1 = I(t, x, ,\) depends upon a parameter ,\ in a closed set G in Rk, is continuous for (t, x) in D, ,\ in G, and has a local lipschitz constant with respect to x independent of ,\ in G, then for each (to, xo) in D, ,\ in G, there is a unique solution x(t, to, Xo, ,\), x(to, to, Xo, ,\) = Xo, passing through (to, xo) and it is a continuous function of (t, to, Xo, ,\) in its domain of definition. PROOF. The proof is essentially the same as the proof of Theorem 3.1 except one chooses M = sup{ll(t, x, ,\)1: (t, x,,\) in V X G1} andk independent of ,\ to be a local lipschitz constant with respect to x on V X G1, where G1 is an arbitrary closed bounded set in G. Since the contraction principle was used in Theorems 3.1 and 3.2 to prove the 'local existence and uniqueness of the solution passing through (to, xo) the solution can be obtained by the method of successive approximations (3.2)
x(n+l) = Tx(n),
n=0,1,2, ... ,
+ f I(s, x(s)) ds, t
Tx(t) = Xo
It -tol ~ ti,
to
where x(O) is an arbitrary continuous function taking the interval It - tol ~ ti into Rn with Ix(O)(t) - xol ~ /3 for It - tol ~ ti. The constants ti, /3 were chosen so that Mti ~ /3 and k/3 < 1 where M and k were bounds on I(t, x) and its lipschitz constant with respect to x over a certain compact set. An obvious choice for x(O)(t) is the constant function Xo. Now, if one works directly with the successive approximations (3.2), one can prove the iterations converge and the equation (1.1) has a unique solution on the interval It - tol·~ ti under only 'the assumption Mti ~ /3 with the restriction kti < 1 being unnecessary. This makes it appear that the contraction principle is not as effective as successive approximations whereas we implied in Chapter 0 that this principle was introduced so as to replace successive approximations'. The discrepanc), is resolved by simply observing there are many e;~uivalent norms on the space of continuous functions satisfying Ix(t) - xol ~/3 for t in [to, to + ti]. In fact, with the constants as above, the norm 1·1 defined by
Ixl
=
sup {:e- 2k(t-to )lx(t)i}, to:;;;t:;;;to+Z!i
GENERAL PROPERTIES OF DIFFERENTIAL EQUATIONS
21
is equivalent to the supremum 'norm used earlier., In this norm, one shows (}asily that T is a contraction only'unoAf the assumption MiX ~ whic}> il'l the same result obtained by successive approximations. How do you resolve the above discrepancy on [to - iX, to]? Another remark on successive approximations is the fact that one can obtain convergence of the sequence defined by (3.2) for any initial x(O) satisfying appropriate bounds. The sequence obviously would converge more rapidly by a more clever choice of x(O)(t) depending upon 1 since choosing x(O)(t) as the solution itself would yield the, fixed point in one iteration. Effective use of such a procedure would require tremendous ingenuity.
p,
EXERCISE 3.1. Prove directly that the sequence of successive approximations (3.2) converges for M iX ~ where M is a bound for 1 on {(t, x): It - till ~ iX, Ix - xol ~ f1}. In the following, we need other results on the dependence of solutions on parameters and initial data.
P
THEOREM 3.3. If1(t, x, ,\) has continuous first derivatives with respect to x, ,\ for (t, x) ED, ,\ in an open set G c Rk, then the solution x(t, to, Xo, '\), x(to, to, Xo, ,\) = Xo, of (3.3)
i; =
1(t, x; ,\),
is continuously differentiable with respect to t, to, Xo, ,\ in its domain of definition. The matrix ox(t, to, Xo, '\)/0'\, ox(to, to, Xo, ,\)/0,\ = 0, satisfies the matrix differential equation
. 01 (t, x(t, to, Xo, ,\), ,\)
(3.4)
y=
01 (t, x(t, to, Xo, ,\), ,\) y+
ox
0'\
The matrix ox(t, to, Xo, '\)/oxo, ox(to, to, Xo, ,\)/oxo the linear variational equation, .
(3.5)
Y=
ol(t, x(t, ,to , Xo, ,\), ,\) 'ox
.
= I, the identity, satisfies
y.
Furthermore,
(3.6)
ox(t, to, Xo, ,\)
. ()x(t, to, Xo, ,\)
oto
oxo
---'----= -
1
\
(to, Xo, /\).
As in Theorem 3.2, we apply Theorem 3.2, Chapter O. If y = = Ty is defined by (3.1), then we must compute the derivatives A(y, cp), B(y, cp) of T ycp with respect to y, cp, respectively. One shows easily that A(y, cp) can be represented by the n X (n + k) matrix PROOF.
(xo,
M, and T
A(y,
cp)(t) =
[(0+1 C/(S, cp(s ~:o) +
Xo, ,\))
ds,
{+I (OI(S, cp(s ~~o)
+xo), ,\)
dS] ,
22
ORDINARY DIFFER·ENTIAL EQUATIONS
and the differential B(y, cp)rp is given by [B(y, cp)rp](t)
=
f
to +! 01'(8 -1..(8 'J
''f'
-
to
t)
ox
0
+ x 0, A) rp(8) dB.
As in the proof of-Theorem 3.2, the constants M, k are chosen as before relative to the set V X G1 with G1 a closed bounded set in G. The proof of the differentiability of cp(t, to, xo, A) with respect to xo, A now proceeds exactly as iil the proof of Theorem 3.1. The function x(t, to, xo, A) = cp(t - to, to, xo, A) + Xo is obviously differentiable in (t, xo, A). Knowing the differentiability immediately implies relations (3.4) and (3.5). To obtain relation (3.6) observe that uniqueness of the solution implies x(t, to, xo, A) = x(t, to + h, x(to + h, to, xo, A), A) for any h sufficiently small since at to + h, these two functions satisfy the equation and take on the same values. Therefore, x(t,
to + h, xo, A) - x(t, to, xo, A) = x(t, to
+ h, xo, A) -
x(t, to
+ h, x(to + h, to, xo, A), A)
and this implies x(t, to, xo, A) is differentiable in to and relation (3.6). The proof of the theorem is complete. By repeated application of the above theorem, one can obtain higher order derivatives of x(t, to, xo, A) under appropriate hypotheses of f. EXERCISE 3.2. If f(t, x) has continuous partial derivatives with respect to x up through order k, show that the solution x(t, to, xo), x(to, to, xo) = Xo, of (1.1) has continuous derivatives of order k with respect to Xo. Find the differential equation for the jth derivatives with respect to Xu and observe that the Taylor series for x(t, to, y) in y in a neighborhood of Xo is obtained by solving only nonhomogeneous linear equations. EXERCISE 3.3. If f(t, x) is analytic in x in a neighborhood of Xo, show there is an interval around to such that the function x(t, to, xo) of Exercise 3.2 is analytic in a neighborhood of Xo . EXERCISE 3.4. If f(t, x, A) has continuous partial derivatives with to x, A up through order k, show that the solution x(t, to, Xo, A), x(to, to, Xo, A) = Xo , of (3.3) has continuous derivatives of order k with respect to Xo, A. Find the differential equations for the partial derivatives with respect to A. Discuss the determination of the Taylor series in the y.eighborhood of some point. ;!
_
respec~
EXERCISE 3.5. As in Exercise 3.2, discuss the analyticity properties of Xo, A) in A when f (t, x, A) satisfies some analyticity conditions.
x(t, to,
GENERAL PROPERTIES OF DIFFERENTIAL EQUATIONS EXERCISE 3.6.
Show that the solutions of the equation
23
x=
(A(t)
+ AB(t))x where IB(t)l, IA(t)1 are continuous and bounded are entire functions of A. Can you generalize your result? EXERCISE 3.7. Suppose f (t, x, y) is continuous and has continuous first derivatives with respect to x, y for 0 ~ t ~ 1, X, yE (-00,00), and the boundary value problem X =f(t, x, x),
x(I) =b,
x(O) =a,
has a solution c/>(t), 0 ~ t ~ 1. If of (t, x, y)/ox > 0 for t E [0, 1] and all x, y. prove there is an -e > 0 such that the boundary value problem x =f(t, x, x),
x(O) =a,
IfJ - bl
has a solution for 0 ~ t ~ 1 and tf;(t, oc) of the initial value problem
and let tf;(t, oco) o ~ t ~ 1. If u(t)
t-, I
= c/>(t).
=
= fJ,
e. Hint: Consider the solution
x(O) =a,
x=f(t, x, x),
J
~
x(I)
x(O)
= oc,
For oc - oco sufficiently small, tf;(t, oc) exists for
otf;(t, oco)/ooc, then
U _ of(t, c/>(t), ¢(t))
u _ of(t, c/>(t), ¢(t)) u = 0,
oy
ox
with u(O) = 0, u(O) = 1. Show that u is monotone nondecreasing and, thus, utI) > O. Use the implicit function theorem to solve tf;(I, oc) - fJ = 0 for oc as a function of fJ. Equation (3.5) is called the linear variational equation for t,he following reason. If x(t) = z(t) x(t, to, Xo, A) is a solution of (3.3), then
+
..
itt)
+ ox(t, to,ot Xo, A) =
I(t, z(t)
+ x(t, to, xo, A), A).
This implies itt)
= f(t, z(t) + x(t, to,
= "of (t, x(t, to, Xo,
xo, A), A) - f(t, x(t, to, xo, A), A)
A), A) z(t)
+ a( Iz(t)) D
ox
as Iz(t)l approaches zero. Equation (3.5) therefore representsthefirstapproximation to the variation z(t) of the true solution of (3.2) from a given solution x(t, to, xo, A). The linear variational equation is obtained by neglecting the terms of order higher than the first in the equation for z. The conclusions of Theorems 3.1 and 3.2 concerning the continuity properties of x(t, to, xo, A) are valid under much weaker hypotheses than
24
ORDINARY DIFFERENTIAL EQUATIONS
stated. In fact, if \ve assu~e the uniqueness of the solution x(t, to, xo) (which is implied when you have a local lipschitz condit.ion in x) then one can prove directly that it must be continuous in (t, to, xo). We need one such result in the study of differential inequalities which we now formulate. LEMMA 3.1. Suppose {fn}, n = 1, 2, ... , is a sequence of functions defined and continuous on an open set Din Rn+l with limn_a:> fn =fo uniformlyon compact subsets of D. Suppose (tn, x n ) is a sequence of points in D converging to (to, xo) in D as n ~ 00 and let cpn(t), n = 0, 1, 2, ... , be. a solution of the equation x =fn(t, x) passing through the point (tn, xn). If cpo(t) is defined on [a, b] and if'! unique, then there is an integer no such that each cpn(t), n ~ no, can be defined on [a, b] and· converges to cpo(t) uniformly on [a, b]. PROOF. Let U be a compact subset of D which contains in its interior the set {(t, cpo(t), a ~ t ~ b} and suppose Ifol < M on U. Then necessarily Ifni < M on U if n ~ no where no is sufficiently large. Choosing ci, $ as in the proof of the basic existence theorem such that M ci ~ $, one can be assured for n sufficiently large that all the cpn(t) are defined on [tn, tn + ci]. If [8, y] = nn [tn, tn + ci], then the fact that tn ~to implies 8 < y. Also, all the CPn for n sufficiently large are defined on [8, y]. The sequence {cpn} is uniformly bounded and equicontinuous since Ifni < M. Therefore, there. exists a subsequence which we label again as cpn which converges uniformly to a function q; on [8, y]. Using the integral equation, we see that q; is cpo. Since every convergent subsequence of cpn on [8, y] must also converge to cpo (by uniqueness of cpo) it follows that cpn converges to cpo on [8, y]. Due to the compactness of the set {t, cp(t)}, t in [a, b], one completes the proof by successively stepping intervals oflength y - 8 until [a, b] is covered. THEOREM 3.4. Suppose f (t, x, A) is a continuous function of (t, x, A) for (t, x) in an open set D and A in a neighborhood of AO in Rp. If x(t, to, Xo, Ao), x(to, to, Xo, Ao) = Xo, is a solution of (3.3) on [a, b] and is unique, then there is a solution x(t, 8, 7), A), X(8, 8, 7), A) = 7), of (3.3) which is defined on [a, b] for all 8, 7), A sufficiently near to, Xo, Ao and is a continuous function of (t, 8, 7), A) at (t, to, xo, Ao). PROOF. Lemma 3.1 implies that x(t, 8, 7), A) is a continuous function of A at to, xo, Ao uniformly with respect to t in [a, b]. Thus, for any e > 0 there is a 81 :> 0 such that 8,7),
Ix(t, if 1(8,
7),
8, 7),
e A) - x(t, to, Xo, Ao)1 < 2"
A) - (to, xo, Ao) I < 81 . Since x(t, to, Xo, Ao) is a continuous function
GENERAL PROPERTIES OF DIFFERENTIAL EQUATIONS
25
of t for t in [a, b), there is a 8 2 such that
Ix(t, to, xo, '\0) -X(T, to, xo, '\0)1
8
<"2
if It - TI < 82 . Let 8 = min(81, (2). Then
Ix(t, s, ~
T), ,\) -
Ix(t, s,
X(T, to, Xo, '\o}l T), ,\). -
x(t, to, xo, '\0)1
+ Ix(t, to, xo, '\0) ~ X(T, to, xo, '\0)1
<8,
provided that I(s, T), ,\) - (to, xo, '\o}l + It - TI < 8. This proves Theorem ~.4. Under appropriate hypotheses, the solutions of differential equations define homeomorphisms of subsets of Rn into Rn. In fact, let us assume for any (to, xo) in D, the equation (1.1) has a unique solution x(t, to, xo) which is jointly continuous in (t, to', xo) in its domain of definition. If x(t, to, xo) exists on an interval [a, b), then there is a neighborhood U(xo) of Xo for which the solution x(t, to, Xl) exists for t in [a, b) and Xl in U(xo). For fixed to and each t in [a, b), the function x(t, t'{), .) can be considered as a mapping T t of U(xo) into a neighborhood of x(t, to, xo). The mapping T t is one to one and continuous by hypothesis and Tt-1x* = x(to, t, x*). Therefore the inverse function is also continuous which implies T t is a hoineomorphism. We are now in a position to define a general solution of (1.1). Let U be an open connected set in Rn and cp: R X U ~ Rn be such that: (a) for any c in U, cp(t, c) is a solution of (l.l); (b) for any to in R, the mapping cp(to, .): U ~ Rn is a homeomorphism on its range. Such a function cp will be referred to as a (local) general solution of (1.1).
1.4. Continuous Dependence and Stability Consider the equation i; =f(t, x) with f(t, 0) = 0 for all t in (-00, 00) and the function f(t, x) defined for all (t, x) in Rn+l. For any (to, xo) in Rn+l, we suppose· this equation has a unique solution x(t, to, xo), x(to, to, xo) = xo, which is jointly continuous in (t, to, xo) in its region of definition. Since f (t, 0) = 0 for all t, it follows that x = 0 is a solution on ( - 00, 00). The hypothesis of continuous dependence implies. that given any real numbers to in ( - 00, 00), T ~ 0, there exists a 8( T) > 0 such that the solution x(t, to, xo), Ixol ~ 8(T) exists on [to, to + T). Furthermore, Ix(t, to, xo)1 ~O uniformly in t on [to, to + T) as Ixol ~O. In other words, given any T > o and anY.8 > 0, there is a 8 = 8(8, T, to) such 'that the solution x(t, to, xo) exists· on 1= [to, to T), Ix(t, to, xo)1 < 8 on I provided that Ixol < 8(8, T, to) (see the accompanying figure).
+
26
ORDINARY DIFFERENTIAL EQUATIONS %
space
I
/
/
/
-----t
(to +T,%(to + T, to, %0»)
I
I I
I t = to
t=
to
+T
Figure 1.4.2
Let us discuss in detail the meaning of continuous dependence on initial data for the equation x = x 2 . Choose to = 0 and let Xo = c. A solution x(t, to, xo) = x(t, 0, c) of this equation is x(t, 0, c) == -c/(ct -1). Uniqueness implies this is the solution with x(O) = c. It is clear that x(t, 0, c) is continuous in t, c in'the domain of definition of x(t, 0, c). Since x(t, 0, 0) = 0, this implies for any T > 0 and any 8 > 0, there is a 8 = 8(8, T) > 0 so that Ix(t, 0, c)1 < 8 if Icl < 8. The largest value of 8 is 8/(1 - 8T) = 8. As T increases, 8 must decrease and, in fact, 8 must approach zero as T --+ 00. This implies t.hA continuity of x(t, 0, c) in c is not uniform with rA!'Inect to t in the infinite interval 10, 00). For this equation, it is even true that no solution with x(O) = c > 0 exists on [0, 00). Even though continuous dependence on parameters and initial data is important, it gives information only on finite intervals of time as the ~bove remarks show.JAn even more important, p.onCAnt. i" (>()nt,inll()l1!'1 rlependence on initial data on infinite int.ervals of time. This is the concept oflstabilitv.J In the remainder of this sec1;ion we assume f smooth enough to ensure existence, uniqueness and continuous dependence on parameters.
x
Definitions. Suppose f: [0, 00) X Rn --+ Rn. Consider = f (t, x), f(t, 0) ==0, t E [0, 00). The solution x = 0 is called UG;punov 8tahZe if for any 8> 0 and any to ~ 0, there is a 8 = 8(8, to) such that Ixol < 8 implies Ix(t, to, xo) I < 8 for t E [to, 00). The solution x = 0 is uniformly 8table if it is stable and 8 can be chosen independent of to ~ The solution x = 0 is called a.~lJm'TJtoticall1J stahle if it is stable and there exists a b = b(to) such that Ixol < b implies· Ix(t, to, xo) 1--+0 as t --+00. The solution x = 0 is uniformlu_ asymptotically stable if it is uniformly stable, b in the definition of l'j.s~mptotic stability can be chosen independent of to ~ 0, and for every "I > 0 there is a
o.
T(TJ) >0 such that Ixol
-GENERAL PROPERTIES OF DIFFERENTIAL EQUATIONS
27
Pictorially, stability is the same as in the above diagram except the solution must remain in the infinite cylinder of radius e for t ~ to. We can discuss the stability and asymptotic stability of any other solution x(t) of the equation by replacing x by x y and discussing the zero solution of the equationy = !(t, x y) - !(t, x). The definitions of stability of an arbitraJ;y solution x(t) are the same as above except with x replaced by x -x(t).
+
+
LEMMA 4.1. If f is either independent of t or periodie in t, then the solution x =0 of (1.1) being stable (asymptotically stable) implies the solution x = 0 of (1.1) is uniforml stable (uniforml toticall stable. EXERCISE
4.1.
Prove Lemma 4.l.
EXERCISE 4.2. Discuss the stability and asymptotic stability of every solution of the equations x = -x(1 - x), x x = 0, and x + 2- l [x 2 (x4 4x2)'/2]X = O. The latter equation has the family of solutions x = c sin(ct d) where c, d are arbitrary constants. Does stability defined in the above way depend on to in the sense that a solution x = 0 may be stable at one value of to and not at another? The answer is no! For tl ~ to, this follows immediately from continuity with respect to initial data. In fact, stability of x = 0 implies the existence of a 8(to, e) > 0 such that Ixol ~ 8(to, e) implies Ix(t, to, xo) I < e, t ~ to. Continuity with respect to initial data implies the existence of a 81 = 8 l (tl, e, to, 8) > 0 so small that IXII < 8l (h, e) implies Ix(t, tl, Xl) I < 8(to, e), tl ~ t~ to. Then Ix(t, tl, Xl) I < e for t ~ tl, provided that IXII ~ 8l (tl, e); that is, stability at tl' For tl ~ to, it is not quite so obvious. Let V(h, e) ={x in Rn: x = X(tl, to, xo) for Xo in the open ball of radius 8(to, e) centered at zero}. Since the mapping X(tl, to, .) is a homeomorphism, there exists a 8 l (tl, e) such that {x: Ixl ~ Sl(tl, c V(tl, e). With this 8l (h, e) we have Ix(t, h, Xl) I < e for t ~ tl and IXII ~ 8(tl, e); -that is, stability at tl.
+
+
+
+
en
EXERCISE 4.3. In the above definition of asymptotic stability of the solution x = 0, we have supposed that x = 0 is stable and solutions with initial- values in'a neighborhood of zero approach zero as t ~ 00 .. Is it possible to have the latter property and also have the solution x = 0 unstable? Show this cannot happen if x is a scalar. Give an example in two dimensions where all solutions approach zero as t ~ 00 and yet the solution x = 0 is unstable. Is it possible to give such an example in two dimensions for an equation whose right hand sides are independent of t? It is not appropriate at this time to have a detailed discussion of stability, but we will continually bring out more of the properties of this concept.
28
ORDINARY DIFFERENTIAL EQUATIONS
1.5. Extension of the Concept of a Differential Equation In Section 1.1, a differential equation was define.d for continuous vector fields f. As an immediate consequence, the initial value problem for (1.1) is equivalent to the integral equation
+ f 1(s, x(s)) dB. I
(5.1)
x(t) = Xo
10
For 1 continuous, any solution of this equation automatically possesses a continuous first derivative. On the other hand, it is clear that (5.1) will be meaningful for a more general class of functions 1 if it is not required that x have a continuous first derivative. The purpose of this section is to make these notions precise for a class of functions f. Suppose D is an open set in Rn+! and f: D -+ Rn is not necessarily continuous. Our problem is to find an absolutely continuous function x defined on a real interval I such that (t, x(t)) ED for t. in I and (5.1)
x(t) =1(t, x(t))
for all t in I except on a set of Lebesgue measure zero. If such a function x and interval I exist, we say x is a solution 01 (5.1). A solution of (5.1) through (to, xo) is a solution x of (5.1) with x(to) = Xo. We will not repeat the phrase " except on a set of Lebesgue measure zero" since it will always be clear that this is understood. Suppose D is an open set in Rn+!. We say that 1: D-+Rn satisfies the Caratheodorv conditionslon D if 1 is measurable in t for each fixed x, continuous in x for each fi:X:ed t and for each compact set U of D, there is an integrable function mu(t) such that (5.2)
11(t, x)1
~ mu(t),
(t, x)
E
u.
For functions 1 which satisfy the Caratheodory conditions on a domain D, the conclusions of Sections 1 and 2 carryover without .change. If the function 1(t, x) is also locally Lipschitzian in x with a measurable Lipschitz function, then the uniqueness property of the solution remains valid. These results are stated below, but only the details of the proof of the existence theorem are given, since the other proofs are essentially the same. THEOREM .5.1. (Caratheodory). If D is an open set in RnH and 1 satisfies the Caratheodory conditions on D, then, for any (to, xo)i:tl D, there is a solution of (5.1) through (to, xo). c.
PROOF. Suppose (x, {J are positive numbers chosen so that the rectangle B((X, (J) = {(t, x): It - tol ~ (x, Ix - xol ~ {J} is in D. Let fIX = {t: It - tol ~ (X},
29
GENERAL PROPERTIES OF DIFFERENTIAL EQUATIONS
m
= mB(ex, (j),
p,
M(t)
=
fo m(8) ds. Choose
'ii,
p so that 0 < ii ~
(x,
.0 <
p~ {J,
IM(t)1 ~ t E fa.. Let d be the set offunctions cp in ~(fa., Rn) which satisfy cp(to) = xo, Icp(t) - xol ~ for all t in fa.. The set d is a closed, bounded, convex subset of ~(fa., Rn). For any cp in d, define the function Tcp by the relation
P
Tcp(t) = Xo
+ f.
t
I(s, cp(s)) ds,
t E fa..
to
The fixed points of T in d coincide with the solutions in d of (5.1). We now apply the Schauder theorem to prove the existence of a fixed point of Tin d. For any cp in d, the operator T is well defined since 1(8, cp(s)) is integrableforcpind. Also, Tcp(to) = xo, and Tcp(t) is continuous for t E fa.. Using (5.2), we have 1
Tcp(t)
-xol ~
It
I/(s, cp(s))1 ds
to
~ I( m(s) dsl = IM(t)1 ~ p, for all t in fa.. Therefore, T: d --+d. The operator T is continuous on d. In fact, if cpn E A, cpn --+cp in d, then I(t, x) continuous in x for each fixed t implies/(t, cpn(t)) --+ I(t, cp(t)) as n--+ 00 for each t in fa.. Since I/(t, cpn(t)) I ~ m(t), the Lebesgue dominated convergence theorem implies
f.
t
I(s, cpn(s)) ds--+
to
f.
t ·
I(s, cp(s)) ds,
. to
as n --+ 00 for each t in fa.. This proves the assertion. For any cp in d, 1Tcp(t)
- Tcp(r) 1 ~ IM(t) -M(r)1
for all t, r in fa.. Since M is continuous on fa. , it is uniformly continuous and" thus, the set T d is an equicontinuous set of ~(fa., Rn). It is also uniformly bounded, This proves T d is relatively compact and, thus, T is completely continuous. The Schauder fixed point theorem implies the existence of a fixed point in d and the theorem is proved. THEOREM 5.2, If D is an open set in Rn+l, 1 satisfies the Caratheodory conditions on D and cp is a solution of (5.1) on some interval, then there is a continuation of cp to a maximal interval of existence. Furthermore, if (a, b) is a maximal interval of existence of (5.1), then x(t) tends to the boundary of D as t --+ a and t --+ b.
30
ORDINARY DIFFERENTIAL EQUATIONS
PROOF. The proof is essentially the same as the proof of Theorem 2.1 and is left to the r!'l8.der. THEOREM 5.3. Suppose D is an open set in Rn+l,f satisfies the Caratheodory conditions on D and for each compact set U in D, there is -an integrable function ku(t) such that
If(t, x) - f(t, y)1 ~ ku(t)
(5.3)
Ix -
yl,
(t, x) E U,
(t, y) E U.
Then, for any (to, xo) in U, there exists a unique solution x(t, to, xo) of (5.1) passing through (to, xo). The domain E in Rn+2 of definition of the function x(t, to, xo) is open and x(t, to, xo) is continuous in E. PROOF. This proof is essentially the same as the proof of Theorem 3.1 and the details are left to the reader. One only needs to choose M(t) as in
the proof of Theorem 5.1, let K(t) =
p
0< ci ~ ex, 0 < ~~, IM(t)l ~ Any linear system (5.4)
i
Lt. kB(/X,ln(s) ds and choose iX, pso that
p, Kl
where A(t) is an n X n matrix, h(t) is an n-vector whose elements are integrable on every finite interval satisfies the Caratheodory.conditions and (5.3). Therefore, the initial value problem has a unique solution.
1.6. Differential Inequalities Let Dr denote the right hand derivative of a function. If w(t, u) is a scalar function of the scalars t, u in some open connected set n, we say a function v(t), a ~ t ~ b, is a solution of the differential inequality Drv(t) ~ w(t, v(t))
(6,1)
on [a, b) if v(t) is continuous on [a, b) and has aright hand derivative on [a, b) that satisfies (6.1). LEMMA 6.1. If x(t) is a continuously differentiable n-vector function on a ~ t ~ b, then Dr Ix(t)1 exists on a ~ t < b and IDr(lx(t)1>1 ~ li(t)1 ' a ~ t
For any two n-vectors x, u and 0 <
O·~
1, h >0, we haye
Ix+ Ohul-IOx+ Ohul ~ (1 - 0) Ixl.
Therefore,
Ix + Ohul-Ixl Oh
~
Ix + hul -Ixl h
;
GENERAL PROPERTIES OF DIFFERENTIAL EQUATIONS
31
that is, the difference quotient Ix+hul-Ixl
h
is a nondecreasing function of h. Furthermore, this difference quotient is bounded below by -lui. Consequently, .
hm
Ix+hul-Ixl
h
h->O+
exists. If x(t) is continuously differentiable for a ~ t.~ b, then this latter relation implies
r
+ hx(t)l -lx(t)1
Ix(t)
h
1m
h->O+
exists. Since l[Jx(t
+ h)l -lx(t)ll- [Jx(t) + hx(t)l -lx(t)IJ1 -Ix(t) + hx(t)IJ1 ~ Ix(t + h) - x(t) - hx(t)l = o(h)
= 1[lx(t+ h)l as
h~O+,
it follows that Dr(lx(t)i) exists and
. I I = hm
Ix(t)
Dr x(t)
+ hx(t)1 -lx(t)1 .. h
h->O+
It is clear that IDr(lx(t)i)1 ~ li(t)l and the lemma is proved. The same proof also shows that the conclusion of Lemma 6.1 is valid for absolutely continuous functions in the sense that Dr(lx(t)i) exists and satisfies IDr(lx(t)i)1 ~ li(t)l almost everywhere for a ~ t ~ b. Q c
THEOREM 6.1. Let w(t, u) be continuous on an open connected set R2 and such that the initial value problem for the scalar equation
(6.2)
u
=
w(t, u)
has a unique ·solution. If u(t) is a solution of (6.2) on a ~ t ~ b. and v(t) is a solution of (6.1) on a ~ t < b with v(a) ~ u(a), then v(t) ~ u(t) for a ~ t ~ b. PROOF. (6.3) for n
Consider the family of equations u=w(t, u)
+-n1
= 1,2, .... We now apply Lemma 3.1 to (6.3).
ORDINARY DIFFERENTIAL EQUATIONS
32
If un(t) designates the solution of (6.3) with Un (a) = u(a), then Lemma 3.1 implies there is an no such that un(t), n ~ no, is defined on [a, b] and un(t) -~ u(t) uniformly on [a, b]. We show v(t) ~ un(t), for n ~ no, a ~ t ~ b. If this is not so, then there exist t2 < t1 in (a, b) such that v(t) > un(t) on t2 < t ~ t1, V(t2) = U n(t2). Therefore, v(t) - V(t2) > un(t) - Un(t2), t2 < t ~ t1, which implies
which is a contradiction. Consequently, v(t) ~ un(t) for t in [a, b], n Sinc!'l un(t) -i>- u(t) uniformly on [a, b], this proves the theorem. COROLLARY 6.1.
A solution of Drv(t)
~
~ no.
0 on [a, b) is nonincreasing on
[a, b).
COROLLARY 6.2 .. Suppose w(t, u) and u(t) are as in Theorem 6.1. If x(t) is a continuous n-vector function with a continuous first derivative on [a, b] such that Ix(a)1 ~ uta), (t, Ix(t)i) E Q, a ~ t ~ b, and li(t)1 ~ w(t, Ix(t)i),
a ~ t ~ b,
then Ix(t)1 ~ u(t) on a ~ t ~ b. PROOF.
This is immediate from Lemma 6.1 and Theorem 6.1.
COROLLARY 6.3. Suppose w(t, u) satisfies the. conditions of Theorem 6.1 for a ~ t < b, u ~ 0, and let u(t) ~ 0 be a solution of (6.2) on a ~ t < b. If f: [a, b) X Rn -i>- Rn is continuous and I/(t, x)1 ~ w(t, lxi),
a ~t
< b,
x ERn,
then the solutions of i = I(t, x),
Ix(a)1 ~ uta),
exist on [a, b) and Ix(t)1 ~ u(t), t in [a, b). PROOF. From Corollary 6.2, Ix(t)1 ~ u(t) as long as x(t) exists. From Theorem 2.1, the solution x(t) can fail to exist on [a, b) only if there is a c, £1,< c < b, such that x(t) is defined on [a, c) and lim Ix(t)1 = 00 as t , c - O. On the other hand, this is impossible since Ix(t)1 ~ u(t), t E [a, c) and tim u(t) exists as t -i>- c - o.
GENERAL PROPERTIES OF DIFFERENTIAL EQUATIONS
33
Notice that Corollary 6.3 gives existence of the. solution on [a, b), as well as upper bounds on the solutions. Another simple application of Corollary 6.2 yields COROLLARY 6.4. Suppose D is an open connected set in Rn+l,f: D ~ Rn is continuous and f(t, x) is locally lipschitz ian in x. If K is any compact set in D and L is the lipschitz constant of f in K, then Ix(t, to, xo) - x(t, to, xI)1 ~
eL(t-to) Ixo
as long as the solutions x(t, to, xo), x(t, to, (t, x(t, to, xo», (t, x(t, to, Xl)) remain in K.
Xl)
- xII
of (1.1) are such that
PROOF. If z(t) = x(t, to, xo) - x(t, to, Xl), then Iz(t)1 ~ L Iz(t)1 as long as (t, x(t, to, xo», (t, x(t, to, Xl» remain in K and Corollary 6.2 gives the result. As a first illustration of the above results, we prove a result on existence in the large. Suppose f: (0(, (0) X Rn -'; Rn is continuous If(t, x)l ~ 4>(t)¢'(lxi),
(6.4)
where 4>(t) ~ 0 is continuous for all t > 0( and ¢'(u) is continuous for u ~ 0 and positive for all u > O. Suppose u(t, to, uo), to> 0(, is a solution of 'It = 4>(t)¢'(u), u(to) = Uo > 0 and this solution is unique for any Uo > O. If
f
(6.5)
CO du ¢'(u)
=
+00,
then the solution u(t, to, uo) exists for all t
f
u
dv
-:;:- =
u. 't'(v)
>
In fact, u satisfies the equation
0(.
It 4>(s) ds, t.
and if u did not exist for all t > 0(, then the continuation theorem implies there would exist a ,. and a sequence {tn },tn ~,. as n ~ 00 such that u(tn ) -'; 00 as n~ 00. But this is impossible since [<4>
Jto
<
00
and [CO dvl¢,(v)
Juo
=
+00.
For any given xo.* 0 in Rn, choose Uo = Ixol and for Xo = 0 choose· any Uo > O. From ..Corollary 6.3, it follows that any solution x(t, to, xo),
of i;
f
=f
x(to, to, xo)
= Xo , to >
0(.
(t, x) exists 'and satisfies Ix(t, to, xo)1 ~ u(t, to, uo) on [to, (0) provided
sati~fies
(6.4) and (6.5). As a special case suppose that f(t, x) = A(t)x + h(t) where A(t), h(.t) are continuous for all values of t. Then any solution of the linear equation
(6.6)
i;
=
A(t)x
+ h(t)
ORDINARY DIFFERENTIAL EQUATIONS
34
exists on (-00, 00). In fact, IA(t)x
+ h(t)1 ~ IA(t)llxl + Ih(t)1 ~ max{IA(t)1 ' Ih(t)}(lxl
+ 1)
= 4>(t)if(lxi),
if(u)
Since 4>(t) is continuous and
= u + 1.
FO dulif(u) = + 00, we have the desired result.
Another interesting special case is when If(t, x)1 ~ K Ixl for all t in (-00, 00) and x in Rn. For 4>(t) = 1 and if(u) = Ku, the above example shows that all solutions of x =f(t, x) exist on (-00, 00). For later reference, part of these results are summarized in THEOREM 6.2. Iff: (at, 00) X Rn -+ Rn is continuous, satisfies (6.4) and (6.5) and the solution of 'Ii = 4>(t)if(u), u(to) = Uo > 0, to > at exists and is unique in (at,.oo), then any solution of the equation (1.1) through (to, xo) exists on (at, 00). In particular, every solution of (6.6) exists on (-00, 00) provided that A(t), h(t) are continuous. As another illustration, we prove a simple result on stability. Suppose f: Rn+l -+ Rn and there exists a continuous function A(t), - 00 < t < 00, such that X· f(t, x) ~ -A(t)X . x where"·" denotes the scalar product of two vectors. Notice this implies f(t, 0) = O. If we let Ixl2 = x . x and suppose x is a solution of x = f (t, x) on an interval I containing to, then
!:..lxl2 =!:.. (x . x) = 2x . f(t, x) ~ dt dt If w(t, u)
-2X(t)lxI2.
= -2A(t)U and u is the solution of 'Ii = w(t, u), u(to) = Ix(to)1 2, then
u(t) = {exp( -2
rt A(S) ds)} Ix(to)1 2 exists for all t in ( Jto
00, .00). Therefore, Corol-
lary 6.2 and the continuation theorem implies that the solution x(t) not only exists on I but exists for all t in .( - 00, 00) and Ix(t)1
If A(t)
~
~ exp (- It: A(S) ds) Ix(to)l,
t
~ to·
0, then the solution x = 0 of x = f (t, x) is stable and actually uni-
formly stable. If A(t) ~ 0 and L7A(S) ds = asymptotically stable. EXERCISE 6.1.
If A(t)
+ 00,
then the solution x = 0 is
~. 0 and Jtoroo A(S) ds = + 00 for all to, is the's~lution
x = 0 of the previous discussion uniformly asymptotically stable? Discuss the case where A(t) is not of fixed sign.
GENERAL PROPERTIES OF DIFFERENTIAL EQUATIONS
35
EXERCISE 6.2. Suppose f: Rn+l -+ Rn is continuous and there exists a positive definite matrix B such that x . Bf(t, x) ~ -A(t)X' x for all t, x where A(t) is continuous for t in (-00,00). Prove that any solution of the equation i = f (t, x), x(to) = xo, exists on [to, 00) and give sufficient conditions for stability and asymptotic stability. (Hint: Find the derivative of the function V(x) = x . Bx along solutions and use the fact that there is a positive constant IL such that x . Bx ;:;; iLX • x for all x.)
r
EXERCISE 6.3.
Consider the equation i =f(t, x), If(t,
x)1
~ rp(t)
Ixl
for
all t, :c inR X R, rp(t) dt < 00. (a) Prove that every solution approaches a constant as t -+ 00. (b) If, in addition, If(t, x) - f(t, y)1 ~ rp(t) Ix - yl
for all x, y,
prove there is a one to one correspondence between the initial values and the limit values of the solution. (c) Does the above result imply anything for the equation
i = -x
+ a(t)x,
fro la(t)1 dt < oo?
(Hint: Consider the transformation x = e-ty.) (d) Does this imply anything about the system
j'" la(t)1 dt < 00, where Xl,
X2
are scalars?
EXERCISE 6.4. z
Consider the initial value problem
+ a(z, z)z + {3(z) = u(t),
z(O)
= g,
z(O)
= 7],
with a(z, w), {3{z) continuous together with their first partial derivatives for all z, w, u continuous and bounded on (-00, 00), a;:;; 0, z{3(z) ;:;; 0. Show there is one and only one solution to this problem and the solution can be defined on [0, 00). Hint: Write the equation' as a system by letting z = X, Z = y, define V(x, y)
= y2J2 +
f:
{3(s) ds and study the rate of change of
V(x(t), y(t)) along the solutions of the two dimensional system.
COROLLARY 6.5. Let w(t, u) satisfy the conditions of Theorem 6.1 and in addition be nondecreasing in u. If u(t) is the same function as in Theorem
36
ORDINARY D,IFFERENTIAL EQUATIONS
6.1· and v(t) is continuous and satisfies (6.6)
~ Va + f~ w(s, v(s»
v(t)
then v(t)
~
ds,
a ~ t ~ b,
Va
~ u(a),
u(t), a ~ t ~ b.
PROOF. Let V(tl be the right hand side of (6.6) so that v(t) ~ V(t). Then .V(t) = w(t, v(t» ~ w(t, V(t)), V(a) = Va ~ u(a). Theorem 6.1 i~plies V(t) ~ u(t) for a ~ t ~ b and this proves the corollary. COROLLARY 6.6. (Gronwall's inequality). If ce is a real constant, (3(t) ~ 0, and cp(t) are continuous real functions for a ~ t ~ b which satisfy cp(t)
~ ce + f~ (3(s)cp(s) ds,
a
~ t ~. b,
cp(t)
~
a
~ t ~ b.
then
PROOF. 'Ii
= (3(t)u,
( ex:(> f~ (3(s) ds )ce,
Va = ce, w(t, u) = {3(t)u. Then = (expf~ (3(s) ds )ce, which proves the
We apply Corollary 6.5 with
u(a)
= ce is given by
u(t)
corollary. Actually, Gronwall's inequality is easily proved by other techniques. In the applications to follow, we actually need a generalization of this inequality so we state and prove it without using Theorem 6.1. LEMMA 6.2. (Generalized Gronwall inequality). If cp, ce are real valued and continuous for a ~ t ~ b, (3(t) ~ 0 is integrable on [a, b] and cp(t)
~
ce(t)
+ f)(S)cp(s) ds,
a
~ t ~ b,
then cp(t)
PROOF.
~
ce(t)
+ f)(S)ce(S) (exp f:{3(U) dU) ds,
a
~ t ~ b.
Let R(t) = f~ (3(s)cp(s) ds. Then dR(t)
----a;t =
(3(t)cp(t) ~ (3(t)ce(t)
+ (3(t)R(t)
except for a set of measure zero. Thus, dR(t)
de -
!
(exp -
(3(t)R(t) ~ (3(t)ce(t),
S: (3(u) du )R(S) ~ (exp - s: (3(u) du )(3(S)ce(S),
GENERAL PROPERTIES OF DIFFERENTIAL EQUATIONS
37
except for a set of measure zero. Integrating from a to t, we obtain (exp -
J~ {1(u) d~ )R(t) ~ J~ (exp -
{{1(U) du ){1(S)CX(S) ds.
and thus R(t)
~ J~ (exp s: {3(u) du ){1(S)CX(S) ds.
This estimate proves the lemma. If 0: is continuous with its first derivative 0: ~ 0, then integrating by parts in Lemma 6.2 gives
~(t) ~ o:(a)exp
(s: (3) + (&(S) exp(
f>
)ds
~[exp S:f3~[o:(a) + (&(s)ds] ~ o:(t) exp
t
fa f3
since f3 ~ O. Gronwall's inequality is now a special case.
1.7. Autonomous Systems-Generalities If x( t) is a solution of (1.1) defined on an interval (a, b), we have previously introduced the concept of a trajectory associated with this solution as the set in Rn+l defined by Ua~t~b(t, x(t)). The path or orbit of a trajectory is the projection ofthe trajectory into Rn, the space of dependent variables in (i.l). The space of dependent variables is usually called the state space or phase space. The phase coordinates for a scalar nth order equation in x is the vector (;1:, xCI), x(2), ... , x(n-I»). System (1.1) is called autonomous if the vector field, that is, the function f in (1.1), is independent of t. In this section we consider some general properties of autonomous systems; namely, the differential system . (7.1)
x =f(x)
where f: il---+ Rn is continuous and Q is an open set in Rn. A basic property of autonomous systems is the following: if x(t) is a solution of (7.1) on an interval (a, b), then for any real number T, the function x(t - T) is a solution of (7.1) on the interval (a + T, b + T). This is clear since the differential equation remains unchanged by a translation of the independent variable. Thus, from a single solution of (7.1) one can define a one parameter family of solutions.
38
ORDINARY DIFFERENTIAL EQUATIONS
We shall henceforth assume that for any p in n, there is a unique solution cp(t, p) of (7.1) passing through p at t = 0. The function cp(t, p) is defined on an open set ~ c Rn+l and satisfies the properties: (i) cp(O,p) =P;
(ii) cp(t, p) is continuous in ~; (iii) cp(t + T, p) = cp(t, cp(T, p)) on ~ In fact, it follows from Theorem 3.4 that cp(t, p) is continuous. Relation (iii) holds since both functions satisfy the equation, are equal for t = and we have assumed uniqueness. From the above definition, the path or orbit y ~ y(p) through a fixed pEn is the set in Rn defined by y(p) = {x ERn: there exist (t, cp) E ~ with cp(t, p) = x}. It is clear that cp(t, p) and cp(t T, p) are different parametrizations of the same orbit y(p). There is a unique path y through a given p in Q. Indeed, paths through p are projections of all solutions of (7.1) which pass through any' of the points on the line (T, p), -00 < T < 00. But, cp(t T, p) is the unique solution of (7.1) passing through (T, p) and we have soon above these functions are all parametrizations of the same curve. Notice this last conclusion implies that no two aths can intersect.
°
+
+
dXI /I(x)
(7.2)
=
dX2 dXn h(x) = ... = fn(x)
if A is differentiable and the differential dx along A is parallel to f(x) when f (x) 0, and A is a point when f(x) = 0.
*
LEMMA 7.1. (7.1) through p.
The solution of (7.2) at any point p of
n
is the
~bit
of
e'
PROOF. If P = (PI, ... ,Pn) is a critical point, then y and A are both equal to p. If p is not a critical point, then one of the components of f say /I
GENERAL PROPERTIES OF DIFFERENTIAL EQUATIONS
39
is such that /I(p) -# o. Therefore, /l(x) -# 0 for x in a neighborhood U of p. In U system (7.2) is therefore equivalent to the ordinary differential' system dxa.
Ja.(x)
dxl
JI(X)
-=--,
oc
= 2, 3, ... , n.
From the existence Theorem l.1, these equations have a solution Xet:(XI) which exists for IXI - PII sufficiently small. We parametrize A in the following way. Consider the autonomous scalar equation dXI
-
dt
=/I(XI, X2(XI), ... , Xn(XI)).
This equation has a solution XI(t), XI(O) = PI, which exists for ItI small. One now easily shows that XI(t), Xa.(XI(t)), oc = 2, ... , n, is a solution of (7.1). Since the orbit of (7.1) through any point of n is a solution of (7.2), this proves A = y and the lemma. A homeomorphic image of a closed or open line segment is called an arc. A homeomorphic image of the circumference of a circle is called a Jordan curve. A path.y is said to be closed if it is a Jordan curve. LEMMA 7.2. A necessary and sufficient condition for a path of (7.1) to be closed is that it corresponds to a nonconstant periodic solution of (7.1). PROOF. If y is a closed path of (7.1) and P is a point of y, there is a T -# 0 such that cp(T, p) = cp(O, p) = p. By uniqueness of solutions cp(t + T, p) = cp(t, p) for all t which says cp(t, p) has period T. Conversely, suppose cp(t, p) = cp(t + T, p) for all t, cp(t, p) nonconstant, and T is the least period of cp(t, p); i.e. p -# cp(t, p), 0 < t < T. As t varies in [0, T), cp(t, p) describes a curve in Rn which is the homeomorphic image of the segment [0, T) with cp(O, p) = cp(T, p). On the other hand, the line segment [0, T] with 0 and T identified is homeomorphic to the unit circle. This completes the proof. EXERCISE 7.l. Suppose the autonomous· equation x =J(x) has a nonconstant periodic solution xO(t). Define stability of this solution. Can the stability ever be asymptotic? What is the strongest type of stability that you would expect in such a situation? We now give a few examples· to illustrate the above cOJlcepts.
Example 7.1. If x is a real scalar, x = x, then cp(t, p) = etp, the trajectory through p is the set (t, etp), - 00 < t < 00 and the path through p is the set {x >O} if P >0, {x =O} if p =0 and {x
-
'IJ
+ pe- t).
If x is a real scalar, x = -x(1 -- x), then cp(t, p) = The paths are the sets {x > I}, {x = I}, {O < x < I},
40
ORDINARY DIFFERENTIAL EQUATIONS
x
Figure 1.7.1
x=l
I I I I I
"0
I t = l n -1
"0-
Figure 1.7.2
= O}, {x < o}. See Fig. 7.2. The equilibrium point x x = 0 is asymptotically stable.
{x
= 1 is unstable and
Example 7.3. If y is a real soalar, then the. equation ii + y = 0 is equivalent to the system Xl = X2, X2 = -Xl where Xl = y. The phase space is R2. For any real <'onstants a, b one verifies that cpI(t)·= a sin(t'+ b), cp2(t) = a cos(t + b) is a solution of this system and any solution of the equation can be written in this. form. Any trajectory lies on a circular cylinder and the path through any point is a circle passing through this point with center at the origin. See Fig. 7.3. An equilibrium point of a two dimensional autonomous system which has the property that every neighborhool of the point contains an orbit which is a closed curve (periodic solution) iJ!::called a . center. Thus, the solution Xl =r x2 = 0 of this example is a center. In this example, we did not need to integrate the equations to obtain the parametric representation of the orbits in phase space. In fact, Lemma 7.1 implies that the orbits are the solutions of the scalar equation
41
GENERAL PROPERTIES OF DIFFERENTIAL EQUATIONS
\
I
, ,
,
\
I
\
\
\
\
:
\ \
I I
\
\
\ \
/
_ \
\\ \
\
, ,,
--~----~-----r--Xl
\
,,
,,
\
-----\\---~t
..L-4--1.-;'--.--/
\
\
\
\
\
\
\
\
\
X2
Figure 1.7.3
+
which has the solutions xi X~ = constant; the constant of course being determined by the initial values. The manner in which the orbits are described with increasing time is easily obtained from the original equations. Suppose
Ex'ample 7.4.
e
> 0 is given,
xi + x~ ·and consider the system
Xl, X2
are real scalars,
r2
=
lh = -X2 + eXI(l - r2), X2
=
Xl
+ eX2(1 -
r2).
'Ihe phase space for this system is R2. If Xl = r cos (), system is equivalent to the system
0=1, r = er(l
X2
= r sin (), then the
- r 2).
One easily verifies that the trajectories and paths are as in Fig. 7.4. In this case, the paths are spirals outside and inside the circle of radius one and the equilibrium point (0, 0) together with the circle of radius one. The equilibrium
\'
I~'\
\
\
{ \\\ \\ 1\\
/ I /
I
\'
I
\' \ \
:
/
/
I
"
I
)...-LV--..... - L \ -I
,"
-
\1
,
\ ...\ , /
\
Figure 1.7.4
ORDINARY DIFFERENTIAL EQUATIONS
42
point (0, 0) is called a focus; that is, solutions in a neighborhood of it spiral toward it as t_+oo (or -00). To say a solution spirals toward zero as t_ +00 (or -(0) is to say that any ray emanating from zero is crossed by the orbit of the solution an infinite number of times and the solution approaches zero as t_+oo (or -00). The orbit (r=l) which is a closed curve in this example is called a limit cycle, the reason for the terminology becoming clear in later discussions. Notice that, for e =0 this example is the same as Example 7.3. However, for any e > 0, no matter how small, the phase portrait of the two equations are completely different. Example "1.5. In this example, we illustrate that the solutions of a differential equation may be much more complicated than any ofthe previous examples. A torus is the homeomorphic image of the cross product of two circles. Suppose c/> are the angles shown in Fig. 7.5 describing a coordinate
e,
Figure 1.7.5
e,
system on the torus, If c/> satisfy the differential equation (j = 1, ~ = w, where w is a constant, then a solution corresponding to an orbit 'Y goes around the torus traversing the angle I/> with period 21T/W and the angle e with period 21T. Therefore, for w irrational, the path is not closed, but it does have the property that the closure of'Y is the whole torus! EXERCISE 7.2. Prove the statement in Example 7.5 concerning the closure of 'Y for w irrational. Can you describe the behavior when w is rational? Example 7.G. In this example, we show in R3 that the com~licated behavior in Example 7.5 can be shared by all solutions of an equation in a solid torus. A solid torus is the homeomorphic image of the cross product
GENERAL PROPERTIES OF DIFFERENTIAL EQUATIONS
43
of a circle and a disk. Suppose the region is that depicted in Fig. 7.5 and the coordinate system (r, (), 4» is as shown with 0 ~ r ~ 2 and 0 ~ (), 4> ~ 21T. The value r = 0 is a circle C which lies in a plane P and the surface r = constant, 0 ~ 4>, () ~ 21T is a torus, which has C at its center. For the differential. equations, choose
¢=w, f =r(I-r),
where w is a constant. The torii r = 0 and r = 1 are invariant in the sense that any solution with initial value in these surfaces remain in them for all t in ( - 00, (0). Except for the periodic solution corresponding to C, all other solutions approach the torus described by r = 1. Therefore, from Example 7.5, the closure of every orbit except C contains the torus r = 1 if w is irrational. EXERCISE 7.3. order equation
Discuss the phase portrait of the solutions of the second
Xl X2
= =
X2(X~ - xi),
-XI(X~ - xi).
What are the orbits and the equilibrium points? Which equilibrium points are stable? What does the application of Lemma 7.1 yield for this system? Suppose rp: Sa. ---J- Rn is a given continuously differentiable function on the ball Sa. = {u in Rn-I, lui < ti} with rp(O) = P and rank [8rp(u)/8u] = n -1 for all u in Sa.. The set {x in Rn, x = rp(u), u in Sa.} is called a differentiable (n -I)-cell En-l through p. Such an En-l is said to be transverse to the path 'YP of (7.1) at p iffor each p' E E~-l, the path 'Yp' through p' is not tangent to En-l at p'. This is equivalent to saying that dx along 'Yp' at the point p' is not a linear combination of the columns of 8rp(u)/8u for lui < ex and this in tum is equivalent to saying that def [8rp(U) ] D(x, u) = det ~' f(x) #0, "
for u < ex. Suppose D(p, 0) # O. Since D(x, u) is continuous, there is an ex sufficiently small so that D(x, u) # 0 for Ix - pi < ex, lui < ex. For this ex, E~-l is transverse to 'YP at p. If p is a regular point of (7.1), there always exists an E~-l transverse to 'Y at p. A closed transversal through 'Y at p is defined in the same way using the closed ball of radius ex. An open path cylinder of (7.1) is a set which is. homeomorphic to an open cylinder (the cross product of an open ball in R n - l and an open interval)
44
ORDINARY DIFFERENTIAL EQUATIONS
and consists only of arcs of paths of (7.1). A clo8ecl poJh cylirulerof (7;1) is a set which is homeomorphic to a closed cylinder and consists only of arcs of paths of (7.1.) The ba,se8 of a closed path cylinder are the images under the homeomorphism of the bases of the closed cylinder. If a is a path cylinder (either open or closed) the arcs of paths of (7.1) lying in C will be called the generator8 010. LEMMA 7.3. Suppose 1 has continuous first partial derivatives in O. If P is a regular point of (7.1) and En-l is a differentiable (n -I)-cell "transve~e at p to the path y of (7.1) through p, then there is a path cylinder a containing p in its interior. In particular, every path y' of (7.1) with a point in a must cross En-l at a p'where En-l is transverse to y'.
Let En-l have a parametric representation given by En-l = u in Ba}. where Ba = {u in Rn-l;lul < oc}, and .p has a continuous first derivative. Let q,(t, p'), q,(0, p') =p', be the solution of (7.1) which describes the path y' through p'. There is an interval 1 containing zero in its interior such that each q,(t, p'), p' E En-l is defined for t E 1. The function q,(t, p') = q,(t, .p(u» can therefore be considered as a mapping T of 1 X Ba into Rn. From Theorem 3.3, this mapping is continuously differentiable in 1 X Ba. If we define PROOF.
{x in Rn: x
= .p(u),
.
F(x, t, u)
f
= -
x
+ .p(u) + f 1(q,(8, .p(u» o
dB
for x ERn, (t, u) E 1 X B a , then the fact that q,(t, .p(u» is a solution of (7.1) implies F(q,(t, .p(u», t, u) = 0 for (t, u) E 1 X Ba. We now consider the relation F(x, t, u) = 0 as implicitly defining t, u·as a function of x. Since det [OF(X, t, U)] o(t, u)
equals D(p, 0) for (x, t, u) = (p, 0, 0), this determinant must be different from zero in a neighborhood of (p, 0, 0). The implicit function theorem therefore implies that the inverse mapping of T exists and is continuously differentiable in a neighborhood of p. This shows there are oc 0, T > 0 such that the mapping T is a continuously differentiable homeomorphism (or diffeomorphism) of IT X Ba , IT = {t: It I < T}, into a neighborhood ofp. Furthermore, {T(t, u), tin I-r} coincides with the arc of y' described by the solution q,(t, p'), - T < t < T. The range of T is an open path cylinder. It is clear there is also a closed path cylinder. This proves the lemma. We now exte.nd this local result to a global result in the sense th,a~ the time interval involved in the description of the path may be arbitrarjbr large as long as it i& finite. First we prove a result for paths which are n~t' closed. More precisely, we prove
>-
GENERAL PROPERTIES OF DIFFERENTIAL EQUATIONS
45
LEMMA 7.4. Supposefhas continuous first partial derivatives in il, y is any path, p is a regular point, pq is an arc of y, E~-l, E~-l are differentiable (n -I)-cells transverl!le to y at p, q, respectively. Then there is a closed path cylinder whose axis is pq and whose bases are in E~-l, E~-l. We can assume that E~-l n ~-l is empty. Relative to Oq. Let tp be the time to traverse the path y from p to q; i.e. q,(tp ' p) = q. From continuity with respect to initial data, one can choose an open (n -I)-cell E~-l in E~-l such that q,(tp ' pi) belongs to Oq for every pi in E';,-l . . Lemma 7.3 implies that each point q,(t p , pi) must lie on an arc in Oq of a trajectory of (7.1) and must cross E~-l at a point q'. Let the time to traverse the arc 'Yp' from p' to q' be tp" The mapping p' -+ tp' is continnously differentiable and an application o( the implicit function theorem similar to the above implies the mapping p' -+ ¢(tp ' ,p') is a homeomorphism. Since f in (7.1) is bounded on Cq and Cp , there is a v > 0 such that the time to ,.,n-l leaye Cp along an arc of a path through pI E np as well as to leave Cq along an arc of a path through q' E E~-l is greater than v. Choose v < tp. Let us show that p'q' is a closed arc if the diameter of E~-l is sufficiently small. If p'q' is not an arc, then YP' must be a closed curve. Thus, there is a Tp" V < Tp' < tp' - v such that q,(Tp" p') = p'. If no such E~-l exists so that p'q' is an arc, then there is a sequence of pic E E~-l, Tp'., V< Tp'. < tp'.-v, p,,-o;..p as k-o;..oo such that q,(Tp'k' pIc) =p". The Tp'. must be bounded since tp'. -o;..tp as k -0;.. 00 and v < Tp'. < tp'. - v. Therefore, there is a subsequence which we label the same as before such that Tp'. -0;.. TO as k -0;.. 00 and 0 < TO < tp - v/2. But this clearly implies that the path yp described by q,(t, p) satisfies q,(TO, p) = p. This is a contradiction since pq was assumed to be an arc. The path cylinder 0 is obtained as the union ofthe arcs of the trajectories p'q' with pi in jJJ~-l. It remains only to show that this is homeomorphic to a closed cylinder. For 1=[0, 1], define the mapping G: jJJ~-l X 1-0;.. Rn by G(p', s) = ip(stp' , p'), where t~ is defined above. It is clear that this mapping is a homeomorphism and therefore 0 is a closed path cylinder. This proves the lemma. Now suppose 'Y is a closed-path. Lemma 7.2 implies 'Y is the orbit of a nonconstant periodic solution ¢(t,p) of (7.1) of least period l> 0. Take a h 'IS anoth er transversal E-pn - f at p, E-pn - 1 ~ Epn-l transversal E pn-l at p. T_ere such that, for any q E £;-1, there is a tq > 0, continuously differentiable in q,::qq,q) in EJ,-I,x(t,q) not in for < t < tq , and the mapping F: X [0,1) -+ R n defined by F(q,s) = x(stq ,q) is a diffeomorphism. The set F(Ep-l X [0,1)) is called a path ring enclosing 'Y. We have proved the following result. PROOF.
E~-l, E~-l, we can construct local open path cylinders 0P'
t; _
EJ,-l
°
46
ORDINARY DIFFERENTIAL EQUATIONS
LEMMA
7.5. If 'Y is a closed path, there is a path ring enclosing 'Y.
It may be that a solution of an autonomous equation is not defined for all t in R as the example i; = x 2 shows. In the applications, one is usually only interested in studying the behavior of the solutions in some bounded set G and it is very awkward to have to continually speak of the domain of definition of a solution. We can avoid this situation by replacing the original differential equation by another one for which all solutions are defined on (- 00, (0) and the paths defined by the solutions of the two coincide inside G. When the paths of two autonomous differential equations coincide on a set G, we say the differential equations are equivalent on G. LEMMA 7.6. If fin (7.1) is defined on Rn and G eRn is open and bounded, there exists a function g: Rn -+ Rn such that i; = g(x) is equivalent to (7.1) onG and the solutions of this latter equation are defined on (-00, (0). PROOF. If f = (iI, . , . , fn), we may suppose without loss of generality that G c {x: If,(x)1 ~ 1, j = 1, 2, ... , n}.Define g = (gJ, ... , gn) by gj = ff 4>" where each 4>j is defined by
1 1 4>j(x) =
fj(x)
1 h(x)
if Iff(X)1 ~ 1, if h(x) > 1, if fj(X) < -1.
Corollary 6.3 implies that g satisfies the conditions of the lemma since •• Ig(x)1 is bounded in Rn.
.
1.8. Autonomous Systems-Limit Sets, Invariant Sets In this section we consider system (7.1) and suppose f satisfies enough conditions on Rn to ensure that the solution cp(t, p), 4>(0, p) = p, is defined for all t in R and all pin Rn and satisfies the conditions (i)-(iii) listed at the beginning of Section 1.7. The orbit y(p) of (7.1) through p is defined by y(p) = {x: x = 4>(t, p), -00
(t, p), t ~ O} 3~nd the negative semiorbit through p is T(P) = {x: x = 4>(t, p), t ~ O}. If .~ do not wish to distinguish a particular point on an orbit, we will write y,y+, y- for the orbit, positive semiorbit, negative semiorbit, respectively. The positive or w-limit set of an orbit y of (7.1) is the set of points in
GENERAL PROPERTIES OF DIFFERENTIAL EQUATIONS
47
Rn which are approached along y with increasing time. More precisely, a point q belongs to the w-limit set or positive limit set w(y) of an orbit, y if there exists a sequence of real numbers {tk}, tk _ 00 as k _ 00 such that c/>(tk, p) -q as k _ 00. Similarly, a point q belongs to the a-limit set or negative limit set a(y) if there is a sequence of real numbers {tk}, tk _ - 00 as k_ 00 such that c/>(tk, p) _q as k_ 00. It is easy to prove that equivalent definitions of the w-limit set and a-limit set are
a(y)
=
n
y-(p)
pey
=
n
Te(-a:>. a:»
U c/>(t, p) !;ii;T
where the bar denotes closure.
I THEOREM 8.1.1 The a- and w-Iimit sets of an orbit y are closed and invariant. Furthermore, if y+(y-) is bounded, then the w-(cx-) limit set is nonempty compact and connected, dist(c/>(t, p), w(y(p))) _0 as t _ 00 and dist(c/>(t, p), cx(y(p))) _0 as t_ -00. PROOF. The closure is obvious from the definition. We now pr~ve the positive limit sets are invariant. If q is in w(y), there is a sequence {tn}, tn-oo as n_oo such that c/>(tn,p)-q as n_oo. Consequently, for any fixed t in (-00, (0), c/>(t + tn, p) = c/>(t, c/>(tn , p)) -c/>(t, q) as n - 00 from the continuity of c/>. This shows that the orbit through q belongs to w(y) or w(y) is invariant. A similar proof shows that a(y) is invariant. If y+ (y-) is bounded, then the w- (cx-) limit set is obviously nonempty and bounded. The closure therefore implies compactness. It is easy to see that dist(c/>(t, p), w(y(p))) _0 as.t_ 00, dist(c/>(t, p), cx(y(p))) _0 as t_ -00. This latter property clearly implies that w(y) and cx(y) are connected and the theorem is proved. COROLLARY 8.1. The limit sets of an orbit must contain only complete paths. A s~t M in Rn is called a minimal setof (7.1) if it is nonempty, closed and invariant and has no proper subset which possesses these three properties. LEMMA 8.1. If A is a nonempty compact, invariant set of (7.1), there is a minimal set M CA.
48
ORDINARY DIFFERENTIAL Eon ATIONS
PROOF. Let F be a family of nonempty subsets of Rn defined by F = {B: B c: A, B.compact, invariant}. For any BI, B2 in F, we say B2 < Bl if B2 c: B l . For any Fl c: F totally ordered by " < ", let C = nBeF1B. The family Fl has the finite intersection property. Indeed, if B l ; B2 are in Fl, then either Bl < B2 or B2 < Bl and, in either case, Bl n B2 if; nonempty and invariant or thus belongs to Fl. The same holds true for any finite collection of elements in Fl. Thus, C is not empty, compact a~? invariant and for each B in F 1 , C < B. Now suppose an element D of F is such that D < B for each B in Fl. Then D c: B for each Bin Fl which implies D c: C or D < C. Therefore C is the minimum of Fl. Since each totally ordered subfamily of F admits a minimum, it follows from Zorn's lemma that there is a minimal element· of F. It is easy to see that a minimal element is a minimal set of (7.1) and the proof is complete. Let us return to the examples considered in Section 1.7 to help clarify the above concepts. In example 7:1, the w-limit set of every orbit except the orbit consisting ofthe critical point {O} is empty. The a-limit set of every orbit is. {O}. The only minimal set is {O}. In example 7.2, th~ w-limit set of the orbits {O < x < I}, {x < O}, is {O}, the a-limit set of {x> I}, {O < x < I} is {O} and {O} and {I} are both minimal sets. In example 7.3, the w- and a-limit set of any orbit is itself, every orbit is a minimal set and any circular disk about the origin is invariant. In example 7.4, the circle {r = I} and the point {r = O} are minimal sets, the circle {r = I} is the w-limit set of every orbit except {r = O}, while the point {r = O} is the a-limit set of every orbit inside the unit circle. In example 7.6, the torus r = 1 is a minimal set as well as the circle r ~ 0, the w-limit set of every orbit except r = 0 is the torus r = 1 and the a-limit set of every orbit inside the torus r = 1 is the circle r = o. Let us give one other. artificial example to show that the w-limit sets do not always need to be minimal sets. Consider rand 8 as polar coordinates which satisfy the equations 1:1
=sin2 8
+ (1 -r)3,
r=r(l-r). The w-limit set of all orbits which do not lie on the sets {r = I} and {r = O} is the circle r = 1. The circle r = 1 is invariant but the orbits of the equation on r = 1 consist of the points {8 = O}, {8 = 7T} and the arcs of the circle {O < (J < 7T}, {7T < 8 < 27T}, The minimal sets on this circle are just the two points {8 = O}, {8 = 7T} .. EXERCISE 8.1: Give an example of a two dimensional system:fwhich has an orbit whose w-limit set is not empty and disconnected. " THEOREM 8.2. If K is a positively invariant set of system (7.1) and K is hom.eomorphic to the closed unit ball in Rn, there is at least one equilibrium point of system (7.1) in K.
GENERAL PR'OPERTIES OF DIFFERENTIAL EQUATIONS
49
PROOF. For any 'Tl > 0, consider the mapping.takingp in K into c/>('Tl, p) in K. From Brouwer's fixed point theorem, there is a PI in K such that c/>('T1, PI) =PL and, thus, a periodic orbit of (7.1) of period 'T1. Choose a sequence 'Tm > 0, 'Tm -+ 0 as m.-+ 00 and corresponding points Pm such that c/>('Tm, Pm) = Pm. We may assume this sequence converges to a p* in K as m -+ 00 since there is always a subsequence of the Pm which converge. For any t and any integer m, there is an integer km(t) such that km(t)'Tm ~ t < km(t)'Tm 'Tm and c/>(km(t)'Tm, Pm) = Pm for all t since c/>(t, Pm) is periodic of period 'Tm in t. Furthermore,
+
Ic/>(t, p*) - p*1 ~ Ic/>(t, p*) - c/>(t, Pm)1 = Ic/>(t, p*) - c/>(t, Pm)1
+ Ic/>(t, Pm) - Pml + IPm - p*1 + Ic/>(t - km(t)'Tm, Pm) - Pml
+IPm-p*l,
and the right hand side approaches zero as. m -+ 00 for all t. Therefore, P* is an equilibrium point of (7.1) and the theorem is proved. Some of the most basic problems in differential equations deal with the characterization of the minimal sets and the behavior of the solutions of the equations near minimal sets. Of course, one would also like to be able to describe the manner in which the w-limit set of any trajectory can be built up from minimal sets and orbits connecting the various minimal sets. In the case of t,wo dimensional systems, these questions have been satisfactorily answered. For higher dimensional systems, the minimal sets have not been completely classified and the local behavior of solutions has been thoroughly discussed only for minimal sets which are very simple. Our main goal in the following chapters is to discuss some approaches to these questions.
1.9. Remarks and Suggestions for Further Study For a detailed proof of Peano's theorem without using the Schauder theorem, see Coddington and Levinson [1], Hartman [1]. When uniqueness of trajectories of a differential equation is not assumed, the union of all trajectories through a given point forms a type of funnel. For a discussion of the'topological properties of such funnels, see Hartman [1]. There are many other ways to generalize the concept of a differential equation. For example, one could permit the vector field f(t, x) to be continuous in t, but discontinuous in x. Also, f (t, x) could be a set valued function. In spite of the fact that such equations are extremely important in some applications to control theory, they are not considered in this book. The interested reader may consult Flugge-Lotz [1], Andre and Seibert [1], Fillipov [1], Lee and Marcus [1]. The results on differential inequalities in Section 6 are valid in a much more general setting. In fact, one can use upper right hand derivatives in
50
ORDINARY DIFFERENTIAL EQUATIONS
place of right hand derivatives, the assumption of uniqueness can be eliminated by considering maximal solutions of the majorizing equation and even some types of vector inequalities can be used. Differential inequalities are also very useful for obtaining uniqueness theo~ms for vector fields which are not Lipschitzian. See Coppel [1], Hartman [I], Szarski [1], Laksmikantham and Leela [1]. Sections 7 and 8 belong to the geometric theory of differential equa.tions begun by Poincare [1] and advanced so much by the books. of Birkhoff [1], Lefschetz [1], Nemitskii and Stepanov [1]; Auslander and Gottschalk [1]. The presentation in Section 7 relies heavily upon the book of Lefschetz [1]. A function cf>: R X Rn into Rn which satisfies properties (i-iii) listed at the beginning of' Section 7 is called a dynamical system. Dynamical systems can and have been studied in great detail without any reference to differential equations (see Gottschalk and Hedlund [1], Nemitskii and Stepanov [1]). All results in Section 7 remain valid for dynamica:l systems. However, the proofs are more difficult since the implicit function theorem cannot be invoked. The concepts of Section 8 a~ essentially due to Birkhoff [1]. , The definitions of stability give~ in Section 4 are due to Uapunov [1]. For other types of stability see Cesari [1], Yoshizawa [2].
t\oY't\eo
WlOi'
r~c ~
rlAJO 0
bJ&f'.;tS ~..t f/omeO'1l4Of"f k'(;
''I
-{ ~,~ ~ be.. ~{of ~ 1M.-\0 ~ t!;-f~ hj
tA-.
r
~ +r~~~.-oUA) i y\ "g '('{~ ~ O'¥'\ q pi VI. 4" 12t>c)3k \j S f~~.J2-t'~t ct.! D\~ C!..() I-Vt. ~ "l c.L"d ~Je.clotJ til\. \Rt wk~~~t i~~S t+.st1.f·
CHAPTER II Two Dimensional Systems
The purpose of this chapter is to discuss the global behavior of solutions of differential equations in the plane and differential equations without critical points on a torus. In particular, in Section 1, the w-limit set of any bounded orbit in the plane is completely characterized, resulting in the famous Poincare-Bendixson theorem. Then this theorem is applied to obtain the existence and stability of limi.t cycles for some special types of equations. In Section 2, all possible w-limit sets of orbits of smootli differential equations without singular points on a torus are characterized, yielding the result that the w~liIl?-it set of an orbit is either a periodic orbit or the torus itself. Differential equations on the plane are by far the more important of the two types discussed since any system with one degree of freedom is described by such equations. On the other hand, in the restricted problem of three bodies in celestial mechanics, the interesting invariant sets are torii and, thus, the theory must be developed. Also, as will be seen in a later chapter, invariant torU arise in many other applications.
n.l.
Planar Two Dimensional Systems-The Poincare-Bendixson Theory In this section, we consider the two dimensional system
(1.1)
I
x =f(x)
where x is in R2, f: R2 _ R2 is continuous with its first partial derivatives and such that the solution c/>(t, p), c/>(O, p) = p, of (1.1) exists for -00 < t < 00. The solution c/>(t, p) is the unique solution through (0, p) and, therefore, is continuous for (t, p) in R3. For each fixed t, recall that the mapping c/>(t, .): R2 _ R2 is a homeomorphism. The beautiful results for 2-dimensional planar systems are made possible because of the Jordan curve theorem which is now stated without proof. Recall that a Jordan curve is the homeomorphic image of a circle.1
52
ORDINARY DIFFERENTIAL EQUATIONS
I JORDAN CURVE THEOREM .• Any Jordan curve J in R2 separates the plane; more precisely, R2\J = Se U S, where Se and S, are disjoint open sets, Se is unbounded and called the exterior of J, S, is bounded and called the interior of J and both sets are arcwise connected. A set B is arcwise connected if p, q in B implies there is an arc pq joining p and q which lies entirely in B. Let p be a regular point, L be a closed transversal containing p, Lo be its interior,
v=
> 0 with 0 < t < tp},
{p in Lo: there is a tp cp(t, p) in R2\L for
cp(tp, p) in Lo and
and let W = h-1(V) where h: [-1, 1] -?-L is a homeomorphism. Also, let g: W -?- (-1, 1) be defined by g(w) = h-1cp(th(W) ' h(w)). See Fig. 1.1. L
__________ .!!:.-l P = h(w) .__-- ___
- -..............
-----. h ........
'
........
"
'....... '-'-,- "- '\ • • • -1 w g(w)
•1
Figure II.I.I
LEMMA 1.1. The set W is open, g is continuous and increasing on W and the sequence {gk(w)}, k = 0, 1, ... , n ~ 00 is monotone, where gk(w) = g(gk-l(w)), k = 1,2, ... , gO(w) =w. PROOF. For any p in V c Lo let q = cp( tp , p) in Lo. From Section 1. 7, we have proved that the arc pq of the path through p can be enclosed in an open path cylinder with pq as axis and the bases of the cylinder lying in the interior Lo of the transversal L. This proves W is open. From continuity with respect to initial data, tp is continuous and we get continuity of g. To prove the last part of the lemma, consider the Jordan curve J given by C = {x: x= cp(t, p), 0 ~ t ~ t p } and the segment of L joining p and q. If P = q, then y(p) is a periodic orbit and the sequence {gk(W)} , w = h-1(p), consists of only one point. Thus, suppose for definiteness h-l(p) wo', Wo = h-l(p), W = h-1(q). Let Si, Se denote the inteflor and exterior of J, respectively. From the definition of a transversal L, p~ths can cross L in only one direction. Therefore, that part of Lo given by h[(g(w), 1)] must be completely in either S, or Se since otherwise an orbit would cross the
TWO DIMENSIONAL SYSTEMS
53
segment pq of Lo in a direction opposite to the direction the orbit through P crosses L. Therefore, if g2(w) is defined, it must belong to (g(w), 1) which by induction shows that the sequence above is monotone. Suppose WI > w, WI in W. If g(WI) is defined, then g(WI) > g(w) for the same reason as before. This completes the proof of the lemma. We used the differentiability of f(x) in the above proof when we proved the existence of a path cylinder in Section 1. 7. As remarked at the end of Chapter I, this assumption is unnecessary and one can prove the existence of a path cylinder only under the assumption of uniqueness of solutions. In all of the proofs that follow in this section, this is the only place differentiability of f(x) is used. In particular, the Poincare-Bendixson theorem below is valid without differentiability of f(x). COROLLARY 1.1. The w-limit set w(y) of an orbit y can intersect the interior Lo of a transversal L in only one point. If w(y) n Lo = Po, Y = y(p), then either w(y) = y, and y is a periodic orbit or there exists a sequence {tk}' tk ~ 00 as k ~ 00 such that I{)(tk,p) is in L o , cp(tk,p) ~ Po monotonically; that is, the sequence h- 1 (CP(tk'P» is monotone. PROOF. Suppose w(y) n Lo contains a point Po. From the definition of w-limit set, there is a sequence {tk}, tk ~ 00 as k ~ 00 such that c/>(tk, p) ~Po as k~ 00. But from Section 1.7, there must be a path cylinder containing Po such that any orbit passing sufficiently near Po must contain an arc which crosses the transversal L at some point. Therefore, there exist points qk=c/>(tk,P) in Lo,tk~oo as k~.oo such that qk~PO.as k~oo. But Lemma 1.1 implies that the qk approach Po monotonically in the sense that h- 1 (qk) is a monotone sequence. Suppose now Po is any other point in w( y) n Lo. Then the same argument holds to get a sequence ql., ~ Po monotonically. Lemma 1.1 then clearly imples that Po = Po and the corollary is proved. COROLLARY 1.2.
If y+ and w(y+) have a regular point in common, then
y+ is a periodic orbit.
PROOF. If Po in y+ n w(y+) is regular, there is a transversal of (1.1) containing Po in its interior. From Corollary 1.1, if w(y+) =j::. y+, there is a sequence qk = ~(tk' p) ~Po monotonically. Since Po is in y+, this contradicts Lemma 1.1. Corollary 1.1 therefore implies the result. THEOREM 1.1. If M is a bounded minimal set of (1.1), then M is either a critical point or a periodic orbit. PROOF. If Y is an orbit in M, then a.(y) and w(y) are not empty and belong to M. Since a.(y) and w(y) are closed and invariant we have oc(y) = w(y) = M. If M contains a critical point, then it must be the point itself
ORDINARY DIFFERENTIAL EQUATIONS
54
since, it is equal to we,,) for some ". If M = we,,) does not contain a critical point, then" c we,,) implies" and we,,) have a .regular point in common which implies by Corollary 1.2 that" is periodic. Therefore" = we,,) == M and this ·proves Theorem 1.1. LEMMA 1.2.
If w(,,+) contains regular points and also a periodic orbit
"0, then t.,I(,,+) =,,0· PROOF. If not, then the connectedness .of w(,,+) implies the existence of a sequence {p,,}, p" in w( ,,+)\"0 and a Po in "0 such that p,,-+ Po as n -+ co. Since Po is regular, there is a closed transvel'8&1 L such that Po is in the interior Lo of L. From Corollary 1.1, w(,,+) ('\ Lo = {po}. From the existence of a path cylinder in Section 1.7, there is neighborhood N of Po such that any orbit entering N must intersect Lo. In particular, ,,(p,,) for n sufficiently large must intersect Lo. But we know this occurs' at Po. Thus p" belongs to for 1t sufficiently large which is a contradiction.
"0
THEOREM 1.2 (Poincare-Bendixson Theorem). If,,+ is a bounded positive semiorbit and w(,,+) does not contain a critical point, then either (i)
,,+ = w(,,+),
or (ii) w(,,+) = y+\,,+.
In either case, the w-limit set is a periodic orbit. The same result is valid for a negative semiorbit. PROOF. By assumption and Theorem I.8.1, w(,,+) is nonempty; compact invariant and oontains regular points only. Therefore, by Lemma 1.8.1, there is_a bounded minimal set M in w(,,+) and M contains .only regular points. Theorem 1.1 implies M is a periodic orbit Lemma 1.2 now implies the theorem. An invariant set M of (1.1) is said to be stable if for every e-neighborhood Us of M there is as-neighborhood U6 of M such that pin U6 implies ,,+(p) in Us. M is said to be asymptotically stable if it is stable and in addition there is a b > 0 such that p in Ub implies dist(r/>(t, p), M) -+0 as t -+co. If M is a periodic orbit, one can also define stability from the inside and outside of M in an obvious manner.
"0.
"0
COROLLARY 1.3. For a periodic orbit to be asymptotically stable it is necessary and sufficient that there is a neighborhood G of such that w(,,(p» =,,0 for any p in G. .
"0
1
~"
PROOF. We first prove sufficiency. Clearly dist(r/>(t, p), ,,0) -+0 as t -+ co for every pin G. Suppose L is a transversal at Po in"o and suppose p is in
TWO DIMENSIONAL SYSTEMS
55
G ('\ Se, q is in G ('\ Si, where Se and Si are the exterior and interior of yo , respectively. From Corollary 1.1, there are sequences Pk = cp(t k, p), qk = cp(t", q) in L approaching Po as k ~ 00. Consider the neighborhood Uk of Yo which lies between the curves given by the arc P k P k+l of y(p) and the segment of L between Pk and PHI and the arc qkqk+l of y(p) and the segment of L between qk and qk+l· Uk is a neighborhood of Yo· The sequences {tk}' {tIc} satisfy tk+ 1 - tk ~ 0.:, t,,+ 1 - tIc -+ 0.: as k ~ 00 where 0.: is the period of Yo. This follows from the existence of a path ring around yo. Continuity with respect to initial data then implies for any given e-neighborhood U £ of yo, t here is a k sufficiently large so that P in Uk implies cp(t, p) in U £ for t ;;;; 0 and yo i$ stable. To prove the converse, suppose yo is asymptotically stable. Then there must exist a neighborhood G of yo which contains no equilibrium points and G\yo contains no periodic orbits. The Poincare-Bendixson theorem implies the w-limit set of every orbit is a periodic orbit. Since yo is the only such orbit in G, this proves the corollary.
COROLLARY 104. Suppose 1'1, 1'2 are two periodic orbits with 1'2 in the interior of 1'1 and no periodic orbits or critical points lie between 1'1 and 1'2 . Then both orbits cannot be asymptotically stable on the sides facing one another. PROOF. Suppose 1'1, 1'2 are stable on the sides facing one another. Then there exist positive orbits y~, 1'2 in the region between 1'1' 1'2 such that 1'1 = Y~\Yi, 1'2 = yz\yz· For any PI in 1'1' P2 in 1'2 construct transversals L 1, L 2 · There exist pi =f; pi in y~ ('\ L 1 , P2 =f; p~ in 1'2 ('\ L 2 • Consider the region S bounded by the Jordan curve consisting of the arc pi pi of yi and the segment of the transversal Ll between pi and pi and the curve consisting of the arc pzp; of y~ and the segment of the transversal L2 between P2 and P; (see Fig. 1.2). The region S contains a negative semiorbit. Thus, the PoincareBendixson Theorem implies the existence of a periodic orbit in this region. This contradiction proves the corollary. THEOREM 1.3. Let 1'+ be a positive semiorbit in a closed bounded subset K of R2 and suppose K has only a finite number of critical points. Then one of the following is satisfied: (i) w(y+) is a critical point; (ii) w(y+) is a ,Jeriodic orbit; (iii) w(y+) contains a finite number of critical points and a set of orbits Yi with o.:(Yi) and W(Yi) consisting of a critical point for each orbit Yf. See Fig. 1.3. PROOF. w(y+) contains at most a finite number of critical points. If w(y+) contains no regular points, then it must be just one point since it is
ORDINARY DIFFERENTIAL EQUATIONS
56
Figure II.l.2
(i)
(ii)
(iii)
Figure II.l.3
connected. This is case (i). Suppose w(y+) has regular points and also contains a periodic orbit yo. Tnen w(y+) = yo from Lemma 1.2. Now suppose w(y+) contains regular points and no periodic orbits. Let yo be an orbit in w(y+). Then w(yo) c w(y.J:). If Po in w(yo) is a regular point and L is a closed transversal to Po with interior L o , then Corollary 1.1 implies w(y+) II Lo = w(yo) II Lo = {Po} and yo must meet Lo at some qo. Since yo belongs to w(y+) we have qo = Po which implies by Corollary 1.2 that yo is periodic. This contradiction implies w(yo) has no regular points. But, w(yo) is connected and therefore consists of exactly" one point, a critical point. A similar argument applies to the a-limit sets and the theorem is proved. COROLLARY 1.5. If y+ is a positive semiorbit contained in a compact set in nand w(y+) contains regular points and exactly one critical P}ht Po, then there is an orbit in w(y+) whose a- and w-limit sets are {Po}. ~ . ;"' We now discuss the possible behavior of orbits in neighborhood of a periodic orbit. Let yo be a periodic orbit and Lo be a transversal at Po in yo ,
a
TWO DIMENSIONAL SYSTEMS
57
h: (-1, 1) -* Lo be a homeomorphism with h(O) = Po. If g is the function defined in Lemma l.1, then g(O) = 0 since yo is periodic. Since the domain W of definition of g is open, 0 is in W, g is continuous and increasing, there is an e > 0 such that g is defined and g(w) > 0 for w in (0, e) and g(w) < 0 for w in (- e, 0). We discuss in detail the case g(w) > 0 on (0, e) and the case g(w) < 0 on (- e, 0) is treated in a similar manner. Three possibilities present themselves. There is an el, 0 < el < e, such that
(i) (ii) (iii)
g(w) g(w) g(w)
< w for w in (0, ed; > w for w in (0, el); = w for a sequence Wn > 0,
Wn -*0 as n -* 00.
In case (i), glc(w) is defined for each k > 0, is monotone decreasing and glc(w)-*O as k-*oo. In fact, it is clear that glc(w) is defined for k>O. Lemma l.1 states that glc(w) is monotone and the hypothesis implies this sequence is decreasing. Therefore, glc(w) -*wo ~ 0 as k -* 00. But, this implies g(wo) = Wo and therefore Wo = O. Similarly, in case (ii), if we define g-lc(w) to be the inverse of glc(w) then g-lc(w), is defined for each k > 0, is decreasing and g-lc(w) -*0 as k -* 00. If we interpret these three cases in terms of orbits and limit sets, we have
THEOREM 1.4. If yo is a periodic orbit and G is an open set containing yo, Ge = G n Se, Gi = G n Si where Se and Si are the interior and exterior of yo, then one of the following situations occur: (i) there is a G such that either yo = w(y(p)) for every pin Ge or yo = Q((Y(p)) for every pin Ge ; (ii) for each G, there is a pin Ge , p not in yo, such that Q((Y(p)) = y(p) is a periodic orbit. Similar statements hold for Gt . We call 'Yo a It'mit cycle if there is a neighborhood G of 'Yo such that either w('Y(P)) = 'Yo for every pEG or a(-y(p) = 'Yo for every pEG. The Pomcare-Bendixson theorem suggests a way to determine the existence of a nonconstant periodic solution of an autbnomous differential equation in the plane. More spe0ifically, one attempts to construct a domain D in R2 which is equilibrium point free and is positively invariant; that is, any solution 0{(1.1) with initial value in D remains iri D for t ~ O. In such a case, we are assured that D contains a positive semiorbit y+ and thus a periodic solution from the Poincar~-Bendixson theorein~ Furthermore, if we can ascertain that there is only one periodic Ol;bit in D, it will be asymptotically stable from Theorem 1.4 and Corollary 1.3. These ideas are illustraten. for the Liimard type equation (1.2)
ii
+ g(u)u + u = 0
58
ORDINARY DIFFERENTIAL EQUATIONS
where g( u) is continuous and the following conditions are satisfied: (1.3)
f:
(a) G(u) = def 9(8) tis is odd in u, (b) G(u) _ 00 as lul- 00 and there is a fJ > 0 such that G(u) > 0 for u> fJ and is monotone increasing. (c) There is an 0: > 0 such that G(u) < 0 for 0 < u < a, G(a) = O. Equation 1.2 is equivalent to the system of equations
u =v -G(u), v=-u.
(1.4)
System (1.4) is a special case of system (1.1) with x = (u, v) and has a unique orbit through any point in R2 since G has a continuous first derivative. System (1.4) has only one critical point; namely, u = 0, v = 0, and the orbits of (1.4) are the solutions of the first order equation (1.5)
dv u -=----
du
v -G(u)'
From (1.4), the function u = u(t) is increasing for v > G(u), decreasing if v(t) is decreasing if u > 0, increasing if u < O. Also, the slopes of the paths v = v(u) described by (1.5) are horizontal on the v-axis and vertical on the curve v = G(u). These facts and hypothesis (1.3b) on G(u) imply that a solution of (1.4) with initial value A = (0, vol for Vo sufficiently large describes an orbit with an arc of the general shape shown in Fig. 1.4. v
< G(u) and the function v =
Figure 11.1.4
TWO
59
DIMENSIONAL SYSTEMS
Observe that (u, v).a solution of (1.4) implies (-u, -v) is also a solution from hypothesis (1.3a). Therefore, if we know a path ABECD exists as in Fig. 1.4, then the reflection of this path through the origin is another path. In particular, if A = (0, vo), D = (0, -VI), VI < vo, then the complete positive semiorbit of the path through any point A' = (0, v~), 0 0 sufficiently large so that a solution as in Fig. 1.4 exists with A = (0, vo), D = (0, -VI), VI < Vo. Consider the function V(u, v) = (u 2 + v2)j2. If u, V are solutions of (1.4) and (1.5), then (1.6)
dV
(a)
dt =
(b)
dV du
(c)
dV dv
-uG(u), uG(u) v -G(u)
=
G(u).
Using these expressions, we have V(D) - V(A)
=
I
dV ABEOD
=
(I AB
+ Jc ) V-u:(U) du + I G(u) dv (u) BEO OD
along the orbits of (1.4). It is clear that this first expression approaches zero monotonically as Vo -+ 00. If F is any point on the u-axis in Fig. 1.4 between
(~, 0) andE, and
cp(vo)
-cp(vo)=-I BEO
= IBEO G(u) dv, then
G(u)dv=C-
JOEB
G(u)dv>I
G(u)dv>FJxFK
EK
where F J, F K are the lengths of the line segments indicated in Fig. 1.4. For fixed F, F K -+ 00 as Vo -+ 00 and this proves cp(vo) -+ - 00 as Vo -+ 00. Thus, there is a Vo su,ch that V(D) < V(A). But this implies VI < Vo and the semiorbit through A must be bounded. On the other hand, this semiorbit must also be bounded away from the origin since (1.6a) and hypothesis (1.3c) implies that dVjdt ~ 0 along solutions of (1.4) if lui < IX. Finally, the PoincareBendixson Theorem implies the existence of a periodic solution of (1.4) and we have . THEOREM 1.5. If G satisfies the conditions (1.3), then equation (1.2) has a nonconstant periodic solution.
60
. ORDINARY DIFFERENTIAL EQUATIONS
v t-_....;;;B:........
V
=G(u)
----------~~--~~--~~--~~-----u
Figure II.l.5
If further hypotheses are made on G, then the above method of proof will yield the existence of exactly one nonconstant periodic solution. In fact, we can prove' THEOREM l.6. If G satisfies the conditions (l.3) with IX = p, then equation (l.2) has exactly one periodic orbit and it is asymptotically stable. PROOF. With the stronger hypotheses on (1, every solution with initial value A = (0, vo), Vo > 0, has an arc of an orbit as shown in Fig. 1.5. With the notations the same as in the proof of Theorem l.5 and with E = (uo, 0), we have
V(D) - V(A) =
f
G(u)
d~ > 0,
ABEOD
if Uo < IX. This implies' no periodic orbit can have Uo < IX. For Uo > IX, if we introduce new variables x = G(u), y = v to the right of line BO in Fig. l.5 (this is legitimate since G(u) is monotone increasing in this region), then the arc BEO goes into an arc B*E*O* with end points on the
f
f
y-axis and the second expression cp(vo) = G(u) dv = x dy is the BEO B*E"O· negatIve of the area bounded by the curve B* E*O* and the y-axis. Therefore, cp{vo) is a monotone decreasing function of Vo. It is easy to check that
JAB
+ . fBCG{u)du is decreasing in Vo
and so V{D) - YeA) is decreasing in Vo·
Also, in the proof of Theorem l.5,it was shown that V(D) - YeA) approaches - 0 0 as Vo ~ 00. Therefore, there is a unique Vo for which V{D) = V{4> and thus a unique nonconstant periodic solution. Theorem 1.4 and Coronary 1.3 imply the stability properties of the orbit and the proof is complete. An important special case of Theorem l.6 is the van der Pol equation (l.7)
it - k(1 - u 2 )u
+u =
0,
k
> 0.
61
TWO DIMENSIONAL SYSTEMf'
In the above crude analysis, we obtained very little information concerning the location of the unique limit cycle given in Theorem 1.6. When a differential equation contains a parameter, one can sometimes discuss the precise limiting behavior as the parameter tends to some value. This is illustrated with van der Pol's equation. (1.7). Suppose k is very large; more specifically, suppose k = 8- 1 and let us determine the behavior of the riodic solution as 8 ~O+. Oscillations of this type are called relaxation oscillations System (1.7) is equivalent to 8U =v -G(u),
(1.8)
V= -
8U,
where G(u) = u 3 /3 - u. From Theorem 1.6, equation (1.8) has a. unique asymptotically stable limit cycle r(8) for every 8> O. From (1.8), if 8 is small and the orbit is away from the curve v = G(u) in Fig. 1.6, then the u
v /
/
/
/
/
/
/
v=G(u)
/
,
----------~~------~r-------~------------u
/
/
/
/
/
"
/ /
/
/
Figure II.l.B
coordinate has a large velocity and the v coordinate is moving slowly. Therefore, the orbit"has a tendency to jump in horizontal directions except when it is very close to the curve v = G(u). These intuitive remarks are made precise in THEOREM l.7. As 8 ~O, the limit cycle of (l.8) approaches the Jordan curve J shown in Fig. l.6 consisting of arcs of the curve v = G(u) and horizontal line segments. To prove this, we construct a closed annular region U containing J such that dist( U, J) is any preassigned constant and yet for 8 sufficiently small, all paths cross the boundary of U inward. U will thus contain (from the Poincare-Bendixson theorem) the limit cycle r( 8). The construction of
62
ORDINARY
DIFFERENTI~L
EQUATIONS
U is shown in Fig. 1.7 where h is a positive constant. The straight lines 81 and 45 are tangent to v = G(u) + h, v = G(u) - h respectively and the lines 56, 12, 9-10, 13-14, are horizontal while 23, 67, 11-12, 15-16 are vertical. The remainder of the construction should be clear. The inner and outer v
------F,~------------~~----------
__
~~-----u
Figure II.l.7
boundaries are chosen to be symmetrical about the origin. Also marked on the figure are arrows designating the direction segments of the boundaries crossed. These are obtained directly from the differential equation and are independent of e > o. It is necessary to show that the other segments of the boundary are also crossed inward by orbits if e is small. By symmetry, it is only necessary to discuss 34, 45 and 10-11. At any point (u, G(u) - h) on 34, along the orbits of (1.8), we have dv
- e2u
du
v -G(u)
-=
eZu
e2u(3)
h
h
=-<--
where u(3) is the value of u at point 3. Hence for e small enough, this is less than g(4) < g(u) which is the slope of the curve G(u) - h. Thus, v < 0 on this arc implies the orbits enter the region along this arc. Along the arc 45, we have Iv - G(u)1 > h and, hence, the absolute value of the slope of the path Idvfdul = l-e 2uf[v - G(u)]1 < e2u(4)fh approaches zero as e _ O. For' e small enough this can be made < g( 4) which is t~e slope of the line 45. Thus, v< 0 on. this arc implies the orbits enter intd'Uif e is small enough.
TWO DIMENSIONAL SY.STEMS
63
LetK be the length ofthe arc 11-12. For K small enough, Iv -G(u)1 > K along the arc 1O-1l. Hence, Idvldul along orbits of (1.8) is less than e2'11-/K < e2u(1l)/K, which approaches zero as e ~o. Thus, for e small, the orbits enter U since 'Ii, > 0 on this arc. This shows that given a region U of the above type, one can always choose e small enough to ensure that the orbits cross the boundary of U inward. This proves the desired result since it is "Clear that U can be made to approximate J as well as desired by appropriately choosing the parameters used in the construction. EXERCJTRlll 1.1. Prove the following Theorem. Any open disk in R 2 which contains a bounded semiorbit of (1.1) must contain an equilibrium point. Hint: Use the Poincani-Bendixson Theorem and Theorem 1.8.2: EXERCISE 1.2. Give a generalization of Exercise 1.1 which remains valid inR 3 ? Give an example. EXERCISE 1.3. Prove the following Theorem. If div f has a fixed sign (excluding zero) in a closed two cell n, then n has no periodic orbits. Hint: Prove by contradiction using Green's theorem over the region bounded by a periodic orbit in n. EXERCISE 1.4. Consider the two dimensional system x =f(t, x), f(t 1, x) = f (t, x), where f has continuous first derivatives with respect to x. Suppose n is a subset of R2 which is homeomorphic to the closed Vnit disk. Also, for any solution x(t, xo), x(O, xo) = Xo, suppose there is a T(xo) such that x(t, xo) is in n for all t ;;;; T(xo). Prove by Brouwer's fixed point theorem that there is an integer m such that the equation has a periodic solution of period m. Does there exist a periodic solution of period 1 ?
+
EXERCISE 1.5. Suppose f as in exercise (1.4) and there is a ,\ > 0 such that x'f(t, x) ~ -,\ Ixl2 for all t, x. If g(t) = g(t 1) is a continuous function, prove the equation x = f(t, x) g(t) has a periodic solution of period 1.
+
+
EXERCISE 1.6. Suppose 'Yo is a periodic orbit of a two dimensional system and let"'G e and Gj be the sets defined in Theorem 1.4. Isit possible for an equation to have a(-y(p» = 'Yo for all noncritical points p in Gj and w('Y(q» = 'Yo for all qin G e ? Explain. EXERCISE 1.7 .. For Lienard's equation, must there always be a periodic orbit which is stable from the outside? Must there be one stable from the inside? Explain.
64
ORDINARY DIFFERENTIAL EQUATIONS
EXERCISE 1.8 Is it possible to have a two dimensional system such that each orbit in a bounded annulus is a periodic orbit? Can this happen for analytic systems? Explain.
11.2. Differential Systems on a Torus In this section, we discuss the behavior of solutions of the pair of first order equations
~
(2.1)
= 4>(~,
8),
8 =e(~,
8),
where
+ 1, 8) = 4>(~, 8 + 1) = 4>(~, 8), e(~ + 1, 8) = e(~, 8 + 1).= e(~, 8).
4>(~
(2.2)
e
We suppose 4>, are continuous and there is a unique solution of (2.1) through any given point in the ~, 8 plane. Since 4>, are bounded, the solutions will exist on ( - 00, (0). If opposite sides of the unit square in the (~, 8)-plane are identified, then this identification yields a torus :r and equations (2.1) can be interpreted as a differential equation on a torus. An orbit of (2.1) in the (~, 8)-plane when interpreted on the torus may appear as in Fig. 2.1.
e
Figure II.2.1
We also SUPP9se that (2.1) has no equilibrium points and, in particular, that 4>(~, 8) =1= 0 for all~, 8. The phase portrait for (2.1) is then determ~ed by ~
(2.3)
d8
~
= A(~,
8),
,'(;,
65
TWO DIMENSIONAL SYSTEMS
A(q,
+ 1, B) = A(q" B+ 1) = A(q"
B),
where A(q" B) is continuous for all q" B. The discussion will center around the solutions of (2.3). The torus Y can be embedded in R3 by the relations (2.4)
= (R + r cos 27TB) cos 27Tq,. y = (R + r cos 27TB) sin 27Tq" z = r sin 27TB, o ~ q, < 1, 0 ~ B< 1, 0 < r < R. x
This embedding is convenient for the sake of terminology only. We shall refer to the circle q, = constant as a meridian and B = constant as a parallel. Example 2.1. In (2.1), if = 1, 0 = w, a constant, then A(8, q,) = w in (2.3). The solution B(q" q,o, Bo), B(q,o, q,o, Bo) = Bo of (2.3) is B(q" q,o, Bo) = w(q, - q,o) + Bo· Case (i). If w is rational, say w = p/q, p, q integers, then B(q,o + q, q,o, Bo) = Bo + p. But, on the torus Y, the point (q,o, Bo) is .the same as (q,o + q, Bo + p) for any integers p, q. Thus, the orbit through (q,o, Bo) on Y is a closed curve for every initial point (q,o, Bo). This implies that the functions in (2.4) expressed as a function of t with the parametric represen· tation of B, q, being given by (2.1) are periodic in t. Case (ii). If w is irrational, there do not exist integers p, q such that B(q,o + q, q,o, Bo) = p + Bo for any Bo. Therefore, no orbit on Y will be closed and the functions in (2.4) will not be periodic. We show in this case that every orbit on Y is dense in Y. It is sufficient to show that the orbit is dense in the meridian q, = 0 since the trajectories in the plane all have constant slope w. There is a constant 8 > 0 such that for any y in [0, 1), there is a sequence of integer pairs (q", p,,), q" _ 00 as k _ 00 such that l'w - p,,/ qk - y/q"l< 8/q~. On Y, (q", B(q", 0, 0)) = (0, wq,,) = (0, Pk + y + "1") = (0, y + "1") where "1k _0 as k - 00. This proves the orbit through (0, 0) is dense but the same is obviously true for any other orbit. The functions (2.4) will be quasiperiodic in t (for the definition, see the Appendix) with the parametric representation of B, q, being given by (2.1) since for any initial point (q,o, Bo) they are obviously representable in the form x(t) = 81(t, wt), y(t) = 82 (t, wt), z(t) = 83 (t, wt) where each 8j is periodic in each argument of period 1. Example 2.2. IfA(q" B) = sin 27TB, there are two closed orbits B = 0, B = 1/2 on Y. It is clear that for any Bo , 0 < Bo < 1, the corresponding orbit on Y has w·limit set as the closed orbit B = 1/2 and lX·limit set the closed orbit B = 0 since the sets B = 0 and B = 1 on Yare the same.
66
·ORDINARY DIFFERENTIAL EQUATIONS
There are two striking differences in these examples. In example 2.1 and c.o irrational, the ex- and c.o-limit set of any orbit on ~ is ~ itself. In example 2.1 with c.o rational, every orbit is closed whereas in example 2.2, there are isolated closed orbits which are the limit sets of all other orbits. For smooth vector fields, it will be shown below that the c.o-limit and ex-limit sets of any orbit of a general equation (2.3) on ~ must either be ~ itself or a closed orbit. Since every solution of (2.3) exists on -00 < tP < 00, it follows that every orbit on ~ must cross the meridian C given by tP = and therefore it is sufficient to take the initial values as (0, g). Let 8(tP, e), 8(0, g) = g, designate the solution of (2.3) that passes through (0, g). From the assumption of uniqueness of solutions of (2.3), the function 8(tP, is a monotone increasing function in g for each tP. Also, the mapping g~ 8(1, is a homeomorphism of the real line onto itself and thus'induces a homeomorphism T of C onto itself. In fact,
°
e)
e)
e)) = (0, 8(1, e)). From the uniqueness of solutions of (2.3), 8(1, e) is a monotone increasing TP
=
PI, P
=
(0, e), PI
=
(1, 8(1,
function and thus T preserves the orientation of C. Also, periodicity of (2.3) and uniqueness of solutions imply
(2.5)
8(tP,
g+ m) =
8(tP,
e) + m
for any integer m. It is easy to see that Tnp
def
=
T(Tn-Ip)
=
(0, 8(n, e)); Tm+n(p)
=
Tm(Tnp), TOP
= P,
for· all m, n = 0, ± 1, .... First of all, the definition makes sense for negative values of the integers since 8(1, P) is a homeomorphism and TOP = Pimply that P = T(T-IP) uniquely defines T-'l. One then inductively defines T-2, T-3, etc. To prove the stated assertions, notice that periodicity in (2.3) and uniqueness of solutions imply that
8(m, 8(n,
e)) =
8(n, 8(m,
e)) =
8(m + n, g).
for all integers m, n. This relation immediately yields the above assertions. THEOREM
2.1.
The
rotation number of (2.3), def. 8( n, g) p= hm - - , Inl-- 00 n
~{
exists and is independent of f Also, p is rational if and only if some pbwer of T has a fixed point; that is, there is a closed orbit on the torus. The rotation number can be interpreted as the average rotation about
67
TWO DIMENSIONAL SYSTEMS
the meridian for one trip around the parallel or the average. slope of the line in the (cp, 8)-plane which passes through the origin and the point (n, 8(n, g» on the trajectory through (0, g). PROOF OF THEOREM 2.1. The proof is in four parts. (i) If p exists for a ~, then it exists for every g and is independent of f We need only consider 0;£ g, ~ < 1 since 8(cp, g) satisfies (2.5). For 0;£ g, ~ < 1, relation (2.5) and the monotonicity of 8(cp, g) in g imply that 8(cp, ~) -1
=
8(cp, ~ -1) ;£ 8(cp, g) ;£ 8(cp, ~
+ 1) = 8(cp, ~) + 1,
and this gives the desired result. (ii) p always exists. The following proof was communicated to the author by M. Peixoto. From (i), we need only consider g = O. For any real g there is an integer m such that 0 ;£ g - m ;£ 1 and thus 8(cp, 0) ;£ 8(cp, g - m) ;£ 8(cp, 1) and (2.5) implies 8(cp, 0);£ 8(~, g) - m ;£ 8(cp, 0)
+ 1.
If y = g - m, then 0;£ y ;£ 1, and this last relation implies 8(cp, 0) - y ;£ 8(cp, g) - g ;£ 8(cp, 0) 1 - y. Since 0 ;£ y ;£ 1, we have
+
8(cp, 0) -1 ;£ 8 (cp, g)
for all cp,
f
-.g ;£ 8(cp, 0) + 1
In particular, for any integer m, 8(m, 0) -1 ;£ 8(m, g) - g;£ 8(m, 0)
(2.6)
+ 1.
From this relation, we have 8(2m, 0) = 8(m, 8(m, 0» satisfies 28(m, 0) - 2 ;£ 28(m, 0) -1 ;£ 8(2m, 0) ;£ 28(m, 0) 1 ;£ 28(m, 0) 2. By successively applying (2.6), we obtain
+
+
n8(m, 0) - n ;£ 8(nm, 0) ;£ n8(m, 0)
for all n
~
+n
0, and n8( -m, 0) - n ;£ 8( --:-nm, 0) ;£ n8( :-m, 0)
for all n
~
O. Thus,
0) _ 8(m, 0) I: :; ~ I8(nm, nm m -Iml
for all n, m =1= 0: Interchanging the role of n, m, we have
I
8(nm, 0) _ 8(n,0) nm n
I:::; ~ -Inl
for all n, m =1= O. The triangle inequality implies 8(n,0) _ 8(m'0)1 :::;~+_1 ,
I and thus p exists with
Ip -
n
m
-Inl
Iml
8(m, O)/ml ;£ Il1ml for all m.
+n
68
ORDINARY DIFFERENTIAL EQUATIONS
We now give another proof. The reader may go immediately to part (iii) of the proof of the theorem if he so desires. The idea is best illustrated by another example. The slope of aline L through the origin of the (rfo, 8)-plane is uniquely determined by the manner in which this line partitions the integer pairs (m, n): More specifically, if Ro is the set of rationals nlm such that the pair (m, n) is below L, and Rl the set of rationals such that (m, n) is above L, then one shows that all rational numbers with the possible exception of one are included in Ro or Rl, Ro () Rl = {O, and therefore the cut defined by Ro and Rl defines a real number which is the slope of the line in question. This same idea can be used to show that the rotation number p exists. Let L(m, n) be the trajectory of (2.3) through the point (m, n) in the (rfo, 8)plane and let L = L(O, 0). The curve L(m, n) is the same as L except for a translation. We say L(m, n) is above L if 8(0, n) > 0 and below L If 8(0, m, n) < 0, where 8(rfo, m, n) is the solution of (2.3) with 8(m, m, n) = n. It is clear from uniqueness that this is equivalent to saying that L(m, n) is above (below) L if 8(rfo, m, n) > 8(rfo, 0, 0) « 8(rfo,0, 0». If L(m, n) is a'bove L, then L(km, kn), k a positive integer, is also above L. In fact, L(2m, 2n) is above L(m, n) since L goes to L(m, n) and L(m, n) goes to L(2m, 2n) by the translation (m, n). An induction process gives the result. Applying the translation (-m, -n) to the same curves Land L(m, n), we see that L( -m, -n) is below L and in general L( -km, -kn)' k > 0 is below L if L(m, n) is above L. These remarks show there is no ambiguity to the classification of Ro and Rl as before. Namely, nlm belongs to Ro(Rl) if L(m, n) is below (above) L. The classes Ro and Rl are nonempty because if n > 0 is sufficiently large, then L(I, n) is above Land L(I, -n) is below L. If nlm is not in Ro, and Blr > nlm, then Blr is in Rl. In fact,L(rm, rn) is either above L or coincides with Land L(rm, Bm) is above L(rm, rn) since it is obtained from it by the translation (0, Bm - rn) and Bm - rn > O. Therefore L(rm, Bm) is above L which implies rmlBm = rIB belongs to R l . Similarly, if nlm is not in Rl and sIr < mIn, then Blr is in Ro. Thus, all rational numbers .with possibly one exception are included in Ro or in Rl and Ro and Rl define a real number p. It remains to show that p is the rotation number defined in the theorem. Suppose m is a given integer and let n be the largest integer such that nlm is inRo. Then n ~ pm ~ n 1 and every point of L(m, n) is below L and every point of L(m, n 1) is not below L. There~ore, since the points on L(m, n) are (rfo m, 8(rfo, 0) n), we have
m,
+
+
+
+
8(",,0)
+ n < 8(rfo + m, 0) ~ 8(rfo, 0) + n + 1.
In particular, for rfo = 0,
n< 8(m, 0) ~ n+ 1,
69
TWO DIMENSIONAL SYSTEMS
and
18(: -pi ~ I; -pi + I~I· 0)
One can approximate p by a sequence of rationals njm, Ip - njml < Kjlml for some constant K. Thus, 8 (m, 0) _ p 1
I
~
K
Iml -+ 00, such that
+1.
mimi
This clearly implies that p =limlml-+oo 8(m, O)jm. (iii) If some power of T has a fixed point, then p is rational. In fact, if there is an integer m and a g in [0, 1) such that Tmg = g, there is also an integer le such that 8(m, g) = g le. Since p = limlml .. oo 8(m, g)jm exists we have
+
p=
lim 8( nm, g) = lim g nm
Inl-+ 00
Inl-+ 00
+ nle = !:.. nm
m
and p is rational. (iv) If p is rational, some power of T has a fixed point. Suppose p = lejm, le, m integers, m > 0, and Tm has no fixed points. Then, for every g, o ~ g ~ 1, 8(m, g) =F g le. If we suppose 8(m, g) > g le for gin [0, 1), then there is an a> 0 such that 8(m, g) -:- g -le ~ a> 0, g in [0, 1). For any' in (-00, (0), there are an integer p and a gin [0,1) such that ,=p+g. Relation (2.5) then implies 8(m, ') - , -le ~ a for all , in (-00, (0). A repeated application of this inequality yields 8(rm, g) - g ~ r(le a) for any integer- r. Dividing by rm and letting r -+ 00 we have p ~ lejm ajm which is a contradiction. This completes the proof of the theorem.
+
+
+ +
COROLLARY 2.1. Among the class of functions A(cp, 8) which are Lipschitzian in 8, the rotation number p = p(A) of (2.3) varies continuously with A; that is, for any e > 0 and A there is a 8 > 0 such that Ip(A) - p(B) I< e if maxO;;;;
dz dcp = [A(cp, z(cp)
+ 8B (cp, 0)) -
A(cp, 8B (cp, 0))]
- [B(cp, 8B (cp, 0)) - A(cp, 8B (cp, 0))], and
Drlzl ~
I I~ d: d.,..
Lizi
+0;;;;
70
ORDINARY DIFFERENTIAL EQUATIONS
for all q,. Thus, 18A(q" 0) - 8B(q" 0)1 ~ L-leL~ sup IB(q,. 8-A(q" 8)1 O;:;;~.8::ia
for all 4> ~ O. In the proof of part (ii) of Theorem 2.1, an estimate on the rate of approach of the sequence 8A(m, O)/m to the rotation number p(A) was obtained; namely, 18A(m, O)/m - p(A)1 < I/lml for all m. Therefore,
I
Ip(A) - p(B) I ~ p(A) _ 8A(:, 0)
+
1
I
8A (m, 0) : 8B(m, 0)
< -.:..
=Iml+
1 8A (m,
I+
1
0) - 8B(m, 0)
m
8B (:, 0) _ p(B)
I
I
for all integers m. For any 8 > 0, choose Iml so large that I/lml < 8/3. For any such given but fixed m, choose 8 > Osuch that 18 A(m, 0) - 8B(m, 0)1 ~ 1 if maxO;:;;8.~;:;;1 IA(q,, 8) - B(q" 8)1 < 8. This fact and the preceding.inequality prove the result. The conclusion ofOorollary 2.1 actually is true without assuming A(q" 8) is Lipschitzian in 8. The proof would use a strengthened version of Theorem 1.3.4 on the continuous dependence of solutions of differential equations on the vector field when uniqueness of solutions is assumed. It is an interesting exercise to prove these assertions. THEOREM 2.2. If the rotation number p is rational, then every trajectory of (2.3) on the torus is either a closed curve or approaches a closed curve. PROOF. (Peixoto). Since p is rational, there exists a closed trajectory y on fI which intersects every meridian of fl. Therefore, fI\y is topologically equivalent to an annulus r. The differential equation (2.3) on fI\y is equivalent to a planar differential equation on r. Since there are no equilibrium points, the Poincare-Bendixson Theorem, Theorem 1.2, yields the conclusion of the theorem. The remainder of this chapter is devoted to a discussion of the .behavior of the orbits of (2.3) when the rotation number p is irrational. Let T: 0 -+0 be the mapping induced by (2.3) which takes the meridian o of fI into itself. For any Pin 0, let ~.
D(P) = {Tnp,n =0, ±1, ±2, ... },
'.
and D'(P) be the set of limit points of D(P). Also, let 10 be the empty set.
TWO DIMENSIONAL SYSTEMS
71
LEMMA 2.1. Suppose p is irrational, m, n are given integers, P is a given point in C and IX, f3 are the closed arcs of C with IX n f3 = {Tmp, Tnp}, IX U f3 = C. Then D(Q) n IX O -=1= 0, D(Q) n f3 0 -=1= 0 for every Q in C, where IX O, f3 0 are the interiors of IX, f3 respectively.
Uk
PROOF. The set Tk(m-n)IXO covers C. For, if not, the sequence {Tk(m-nl(Tnp)} would approach a limit Po and T(m-nlPo = Po which, from Theorem 2,1, contradicts the fact that p is irrational. Consequently, for any Q in C, there is an integer p such that Q is in TP(m-nlIXO; that is, TP(n-mlQ is in IXO and D(Q) n IXO -=1= 0. The same argument applies to f3.
TF
THEOREM 2.3. If p is irrational, D'(P) and either
= F is the same for all
P,
=F
= C (the ergodic case)
(i)
F
(ii)
F is a nowhere dense perfect set.
or
PROOF. If S belongs to D'(P), there is a sequence {P k } c: D(P) approaching S as k -*00. For anypointsPk ,Pk+l of this sequence and Q in C, it follows from Lemma 2.1 that there is an integer nk such that TnkQ belongs to the shortest of the arcs IX, f3 on C connecting these two poiuts. Therefore . Tn'Q -*S as k -* 00 and D'(P) c: D'(Q). The argument is clearly the same to obtain D'(Q) c: D'(P) which proves the first statement of the theorem. If Q is in F, then there is a sequence nk and a P such that Tnk P -*Q as n -* 00. This clearly implies TQ belongs to F and T-IQ belongs to F. Therefore TF = F. If R is an arbitrary element of F, then the fact that F = D'(Q) for every Q implies for any Q E F there is a sequence of integers nk such that TnkQ -* R. Therefore, the set of limit points of F is F itself and F is perfect. Suppose F contains a closed arc y of C. Then y contains a closed subarc IX with endpoints Tn P, Tm P for some integers n, m and P in C. Therefore, by Lemma 2.1, TklX covers C and since TIX, T2 1X , ••• belong to F we have F = C. This proves the theorem . . Our next objective is to obtain sufficient conditions which will ensure that T is ergodIc; that is, the limit set F of the iterates of T is C. Let P n = Tnp, n = 0, ±1, ±2, .... If p is irrational and IX is any closed arc of C with P as an endpoint, Lemma 2.1 implies there is an integer n such that either P n or P -n is the only point P k in the interior IX O of IX for Ikl ~ n. Since no power of T has a fixed point, for any N > 0, IX can be chosen so small that n ~ N. For definiteness, suppose P -n is in 1X0. Let Po P- n denote the arc of C with endpoints Po, P -n and which also belongs to IX. We associate an orientation to this arc which is the same as the orientation of
Uk
72
ORDINARY DIFFERENTIAL EQUATIONS
= 0, 1, ... , n -1, designate· the arc of C joining P k , Pk'-n which has the same orientation as C.
G. Also, let Pk P k- n , k
LEMMA
2.2.
The arcs P k P k - n , k
= 0, 1, ... , n - 1, are disjoint.
PROOF. If the assertion is not true, then there exists an t from the set t -n, -n + 1, ... , n-l} and a k from {O, 1, ... , n -I} such that PI belongs to the interior P k P2-n of PkPk - n . Therefore, P'-k is in POP~n' from the orientation preserving nature of-powers of T. This is impossible in case -n ~ [- k < n from the choice of n. Suppose -2n + 1 ~ [- k<-n. Since P;belongs to P k PLn , it follows that P'+nPt and ~Pk-n intersect and, in particular, P k is in P'+nP~. Thus P k - n- t is in PoP~n which is impossible since < k - t - n < n. This proves the lemma.
°
THEOREM 2.4. Let ifJ(g) = 8(1, g), 0;;;;:; g;;;;:; 1. If p is irrational and ifJ possesses a continuous first Qerivative ifJ' > which is of bounded variation, then T is ~rgodic.
°
PRoor. Let gk = ifJk(g), k =0, ±1, ... , ifJO(g) = g, be recursively defined by choosing ifJ- 1 (g) as the unique solution of ifJO(g) = ifJ(ifJ- 1 (g)). Then Tkp = (Of ifJk(g)), P = (0, g). From the product rule for differentiation, we have
difJk(.g)
~=
DoifJ (gj), difJ-k(g) dg =
k-1,
[-1, ]-1 .Do ifJ (gj-k)
where ifJ'(g) = difJ(g)/dg. Suppose P and n are chosen as prior to Lemma 2.2. Since Pk Pk-n, k = 0, 1, ... , n -1, are disjoint we have I log (difJ;;e)
difJ~;(g))1 =
IlOgCfl ifJ'(gj) )-lOgCfl ifJ'(gj-n)) I
I
= ;t:[log ifJ'(gj) -log ifJ'(gj-n))
I
;;;;:;V where V is the total variation of log ifJ'. This function has bounded variation from the hypothesis on ifJ'. Therefore,
e-V < difJn(g) difJ-n(g) < eV =dg dg = uniformly on nand g. Suppose oc is a:ny arc on C of length S. If Sk is the length of Tk oc, then
Sk + S-k = and, therefore, Sk
difJ-k) (difJk difJ- k) ~ :i I..( difJk d[ + df dg ~ 2 fa dg d[" dg ~ 2 SeV 12
+ S-k does not approach zero as k -+ 00.
TWO DIMENSIONAL SYSTEMS
73
If G\F is not empty (that is, T is not ergodic), then take an open arc in G\F with end points in F. This can be done since F is nowhere dense and perfect. Since T F = F and T preserves orientation, all of the arcs Tk lX , k =0, ±I, ... are in G\F. Also TklX n Ti lX = 0, k =l=j, since the end points of these arcs are in F and if one coincided with another the end points would correspond to a fixed point of a power of T. Therefore, compactness of G yields Ok + O-k~O as k~oo. This contradiction implies G\F is empty arid proves the theorem. IX
Remark. The smoothness assumptions on if; in Theorem 2.4 are satisfied if A(r/>, 0) in (2.3) has continuous first and second partial derivatives with respect to O. In fact, Theorem 1.3.3 and exercise 1.3.2. imply that if;'(g), if;//(g) are continuous and, in particular, if;'(g) is of bounded variation for 0 ~ g ~ 1. Also, this same theorem states that fJ(}(r/>, g)/8g is a solution of the scalar equation dy 8A(r/>, 0) dr/> = 80 y, with initial value 1 at r/> = o. Thus, if;'(g) = 80(1, g)f8g > 0, O·~ g ~ 1. Denjoy [I] has shown by means of an example that Theorem 2.4 is false if the smoothness conditions on if; are relaxed. There is no known way to determine the explicit dependence of the rotation number p of (2.3) on the function A(r/>, 0), and, thus, in particular, to assert whether or not p is irrational. However, the result of Denjoy was the first striking example of the importance of smoothness in differential equa. tions to eliminate unwanted pathological behavior. Suppose the notation is the same as in Theorem 2.4 and the proof of Theorem 2.4. LEMMA 2.3. If P is irrational and g is a fixed real number, then the function g(gn + m) = np + m, gn = if;n(g), n, m integers, is an increasing func. tion on the sequence of real numbers {gn + m}. PROOF . . Throughout this proof,. n, m, r, 8 will denote integers. The order ofthe elements in {gn + m} does not depend upon g; that is, gn + m < gr + 8 implies + < + 8 for any'. This is equivalent to saying that gn - gr < 8 - m implies 'n - 'r < 8 - m for any'. If this were not true, there would be an 'Y) such that 'Y)n - 'Y)r is an integer which in turn implies some power of T has a fixed point, contradicting the fact that p is irrational. It suffices therefore to (lhoose g =0. Recall that if;m(o) = O(m, 0). If p ~ O(m, 0) ~ r, then a repeated applica. tion of (2.6) yields
'n m 'r
O(m, 0)
+ (k -I)p ~ O(km, 0) ~ O(m, 0) + (k -l)r,
74
ORDINARY DIFFERENTIAL EQUA.TIONS
for any k > O. Thus, 8(m, 0) (1) 8(km, 0) 8(m, 0) (1 1) --+ 1-- p:S:m :S:--+ --k r. k k km k
Taking the limit as k _ p<mp
00,
we obtain p
+
~
mp ~ r. Since p is irrational,
+
Now suppose 8(n, 0) m < 8(r, 0) 8, that is, 8(n, m) < 8(r, 8). Then 8(n - r, m) < 8(0, 8) or 8(n - r, 0) m < 8. From the preceding paragraph, this implies p(n - r) < 8 -m, or pn m < pr 8, which was to be proved.
+ +
+
THEOREM 2.5. H T is ergodic with (irrational) rotation number p, then T is topologically equivalent to a rotation of the circle 0 by an angle 27TP; that is, there is a homeomorphism G of 0 onto 0 such that GT = RG where R is the rotation of 0 through the angle 27Tp.
Let g be the increasing function defined in Lemma 2.3, A = {e~ + m}, n, m integers. Since T is ergodic, B is dense on the real numbers and since p is irrational, A is also dense on the real numbers. The function g is continuous from B to A. Since B is dense, g has a unique continuous increasing extension to all of the reals. We again designate this function by g. H 1J =en + m, g(1J) =np + m, then PROOF.
{np
+ m}, B =
g(1J+ 1) =np+m+ 1 =g(1J)
(2.7)
g("'(1J»
+ 1,
== g(",(en + m» = g(",(en) + m) = g(en+1 + m) = (n+l)p+m =g(1J)
+ p.
+
+
Since B is dense and g is continuous, it follows that g(1J 1) =g(1]) 1, g("'(1J» = g(1]) p for all real 1]. The homeomorphism G: 0 -0 is defined by G(1]) g(1J), 0 ~ 1J < 1. The relation g("'(1]» g(1]) p implies that GT = RG
=
+
=
+
a.nd the theorem is proved. THEOREM 2.6. (Bohl). H T is ergodic, there exists a function w(y, z) which is continuous in y, z and
w(y
+ 1, z) = w(y, z + 1) = w(y, z)
for all y; z such that every solution 8 of (2.3) satisfies (2.8)
8(c/J) = pc/J
+ c + w(c/J, pc/J + c),
where c is a constant and p is the rotation number. Conversely, for any constant c, iJ(c/J) given in (2.8) satisfies (2.3). Each c in [0, 1) correspo~s to a unique 8(0) (mod 1). -:<:PROOF.
Let
ebe any real number and 8(c/J, e) be the solution of (2.3)
75
TWO DIMENSIONAL SYSTEMS
with 0(0,
e) = f
We know that ()(rp, e ()(rp
+ 1) = ()(rp, e) + 1,
+ 1, e) =
()(rp, ",(e»,
where as before ,pte) = 0(1, e). Let g be the function given in the proof of Theorem 2.5 and let h denote the inverse of g. Fr"m the properties (2.7) of g, we have h(e
+ 1) =h(e) + 1,
"'(h(e» = h(e+ p).
If 8(rp, e) = ()(rp, h(e», then 8(rp, e
+ 1) =
()(rp, h(e
+ 1» =
()(rp; h(e)
+ 1) =
()(rp, h(e»
+1=
8(rp, e)
+ 1,
and 8(rp
+ 1, e) =
() (rp
+ 1, h(e» =
()(rp, "'(h(e») = ()(rp, h(e
+ p» =
8(rp, e +p).
If w(y, z) = 8(y, Z - py) - z for all real y, z, then one easily observes that w(y, Z + 1) = w(y + 1, z) = w(y, z). Therefore, def
O(rp, h(e»
= 8(rp,
e)
= prp + e + w(rp, prp + e),
which proves the theorem. EXERCISE 2.1. All functions below are continuous, periodic in 0, rp of period 1 and smooth enough to ensure a unique solution of the equations for any initial values. (a) What is the rotation number of d()/drp = sin 21T()? (b) If Ig(()1 < 1, 0;;;; () ;;;; 1, what is the rotation number of d()/drp = sin 21T0 + g( () ? (c) Using the Poincare-Bendixson theorem, prove that the rotation number of d()/drp = sin 21T0 + eg((), rp) is rational if lei is small enough. (d) Use the Brouwer fixed point theorem and the construction used in part (c) to show that the rotation number of the equation in (c) is zero if lei is small enough. . EXERCISE 2.2. Suppose w is irrational. For any e > 0, show there exists a function g((), rp) continuous in (), rp of period 1 such that max{lg(O, rp)l, o ;;;; (), rp ;;;; I} < e and the rotation number of d() /drp = w + g( (), rp) is rational and not all orbits are closed on the torus. Hint: Choose sequences of integers {Pk}, {qk}, qk -+ 00 as k -+ 00 such that Iw - Pk/qkl < for some constant y and consider g((), rp) = a sin 21T(b() + erp) for appropriate constants a, b, e.
yM
76
ORDINARY DIFFERENTIAL EQUATIONS
EXERCISE 2.3. In Exercise 2.1, arbitrarily small changes in the vector field did not change the rotation number, whereas in Exercise 2.2, an appropriate small change in the vector field did change this number. Can you explain why this is so? Can you give a general result for equation (2.3) which will exhibit the same properties relative to the rotation number as Exercises 2.1 and 2.2? What would you say is the most typical behavior among all veotor fields (2.3)1 EXERCISE 2.4. (a) For any continuous funotion /(q,) of period 1, show that the equation dOjdq, = 27TO /(q,) has a unique periodic solution of period 1. (Hint: Look at e2n (I(J-8)/(s) ds).
f:
+
(b) Designate this unique solution by K/. If P is the Banach space of oontinuous periodio funotions of period 1 with the topology of uniform oonvergenoe, show that K: P-+P is oontinuous and linear. (0) Suppose g(O, q,) has period 1 in 0, q" is oontinuous in 0, q, and lipsohitzian in O. Using the oontraotion prinoiple and part (b), show there is an eo> 0 suoh that the equation dOjdq, = 27TO + (sin 27TO - 27TO) + eg(O, q,) has a solution which is periodic in q, of period 1 if 1el < eo .
U.3. Remarks and Suggestions for Further Study For many more details on the general theory of two dimensional systems, see Hartman [1], Lefschetz [1], Sansone and Conti [1]. The book of Sansone and Conti [1] also contains a wealth of applications of the Poinoare-Bendixson theory to the existence of periodic solutions of a second order autonomous equation. Finding periodic solutions of nonautonomous equations by extensions of the ideas in Exercises 1.4 and 1.5 can be found, for example, in Cartwright [1], Levinson [1], Lefschetz [1], Sansone and Conti [1]. An interesting discussion of the global stability of the origin in the plane is given by Olech [1]. . One result obtained for two dimensional systems in the plane was that the only compaot minimal sets are points and closed curves. For a singularity free "smooth" veotor field on the torus, the only minimal sets on the torus are closed curves and the' torus itself. An outstanding problem for many years was the oharacterization of the minimal sets of " smooth" veotor fields on an arbitrary oompact two dimensional manifold. This problem was solved by Schwartz [1] when he proved that the only possible minimal sets are points, olosed curves and' a torus, the latter being possible only when the ~nifold is itself a torus. ~, . It is difficult to pose questions in n dimensions with answers as elegant as the ones given in this chapter. One problem that has been studied in
TWO DIMENSIONAL SYSTEMS
77
considerable detail arises from a more careful study of Theorems 2.5 and 2.6. Consider a differential equation on an n-dimensional torus
x =f(X) where x = (Xl, ... , Xn ), f(XI, ... , xn) is periodic in each variable of period 1. Problem: When does there exist a transformation of variables X = g(y) such that the new equation for y is if = w where w is a constant n-vector? If such a transformation exists, then the resulting flow is easy to analyze and one has a generalization of Theorem 2.6. For f(x) = w + eh(x) where e is small and w belonging to a certain class of irrationals,' some results along this line are available (see Arnol'd [1]). The techniques of proof are currently the most important tools in the study of stability in celestial mechanics (see Kolmogorov [1], Arnol'd [2], Moser [1]). In Section 1, the van der Pol equation (1.7) was discussed when the parameter k was large. In the coordinates described by (1.8), it becomes clear that the asymptotically stable periodic orbit has the property that the system moves along its orbit at a reasonable pace as long as it remains close to a given curve and then moves at a very rapid pace away from this curve. Such relaxation oscillations are very important in the applications and the reader may consult van del' Pol [1, 2], LaSalle [1], Minorsky [1], Pontryagin and Mishchenko [1].
CHAPTER III Linear Systems and Linearization
In the previous chll.pter, much information concerning the qualitative behavior of solutions of two dimensional autonomous systems was obtained without using any particular properties of the vector field.· In higher dimensions and even in two dimensions when more specific information is desired, one must resort to techniques which are more analytical in nature and yield information only in a. neighborhood of some solution or an invariant set of solutions. If c/>(t) is a solution of (1)
i =f(x),
where f has continuous first derivative!;!, then x = c/>(t) (2)
iJ =
+ yin (1) implies
A(t)y + g(t, y)
where A(t) = of(c/>(t»/ox and g(t, 0) = 0, og(t, 0)/8y = O. It is quite natural to study the linear equation
(3)
z=A(t)z,
and to inquire about the relationship between the solution z of (3) and the solution y of (2) in a neighborhood of y = O. More specifically, is the linear equation (3) a good" approximation" to equation (2) near y = O? Some questions of this type are -considered in this chapter. More precisely, the theory of general linear systems is given in Section 1 with Section 2 being devoted to characterizations of the concepts of stability for linear systems and the preservation of these stability properties for special perturbations of a linear system. Section 3 is a specializati~n of the general theory to nth order equations. In Section 4, the fundamental solution of linear autonomous systems is characterized in terms of the ele~entary functions with the phase portrait of the two dimensional system~ 'iscussed in Section 5. Section 6 is devoted to an autonomous equation (2)' with no eigenvalue of A on the imaginary axis. It is shown in this section that many of the qualitative properties of the solutions of t~e linear and nonlinear equation
78
LINEAR SYSTEMS AND LINEARIZATION
79
are the same near y = o. Section 7 gives the theory of linear equations with periodic coefficients with the Floquet representation included. Sections 8, 9 and 10 are devoted to stability theory for special classes of linear equations with periodic coefficients.
ill.l. General Linear Systems Consider the linear system of n first order equations n
(1.1)
xi = L ajk(t)xk + hj(t),
j
=
1, 2, ... , n,
k=1
where the ajk and hj for j, k = 1, 2, ... , n are continuous real or complex valued functions on the interval (- 00, 00). In matrix notation, equation (1.1) can be written in more compact form as
+
(1.2)
x = A(t)x + h(t)
where A = (ajk),j, k = 1, 2, ... , n; h = COl{hl' ... , h~)1 The basic characteristic property of linear systems of the form (1.2) is the lPrinciple of Superpositionl If x is a solution of (1.2) corresponding to the" forcing function" or " input" hand y is a solution corresponding to g then ex + ay is a solution corresponding to the forcing or input function ch + r1g for any real or complex numbers c and a. This principle is verified by direct computation. In particular, if c = - a = 1, h = g, then x and y being solutions of (1.2) corresponding to the forcing function h implies that x .,- y is a solution of the homogeneous equation (1.3)
x =A(t)x.
If cp(t, c) is a general solution of the homogeneous equation (1.3) and xp(t) is any particular solution of (1.2), then this latter property states that any solution of (1.2) is of the form cp(t, c) +xp(t) for some c. Below, we describe theoretically how to obtain the general solution of (1.3) and then show that a particular solution of (1.2) can be found from the knowledge of the general solution of (1.3) ,and a quadrature. From the properties of the coefficient matrix A and the forcing function h in (1.2), the basic existence and uniqueness theorem in Chapter I implies that the initial value problem for (1.2) has a unique solution which exists on an interval containing the initial time. Theorem 1.6.2 also asserts that every solution of (1.2) exists on the infinite interval (-00, 00). A set of vectors xl, ... , xn are said to be linearly independent if =1 Cj xi = 0 for any complex constants cj implies cj = 0 for j = 1, ... , n.
+
Lf
80
ORDINARY DIFFERENTIAL EQUATIONS
The vectors Xl, ... , xn are said to be linearly dependent if they are not linearly independent. The proof of the following lemma is left as an exercise. LEMMA 1.1. The vectors xl, ... , xn are linearly independent if and only if det[xl, ... , xn] #0. An n X n matrix X(t), t > 0, is said to be an n X n matrix solution of (1.3) if each column of X(t) satisfies (1.3). A fundamental matrix solution of (1.3) is an n X n matrix solution X(t) of (1.3) such that det X(t) # O. A principal matrix solution of (1.3) at initial time to is a fundaQlental matrix solution such that X(to) = J, the identity matrix. The principal matrix solution at to will be designated by X(t, to). From the above definition of a fundamental matrix solution it is clear that a fundamental matrix solution is simply a matrix solution of (1.3) such that the n columns of X(t) are lineady independent. An elementary but useful property of a matrix solution of (1.3) is LEMMA 1.2. If X(t) is an n X n matrix solution of (1.3), then either det X(t) # 0 for all tor det X(t) = 0 for all t. PROOF. If there exists a real number T such that the matrix X(T) is singular, then there exists a nonzero vector c such that X(T)C = O. For this vector c, linearity of (1.3) implies the function x(t) = X(t)c is a solution of (1.3). Since the function x = 0 is obviously a solution of the equation, uniqueness implies X(t)c =0 for all t in (-00, (0) and thus det X(t) =0 for all t in (-00, (0). This proves the lemma. COROLLARY 1.1. If Xo is an n X n nonsingular matrix and if X(t) is the matrix solution of (1.3) with 'X(O) = X o , then X(t) is a fundamental matrix solution of (1.3). Corollary 1.1 shows that the determination of a fundamental matrix solution consists only of selecting n linearly independent initial vectors and finding the corresponding n linearly independent solutions of the system (1.3). The fact that the matrix X(t) is nonsingular for all t yields the following important LEMMA 1.3. If X(t) is any fundamental matrix solution of (1.3), then a general solution of (1.3) is X(t)c where c is an arbitrary n vector. PROOF. It is clear that the function x(t) = X(t)c is a solution of (1.3) for any constant n-vect{)r c. Furthermore, if to and Xo are given, then X(t)c, c = X-I(to)xO, is the solution of (1.3) which passes through the point (to, xo). LEMMA 1.4 .. If X(t) is a fundamental matrix solution of (1.3), $en the matrix X-I(t) is a fundamental matrix solution of the adjoint equ4ttOn (1.4)
y=
-yA(t)
LINEAR SYSTEMS AND LINEARIZATION
81
where y is a row n-vector; that is, each row of X-I satisfies (1.4) and det X-l(t.) -::1= o. Let Y(t) = X-l(t). Since YX = I, the identity, it follows that Since X is non-singular this implies that Y = X-I is a matrix solution of the adjoint equation (1.4). This is equivalent to saying that each row qfthe matrix X-I is a solution of (1.4). X-I is obviously nonsingular. PROOF.
yX
+ YAX = O.
THEOREM 1.1. If X is a fundamental matrix solution of (l.3) then every solution of (1.2) is given by formula (1.5)
x(t)
for an real number
= X(t) [X-l(T)X(T) + T
in (-00,
f:
X-l(s)h(s)
dB]
+ 00 .
Formula (1.5) is referred to as thehariation of constants formulal for (1.2). The reason for this is clear since it implies that every solution has the form X(t)c(t) where c(t) is the function given in the brackets· in (1.5). This is the same form as the general solution of (1.3) given by Lemma 1.3 except that the vector c now depends upon the independent variable t. PROOF OF THEOREM 1.1. If we rewrite equation (1.2) as x - Ax = h and use the fact that X-I is a matrix solution of the adjoint equation (1.4), then equation (1.2) is equivalent to d
d,s [X-l(s)x(s)]
Integrating this expression from weobtain
= T
X-l(s)h(s).
to t for any real numbers
T
and t,
I. X-l(s)h(s) dB. t
X-l(t)x(t) - X-l(T)X(T) =
T
A rearrangement of the terms in this expression yields equation (1.5) and the theorem is proved. We give another proof of this theorem. Since X(t) is nonsingular for each t, it follows that the transformation of variables x = X(t)y defines a homeomorphism of Rn into Rn for each t. If this transformation is applied to (1.2), then by Lemma 1.4, d
iJ = -
dt
[X-Ix]
=
X-l(Ax
+ h) -
X-lAx
=
This implies that t
y(t) =X-l(T)X(T)
+ I. X-l(s)h(s) T
X-lh.
82
ORDINARY DIFFERENTIAL EQUATIONS
since Y(T) = X-l(T)X(T). The formula x(t) = X(t)y(t) yields relation (1.5). If for any given T in (-00, (0), we let X(t, T), X(T, T) = I, designate the principal matrix solution of (1.3) at T, then X(t, T) = X(t, 8)X(8, T)
(1.6)
for all t, T, 8. In fact, considered as functions of t, both sides of this relation satisfy (1.3) and coincide for t = 8. Uniqueness yields the result. This formula leads to a simplifica.tion of the variation of constants formula; namely t
(1.7)
x(t) = X(t, T)X(T)
+ f..,. X(t, 8)11.(8) ds .
I LEMMA. 1.5.1 If X(t) is an n X n matrix solution of (1.3), then det X(t) = [det X(to)] exp ( ( tr A(8)
(1.8)
ds)
for all t, to in ( - 00, (0), where tr A is the sum of the diagonal elements of A. The proof is easily supplied by showing that det X(t) satisfies the scalar equationz =[tr A(t)]z and is left as an exerpise for the reader. The above theory is valid for linear systems with A(t), h(t) integrab~e in t. If the elements of the matrix A(t) and the components of h(t) are only Lebesgue integrable on every compact subset of R, then the results of Section 1.5 imply that the initial value problem for (1.2) has a unique solution. If x(t) = x(t, to, xo), x(to, to, xo) = Xo, is a solution of (1.2) then Ix(t)1
~ Ixol +
f 111.(8)1 ds + f IA(8)IIX(8)1 t
t
to
to
d8,
t
~ to,
as long as x(t) is defined. The generalized Gronwall inequality (Lemma 1.6.2 and an integration by parts implies
Ix(t) I ~ (exp(IAfs)ldS)[lxol + (lh(S)ldS] for all t ~ to for which x(t) is defined. The continuation theorem implies x(t) is defined for t·~ to. Replacing t by to - u and using similar estimates one obtains existence of x(t, to , xo) for t ~ to. The general structure of the solutions of the homogeneous equation (1.3) and the variation of constants formula are proved exactly as before. We also need some estimates on the dependence of the solutions of (1.2) on the right hand side of the differential equation. Suppose A, Bare n X n integrable f!latrix functions and 11., g are n-vector integrable fur~ctions on compact subsets of R and consider the equations
?
(1.9)
(a)
x=
A(t)x
+ h(t),
(b)
'Ii =
B(t)y
+ g(t).
83
LINEAR SYSTEMS AND LINEARIZATION
If x, yare solutions of (1.9a), (1.9b) respectively, then x - y satisfies the equation
z = Az + (A -
B)y
+h -
g.
Thus,
Ix(t) -
y(t)1
~
Ix(to) - y(to)1
+ Ih(s) -
+ f.
g(s)1J
t
to
[IA(s)llx(s) - y(s)1
+ IA(s) -
B(s)lly(s)1
t ~ to·
ds,
Using the generalized Gronwall inequality (Lemma I.6.2), we have (1.10)
Ix(t) - y(t)1
~ (exp Ie: IA(s)1 ds )Ix(to) -
y(to)1
I:
+ { (exp IA(u)1 dU )[IA(S) + Ih(s) - g(s)1J ds.
B(s)lly(s)l
In particular, if XA(t, to), XA(fe I to) = I, XB(t, to), XB(to, to) = I, are fundamental matrix solutions of x = A(t)x, if = B(t)y, respectively, then (1.11)
IXA(t, to) - XB(t,
to)1 ~ to~!;)XB(U, to)l( exp {IA(s)1
f. IA(s) -
ds)
t
X
B(s)1
ds,
to
for all t ~ to. Relation (1.11) implies that the principal matrix solution of x·= Ax is a continuous function of the integrable matrix functions A defined
I;
on [0, t] with IAI = IA(s)1 ds. When a linear differential equation .contains general time varying coefficients, the remarks in. this section essentially comprise the theory concerning the specific structure of the solutions. For special equations, much more detailed information is available. For equations with constant or periodic coefficients, the general structure of the solutions is known and discussed in Sections 4 and 7.
DI.2. Stability of Linear and Perturbed Linear Systems
In this section, characterizations of stability for linear systems are given and the results applied to stability of perturbed linear systems. We firs~ remark that the stability of any solution of Q011l0VAnAOUS linear.eqpation.ia,
ORDINARY DIFFERENTIAL EQUATIONS.
84
determined by the stability of the zero solution. Therefore, for linear systems, it is' permissible to say system (1.3) is stable, uniformly stable, etc. THEOREM 2.1. Let X(t) be a fundamental matrix solution of (1.3) and let f3 be tny nJmber in (-00, (0). The system (1.3) is (i) stabl for any to in (-00, (0) if and only if there is a K = K(to) > 0 such that IX(t)1 ~ K,
(2.1)
to
~
t < 00;
I
(ii) uniformly stablelfor to ~ f3 if and only if there is a K =K(f3) >0 such that IX(t)X-1(s)1 ~ K,
(2.2)
(iii) (2.3)
to ~ s ~ t
<
00;
I asymptotically stablelfor any to in (-00, (0) if and only if
r
IX(t)1 ~O
as
t~
00;
(iv) uniformly asymptotically stablelfor to ~ f3 if and only if it is exponentially asymptotically stable; that is, there are K= K(f3) > 0, at: = at:(f3) > 0 such that IX(t)X-1(S)1
(2.4)
~
Ke-a(t-sl,
PROOF. (i) Suppose to is any given real number in (-00, (0) and (2.1) holds. Any solution x(t) satisfies x(t) = X(t)X-1(tO)x(to). Let IX-1(to)1 = L. Then Ix(t)1 ~ KL Ix(to)1 < e if Ix(to)1 < elKL and the zero solution of (1.3) is stable. Conversely, if for any e > 0 there is a 8 = 8(e, to) > 0 such that IX(t)X-1(tO)x(to)1 < e for Ix(to)1 < 8, then IX(t)X-1(to)1 = sup IX(t)X-1(tO)gl lei <:;1 =
sup IX(t)X-1(to)8- 1x(to)1
~
e8- 1
1",(toll<:;8
for t
~
to. This proves (i). If (2.2) is satisfied, then x(t) =X(t)X-1(tO)x(to) for any to ~ f3and Ix(t)1 ~ K Ix(to)1 < e, t ~ 7', if Ix(to)1 < elK which is uniform stability. The converse follows as in (i) with the observation that 8 is independent of to. (iii) If (2.3) is satisfied, then for any to in (-00, (0) there is a K = K(to) such that IX(t)1 ~ K, t ~ to and (i) implies stability. Since x(t) = X(t)X-1(tO)x(to) we have Ix(t)1 ~O as t ~ 00. The converse is trivial. (iv) If (2.4) is satisfied, then (1.3) is uniformly stable from (ii).~~uppose Ix(to)l ~ 1. For any TJ > 0, 0 < TJ < K, let T = -at:-1log(TJIK). Thel{'
(ii)
Ix(t)1 = IX(t)X-1(tO)x(to)1 ~ Ke-a(t-tollx(to)1 ~ TJ,
85
LINEAR SYSTEMS AND LINEARIZATION
if t ~ to + T, to ~ {3; that is, uniform asymptotic stability of the zero solution of (1.3). Conversely, suppose the solution x = 0 is uniformly asymptotically stable for to ~ (3. There is a b > 0 such that for any '1, 0 < '1 < b, there is a T = T('1) > 0 such that (2.5)
IX(t)X-l(tO)X(to)1
< '1
for
and all to ~ {3, Ix(to)1 ~ b. Thus, IX(t)X-l(to)1 In particular, (2.6)
IX(t
t
~ to+ T,
< '1b- 1 for
+ T)X-l(t)1 < '1b- 1 < I,
t ~ to
+ T,
to ~ {3.
t ~ (3.
Since the solution x = 0 is uniformly stable, we have from (ii) that there is an M = M({3) such that IX(t)X-l(S)1 ~ M, (3 ~ s ~ t < 00. Suppose oc = -T-1log('1b- 1),
K=Me aT .
For any t ~ to, there is an integer k Thus, using (2.6),
~
0 such that kT
~ t - to
< (k + I)T.
+ kT)1 . IX(to + kT).X-l(to)1 ~ M 'IX(to + kT)X-l(to)1 ~ M('1b-1)IX(to + (k -1)T)X-l(to)1
IX(t)X-l(to)1 ~ IX(t)X-l(to
~
M('1b-1)k = Me- akT
= Ke-a(k+1)T
~
Ke-a(t-to).
This proves (iv) and Theorem 2.1. Using the variation of constants formula and Gronwall's inequality (Corollary 1.6.6), it is very easy to obtain the following stability results for perturbed linear systems. THEOREM 2.2. Suppose {3 is given in (-00, (0) and the system (1.3) is uniformly stable for to ~ (3. If the n X n continuous matrix function B(t) satisfies J/l<X> IB(t)1 dt < 00, then the system
x=
(2.7)
[A(t)
+ B(t)]x
is uniformly stable. PROOF. If X(t) is a fundamental matrix solution of (1.3), then the variation of constants formula implies that any solution of (2.7) has the form
+f
t
(2.8)
x(t) = X(t)X-l(tO)x(to)
to
X(t)X-l(s)B(s)x(s) ds,
86
ORDINARY DIFFERENTIAL EQUATIONS
for all t, to in (-00, 00). If to ~ {3, then Theorem 2.1 implies there is a constant .K =K({3) such that IX(t)X-l(to)1 ~ K for all t ~ to. Consequently,
~K
Ix(t)1
+f
t
Ix(to)1
K IB(s) 1'lx(s)1 ds,
to
t
~ to
.
Gronwall's inequality implies Ix(t)1
~ K( exp K
rift )lx(to)l,
(IB(S)I
This clearly implies uniform stability of (2.7) and the theorem. THEOREM 2.3. Suppose {3 is given in (- 00, 00) and system (1.3) is uniformly asymptotically stable for to ~ {3. If the n X n continuous matrix function B(t) satisfies
f
(2.9)
to
IB(s)1 ds
~ y
to)
+
T,
for some constants T = T({3), y = y({3), then there is an r > 0 such that system (2.7) is uniformly asymptotically stable if y < r. PROOF. If X(t) is a fundamental matrix solution of (1.3) and (1.3) is uniformly asymptotically stable for to ~ {3, then Theorem 2.1 implies there are qonstants K = K({3) > 0, ex = ex({3) > 0 such that (2.4) is satisfieq. Using (2.8), one observes that the solution x(t) of (2.7) satisfies Ix(t)1
~ Ke-a(t-to>lx(to)1 + K
r
t ~ to.
e-a(t-BlIB(s)1 'lx(s)1 ds,
to
If z(t) = eatlx(t)l, this inequality implies z(t)
~ Kz(to) + -
f
t
to
K IB(s)1 z(s) ds,
An application of Gronwall's inequality and (2.9) yield z(t)
~ K( exp K
(IB(S)I ds )Z(to)
~ K 1eK 1'(t-tolz(to),
Kl
= Ke KT ,
which in turn implies Ix(t)1 ~ Kllx(to)1 exp[ -(ex - Ky)(t - to)], t ~ to. If r'= exK-l and y < r, then system (2.7) is uniformly asymptotically stable and the theorem is proved. THEOREM 2.4, Suppose {3 is given in (-00, 00) and system (1.3) is uniformly asymptotically stable for to ~ {3. If f(t, x) is continuous fcj (t, x) in R X R" and for any s > 0, there is a (j > 0 such that }G. (2.10)
If(t, x)1 ~ slxl
for
Ixl
<
(j,
tin R,
LINEAR SYSTEMS AND
then the solution x
~INEARIZATION
87
= 0 of
x = A(t)x + f(t, x)
(2.11)
is uniformly asymptotically stable for to
~
f3.
PROOF. The hypothesis of uniform asymptotic stability for to ~ f3 of the system (1.3) implies there are constants K = K(f3) > 0, ex = ex(f3) > 0 such that (2.4) is satisfied. Any solution x of (2.11) satisfies (2.12)
x(t)
=
X(t)X-l(tO)X(to)
+
t t
X(t)X-l(s)f(s, x(s)) ds.
Choose e so that eK < ex and let a be chosen so that (2.10) is satisfied. For those values of t ~ to for which x(t) satisfies Ix(t)1 < a, we have Ix(t)1
~ Ke-a(t-to)lx(to)1
+f
t
eKe-a(t-s)lx(s)1 ds.
to
Proceeding exactly as in the proof of the previous theorem, we obtain (2.13)
Ix(t)1 ~ K Ix(to)1 e-(a-eK)(t-to)
for all values of t ~ to for which Ix(t)1 < a. Since ex - eK > 0, this inequality implies Ix(t)1 < a for all t ~ to as long as Ix(to)1 < a/K. Consequently (2.13) holds for all t ~ to provided Ix(to)1 < a/K. Relation (2.13) clearly implies uniform asymptotic stability of the solution x = 0 and the theorem is proved. Theorem 2.4 is a generalization of the famous theorem of Lyapunov on stability with respect to the first approximation. The reason for the terminology is the following: Suppose g: Rn+l ~ Rn is continuous, g(t, x) has continuous first partial derivatives with respect to x and g(t, 0) = 0 for all t. If A(t)x = [og(t, O)/ox]x and f(t, x) = g(t, x) - A(t)x; then f(t, 0) = 0, af(t,x)/ax approaches zero as x approaches zero uniformly in t and the conditions of Theorem 2.4 are satisfied. The function A(t)x is clearly the linear approximation to the right hand side of the eqatuion X= g(t,x). There are many more results available on stability of perturbed linear systems and clearly many variants that could be obtained by abstracting the essential elements of the proofs given above. The exercises at the end of this section indicate some of the possibilities. In spite of the fact that the proofs of the above theorems are extremely simple and the results are not surprising to the intuition, extreme care must be exercised in treating arbitrary perturbed linear systems. The example it - 2t- 11i, + u = 0 has a fundamental system of solutions sin t - t cos t, cos t + t SiR t and is therefore unstable. This equation can be considered as a perturbation of the uniformly stable system it + u = o. The following example is an equation (1.3) which is asymptotically
88
ORDINARY DIFFERENTIAL EQUATIONS
stable, but not uniformly, and system (2.7) is unstable for an appropriate B(t) such that
fO IB(s)1 ds <
00
and IB(t)I---+O as t ---+ 00. Consider the system
(1.3) with (2.14) and 1
A(t) =
0 ] [ -a sin log t + cos log t - 2a '
0
< 2a < 1 + e- n . The solutions of (1.3) Xl(t) =
Cl
are
exp( -at),
X2(t) = C2 exp(t sin log t - 2at),
where Cl, C2 are arbitrary constants. For any exponentially. If B(t) is defined by B(t) =
then
f'
Cl,
C2, Xl(t), X2(t) ---+0 as t ---+
00
[e~at ~],
ds < 00 and IB(t)I---+O as t ---+ 00. The solutions of the perturbed
IB(s)1
equation (2.7) with initial time to = 0 are Xl(t)
= Cl
exp( -at),
X2(t) = [C2
+ Cl f~ exp( -s sin log s) ds] exp(t sin log t -
If IX is chosen so that 0 -< IX < 7T/2, tn = exp[(2n - 1/2)7T], n sin log s ~ -cos IX for tn ~ s~ tne IX • Hence,'
fo
tne n
exp( -s sin log s) ds
f ~f >
=
2at). 1, 2, ... , then
tneCl
exp( -s sin log s) ds
tn tne«
exp(s cos
IX)
tn
ds
>tn(e IX -I) exp(tn cos IX).
Choose C2 =0,
Cl
= 1. Since sin log(tne n ) = 1, we have IX2(tnen)1 ~ tn(e IX -1) exp(bt n ),
where b = (1 - 2a)e n + cos IX. Ifwe choose IX so that b > 0, then IX2(tn e~)I---+ 00 as n ---+ 00 and the system is unstable. ,~ EXERCISE 2.1. Suppose there is a constant K such that a fundamental matrix solution X of the real system (1.3) satisfies IX(t)1 ~ K, t ~ f3 and
LINEAR SYSTEMS AND LINEARIZATION
f
t
lim inf t .... 00
tr A(s) ds
-00.
{3
Prove that X-I is bounded on [{3, approaches zero as t _ 00. EXERCISE 2.2.
>
89
00)
and no nontrivial solution of (1.3)
Suppose A satisfies the conditions in Exercise 2.1 and
B(t) is a continuous real n X n matrix for t ~ {3 with f{3°O IA(t) - B(t)1
<
00.
Prove that every solution of iI = B(t)y is bounded on [{3, 00). For any solution x of (1.3), prove there is a unique solution y of iI = B(t)y such that y(t) -x(t)_O as t_ 00. EXERCISE 2.3.
Suppose system (1.3) is uniformly asymptotically stable, 00. Prove there is aT> {3 such that any solution x(t) of
f satisfies the conditions of Theorem 2.4 and b(t) _0 as t _
x = A(t)x + f approaches zero as t _
00
(t, x)
+ b(t)
if Ix(T)1 is small enough.
EXERCISE 2.4. Generalize the result of Exercise 2.3 with b(t) replaced by g(t, x) where g(t, x) _0 as t _ 00 uniformly for x in compact sets. EXERCISE 2.5. [t+1
Jt
c(s) ds
~ y,
t
~
Suppose there exists a continuous function c(t) such that
{3, for some constant y
= y({3)
and f: Rn+1_Rn is con-
tinuous with if(t, x)1 ~ c(t)lxl. Prove there is a constant r >0 such that the solution x = 0 of (2.11) is uniformly asymptotically stable if y < r. EXERCISE 2.6. Generalize Exercises 2.3 and 2.4 with conditions of Exercise 2.5.
m.3.
f
satisfying the
nth Order Scalar Equations
Due to the frequency of occurrence of nth order scalar equations in the applications, it is worthwhile to transform the information obtained in Section 1 to equations of this type. Suppose y is a scalar, aI, ... , an and g are continuous real or complex valued functions on ( - 00, 00) and consider the equation
+
(3.1) where D represents the operation of differentiation with respect to t. The function D2y is the second derivative of y with respect to t, and so forth.
90
ORDINARY DIFFERENTIAL EQUATIONS
Equation (3.1) is equivalent to
(3.2)
i
='Ax + h
~],
x=[ l)n-2y A=[ l)n-ly
~
1
o o
0
o 1
o
-an
From this representation of (3.1), a solution of (3.2) is a column vector of dimension n, but the (j 1)th component of the solution vector is obtained by differentiation of the first component j times with respect to t and this first component must be a solution of (3.1). Consequently, any n X n matrix solution [~l, ... , ~n], ~j an n-vector, of (3.2) must satisfy ~j = col(c/>1, l)c/>1, ... , l)n-l c/>1),where cf>i, j = 1, 2, ... , n, is a solution of (3.1). If c/Jl, ... , c/>n are n-scalar functions which are (n -I)-times continuously differentiable, the Wronskian tl(c/>l, ... ,c/>n) of c/>l, ... , c/>n is defined by
+
J
D~t,:.. c/>n
(3.3)
A set of scalar functions c/>l, ... , c/>n defined on a ~ t ~ b are said to be linearly dependent on [a, b) if there exist constants Cl, ... , Cn not all zero such that Clc/>l(t) cnc/>n(t) = 0 for aUt in [a, b].-Otherwise, the functions are linearly independent on [a, b].
+ .,. +
LEMMA 3.1. If c/>l, ... , c/>n are n -1 times continuously differentiable scalar functions on an interval I, then c/>l, ... , c/>n are linearly independent on I if the Wronskian tl defined in (3.3) is different from zero on I. PROOF. Suppose there is a linear combination of the c/>t which is zero on I. More precisely, suppose that n
L0c/>j(t) =0,
t in I,
j=O
where the Cj are constant. By repeated differentiation of this relation, the following system ?f equations for the constants Cj is obtained: n
L Cj l)kc/>j(t) j=l
Or, in matrix form,
=
0,
k =0,1,2, ... , n-1.
LINEAR SYSTEM'S AND LINEARIZATION
rpl Drpl
[
91
rp2 Drp2
Dn~lrpl Dn~lrp2 This latter relation must be satisfied for all t in I. On the other hand, by hypothesis, the determinant of the coefficient matrix is different from zero for all t in I which implies that the only solution of this system of equations is Cl = ... = Cn = 0; that is, the functions rpj are linearly independent. The converse of this statement is not true. In fact, there can be nfunctions with n - 1 continuous derivatives on an interval I such that the Wronskian is identically zero for all t in I and yet the functions are linearly independent. For n = 2, one merely has to choose two differentiable functions rpl, rp2 on [0, 3] such that rpl(t) = 0, t in [0, 2], rp2(t) = 0, t in [1, 3] and the functions are ::f= 0 otherwise. On the other hand, it is clear from the proof of Lemma 3.1 that linear dependence of rpl, ... , rpn on I implies ~ 0 on I. If rpl, .' .. , rpn are solutions of the homogeneous equation
=
(3.4) then ~(rpl' ... , rpn) is the determinant of an n X n matrix solution of the homogeneous system (3.2); that is, system (3.2) with h =0. Lemmas 1.2, 1.5, formula (3.2) and the remarks above imply THEOREM 3.1. If rpl, ... , rpn are n solutions of (3.4), then ~(rpl' ... , rpn)(t) is ::f= 0 for all t in (- 00, (0) or identically zero. More specifically,
~(rpl' ... , rpn)(t) = ~(rpl' ... , rpn)(O) exp ( - I~ al(s) ds). Thus, the n solutions rpl, ... , rpn of (3.4) are linearly independent ifand only if b(rpl, ... , rpn)(O) ::f= O. Let us determine the variation of constants formula for the solutions of equation (3.1). Since (3.1) is equivalent to (3.2), it is sufficient to determine only the variation of constants formula for the first component of the vector solution of (3.2). If X(t, T), X(T, T) = I, is the principal matrix solution of x = Ax, then any solution of (3.2) satisfies relation (1.7). If the first row of the matrix X(t, T) is designated by (rpl(t, T), ... , rpn(t, T)), then the first component y of the vector x in (1.7) is given by n
(3.5)
y(t)
=
j"f/DHY(T)]rpj(t, T)
t
+ IT rpn(t, s)g(s) ds,
where g is the function given in (3.1). The reason for such a simple expression is due to the fact that the vector h in (2.2) has all components zero except
92
ORDINARY DIFFERENTIAL EQUATIONS
the last which is g. Since the function c/>,,(t, s) together with its derivatives up through order n - I form the last column of the matrix X(t, s) it follows thatc/>,,(t, s) is the solution of the homogeneous equation (3.4) which for t = s vanishes together with all its derivatives up through order n - 2, and the derivative of order n - 1 with respect to t is equal to one. This is summarized in THEOREM 3.2. For any real number s in (- 00, + 00), let c/>(t, s) be the unique solution of the homogeneous equation (3.4) which satisfies the initial data
(3.6)
y(s)
= Dy(s) = ... = D"-2y(S) = 0,
D"-ly(S)
= l.
Then the unique solution of equation (3.1) which vanishes together with all derivatives up through order n - 1 at t = T is given by (;J.7)
y(t)
=
1 .,.
c/>(t, s)g(s) ds.
Another relation which is easily obtained from the general theory of Section 1 is the equation adjoint to equation (3.4). This is derived from the general definition of the adjoint equation given in Section 1. The adjoint to equation (3.2) with h = 0 is giv~n by w = -wA, W = (WI, ..• , w,,), or, equivalently, (3.S)
Dwl =a"w" { ~.~2 = -WI a,,-l W"
+
Dw"
= -W,,-l + al W" .
It is not obvious that equation (3.S) is equivalent to an nth order scalar equation of any type. However, if each of the functions ak, k = I, ... , n, has a sufficient number of continuous derivatives, this is the case. In fact, differentiation of the (k + l)th equation in (3.S) k-times with respect to t yields the equivalent set of equations (3.9)
If Z
Dwl =a"w" , Dk+lWk+l = _Dkwk + Dk(a,,_kw,,),
k = I, 2, ... , n - 1.
= w", then one easily concludes from (3.9) that
(3.10)
. D"z - D"-l(alz) + ... + (-I)"a"z = 0 . ;
ti.
Equation (3.10) is referred to as the adjoint to equation (3.4). If thla's are constant this is the same differential equation as the original one except for some changes in sign in the coefficients.
LINEAR SYSTEMS AND LINEARIZATION
m.4.
93
Linear Systems with Constant Coefficients
In this section, we consider the homogeneous equation (4.1)
x=Ax
and the nonhomogeneous equation (4.2)
x.=Ax+h(t)
where A is an n X n real or complex constant matrix and h is a continuous real or complex n-vector function on (-00, (0). The principal matrix solution P(t) of (4.1) at t =0 is such that each column of P(t) satisfies (4.1) and P(O) = I, the identity. The matrix P(t) satisfies the fonowing property: P(t + 8) = P(t)P(8) for all t, 8 in (-00, (0). In fact, for each fixed 8 in (~oo, (0), both P(t + 8) and P(t)P(8) satisfy (4.1) and for t = 0 these matrices coincide. Thus, the uniqueness theorem implies the result. This relation suggests that P(t) behaves as an exponential function. Therefore, we make the following Definition 4.1. The n X n matrix eAt, - 00 < t < 00, eO is defined as the principal matrix solution of (4.1) at t = O.
= I, the identity,
The matrix eAt satisfies the following properties: (i)
(ii) (iii)
eA(t+8) = eAte AB (eAt)-l = e- At d - eAt = Ae At = eAt A
dt
(iv) eAt (v) (vi)
=I
+ At + -2!1 A2t2 + ... + -n!1 Antn + ...
a general solution of (4.1) is eAtc where c is an arbitrary constant n-vector. If X(t), det X(O) =F 0 is an n X n matrix solution of (4.1), then eAt
=
X(t)X-l(O).
Property W was proved above and (ii) is a consequence of (i). Property (v) is Lemma 1.3 for this special case. (iii) follows from uniqueness and the observation that both Ae At and eAt A are matrix solutions of (4.1). This same argument proves (vi). Before proving (iv), it is necessary to make a few remarks about the meaning of a matrix power series. Iff(z) is a function of the complex variable z which is analytic in a neighborhood of z = 0, the power series f(A) is defined as the formal power series obtained by substituting for each term in the power series of f (z) the matrix A for the complex variable z. Of course, ·such
94
ORDINARY DIFFERENTIAL EQUATIONS
an expression will be meaningful if and only if for the particular matrix A each element of the power series matrix converges. One can actually use such a series as a definition for matrix functions f(A) when f(z) is an analytic function of the complex variable z. On the other hand, it should be noted that property (iv) above is not a definition of eAt but is going to be derived from the fact that eAt is the principal matrix solution of (3.1) at t = O. PROOF OF (IV). Since P(t) ~ eAt is the principal matrix solution of (4.1) at t = 0 it follows that P(t) must satisfy the integral equation
(4.3)
=1+
P(t)
fo t
AP(s) ds.
If one attempts to solve equation (4;.3) by the obvious method of successive approximations (4.4)
P(O)
=1,
+ f AP(k)(S) ds, t
P(k+1)(t) = 1
k =0,1,2, ... , o then one observes that the expression for P(k) is given by the formula
In the proof of the Picard-Lindel6f Theorem in Chapter I, it was shown that this sequence converged for It I ~ IX and IX small. We now wish to show that. the sequence of matrices P(k)(t) converge uniformly for all t in a compact set in (-00, (0). If we let P~f)(t) be the (ij)th element of the matrix P(k)(t), then there is a constant f3 > 0 independent of k such that 1~~)(t)1 ~ f3-1IP(k)(t)l. Thus, 1 1 :::;; 1 IXt 1X 2t 2 IXktk :::;; eat f31 p<,~)(t)1 'J 2! k! -'
+ +-
where for simplicity we have put
IX
=
+ ... + -
IAI. Furthermore,
_ P\~)(t)1 :::;; 1 IXk+1tk+1 f3I P\~+1)(t) tJ tJ -(k+l)! '
and the sequence (P~~)(t)) converges uniformly for all t in a compact set of (-00, (0). This shows that the power series in property (iv) is well defined and thus the sequence of matrices {P(k)(t)} converges uniformly on allcompact subsets of (-00, (0). From (4.4), it follows that the limit is the;iprincipal matrix .solution P(t) = eAt of (4.1). This proves property (iv). . In spite of the above apparent simplicity, the matrix eAt is a rather complicated individual. Many of the operations that are valid for the scalar
LINEAR SYSTEMS AND LINEARIZATION
95
exponential function are also true for the matrix function eAt. However, other operati<;ms do not behave in the 8ame way as for scalars since in general matrix multiplication is not commutative. The following two exercises indicate that caution must be exercised in operating with eAt. EXERCISE
4.1.
Prove that BeAt = eAtB for all t if and only if BA = AB.
EXERCISE 4.2. Prove that eAte Bt = e(A+B)t for all t if and only if BA =AB. With a general solution of equation (4;1) being given by eAtc, where c is a constant n-vector and the matrix eAt having the power series representation given in property (iv), one might ask if there is !J.nything left to discuss about linear equations with constant coefficients. Unfortunately, the previous notation is very misleading and we have not answered the following questions: In what precise sense does the function x(t) = eAtc where c is a constant n-vector behave as the scalar exponential function? What is an effective means for computing eAt? Both of the questions are very closely related and reduce to a detailed discussion of the eigenvalues and eigenvectors of a matrix. The general structure of the solutions of the differential equation (3.1) is determined completely from the sol~tion of an algebraic problem. This fact expresses the simplicity of linear systems of differential equations with constant coefficients and also explains why there is always a definite attempt in the applications to simulate models of physical systems by using equations with constant coefficients. A complex number A is called an eigenvalue (proper value, characteristic value) of an n X n matrix A if there exists a non-zero vector v such that
(4.5)
Av = AV
or
(A - AJ)V = O.
If A is an eigenvalue of the matrix A and v is any non-zero solution of equation (4.5), then v is called an eigenvector (proper value, characteristic vector) associated with the eigenvalue A. Hence, A is an eigenvalue of the matrix A if and only if A is a solution of the characteristic equation (4.6)
det(A -,\I) = O.
The characteristic equation is a polynomial of degree n in A and, therefore, has n solutions, not all of which may be distinct. On the other hand, if AI, ... , Ak are distinct eigenvalues of the matrix A and vI, ... , vk are corresponding eigenvectors, then v!, ... , v k are linearly independent. The following lemma is obvious. LEMMA 4.1. If A is an eigenvalue of the matrix A and v is an eigenvector associated with A, then the function x(t) = eAtv is a solution of the differential equation x = Ax.
96
ORDINARY DIFFERENTIAL EQUATIONS
LEMMA 4.2. If A is an n X n matrix with n-distinct eigenvalues AI, ... , An and vI, ... , v n are the corresponding eigenvectors, then a general solution of equation (4.1) is given by
V(t)c
=
c1v 1eA1t + ...
+ Cn vne Ant ,
where V(t) = (v 1eA1t , ... , vne Ant ) and c = (C1, ... , cn) is an arbitrary complex n-vector. Furthermore, eAt = V(t)V(O)-l. PROOF. If we let V(t) be defined as above, then Lemma 1.3 implies it is sufficient to show that V(t) is a fundamental matrix solution of equation (4.1). Since V(O) = (vI, ... , v n ) and the vectors VI, ... , vn are linearly independent, V(O) is non-singular. This fact, I-emma 4.1 and Corollary 1.1 imply that V(t) is a fundamental matrix solution of (4.1). A general solution is therefore V(t)c where c is an arbitrary n-vector. The last assertion of the lemma is precisely property (vi) above of eAt. If the matrix A has real elements then the eigenvalues will occur in complex conjugate pairs. This will imply that the corresponding eigenvectors can be chosen as complex conjugates~ Therefore, if only real solutions of (4.1) are desired, the constants C1, ... , Cn given in the general solution will need to satisfy some restriction of complex conjugacy. In fact, if Al and A2 are complex conjugates, then the eigenvectors vI, v2 can be' chosen as complex conjugates. In the general solution the constants C1 and C2 then may be chosen as complex conjugates to obtain a real solution of equation (4.1). We now describe the general procedure for obtaining eAt. To do this we need some elementary concepts from linear algebra. Let en be complex n-dimensional space and let 8 1 ,82 be linear subspaces of en. The subspaces 8 1 and 8 2 aI~ linearly independent if C1Y1 + C2 Y2 = 0, Y1 in 81, Y2 in 8 2, implies that C1 = C2 = O. The direct sum, 8i EEl 8 2 , of independent subspaces 8 1, 82 is a subspace 8 of en whose elements yare given by y = Y1 + Y2, Yl in 8 1 , Y2 in 8 2 . If A is an n X n matrix and 8 is a subspace of en, then 8 is invariant under A if for any y in 8 the vector Ay is in 8. If A is an n X n matrix the null space of A, denoted by N(A), is the set of all y in en such that Ay = O. If A is an eigenvalue of the n X n matrix A, r(A) is the least integer k such that N(A - ..\1)k+! = N(A - ..\1)k. The set N(A - ..\1)T(A) is a subspace of en and is called the generalized eigenspace of A corresponding to the eigenvalue A. This generalized eigenspace is denoted by the symbol MA(A). The dimension of M A(A) is equal to the algebraic multiplicity of the eigenvalue A; that is, the multiplicity of the eigenvalue as a zero of the characteristic polynomial' det(A - ..\1). We sayan eigenvalue A of A has ~imple elementary divisors if M A(A) = N(A - ..\1). The following result is basic and a proof can be found in any book on linear algebra. LEMMA 4.3.
If A is an n X n matrix and AI, ... , As are the distinct
LINEAR SYSTEMS AND LINEARIZATION
97
eigenvalues of A, then the corresponding generalized eigenspaces M A,(A), ... , M A,(A) are linearly independent, are invariant under the matrix A and = M A,(A) $ ... ® M A,(A).
en
Since the generalized eigenspaces M A, (A), ... , M A, (A) are linearly independent if ><1, ... , As are the distinct eigenvalues of A, it follows that any vector xo in can be represented uniquely as
en
8
xo=
(4.7)
L xO,j, j~l
Since the generalized eigenspaces M Aj(A) are invariant under the matrix A, any solution of the differential equation (4.1) with initial value in M Aj(A) will remain in M A, (A) for all values of t. This is immediately obvious since the tangent vector to the curve described by the solution of the differential equation in the phase space lies in the subspace M Aj(A). For any xO in and any complex number A,
en
eAtxO = e(A-Al)teAtxO
=
(f
k~O
(A - AI)k
~)xoeAt. k.
If Ais an eigenvalue of A and M A(A) is the corresponding generalized eigenspace, then xO in M A(A) implies that the above infinite series. is actually only a polynomial since (A - AI)kxO = 0 for all k ~ r(A). Therefore, for any xO in M A(A), eAtxo is given by
(4.8)
eAtxo =
[ r(~)-l L (A k~O
tk]
_AI)k"""l xOeAt. k.
In summary, if A is an eigenvalue of the matrix A and M A(A) is the corresponding generalized eigenspace, then any solution of the differential equation with its initial value in M A(A) must lie in M A(A) for all values of t and the solution of the differential equation is a polynomial in t with vector coefficients times eAt. Furthermore, this polynomial in t can be obtained by a direct evaluat'ion of the infinite series for e(A-Al)t. These remarks together with Lemma 4.3 and the decomposition (4.7) yield THEOREM 4.1.. If AI, ... , As are the distinct eigenvalues of the nx n matrix A and M A,(A), .. " M A,(A) are the corresponding generalized eigenspaces, then the solution of the initial value problem (4.9)
x=Ax,
x(O) =xo,
98
ORDINARY DIFFERENTIAL EQUATIONS
is given by (4.10)
X(t)
= L• ;=1
[' CAI)-1
L
(A
tk] -Aj 1)11:..,
xO,je}.Jt,
k.
k=O
where the vector xO.1 belongs to the generalized eigenspace M }.,(A) and is determined by the unique decomposition of the initial vector· xo according to equation (4.7). The specific manner in which one applies this result is as follows. The dimension of the generalized eigenspace M }.(A) of an eigenvalue of A is equal to the algebraic multiplicity of the eigenvalue A. For a given eigenvalue A of the matrix A one first obtains the null space of the matrix A - AI. If the dimension of this null space is not equal to the algebraic multiplicity of the eigenvalue A, one proceeds to calculate the null space of (A - AI)2. If the dimension of this subspace is equal to the algebraic multiplicity of the eigenvalue A, the subspace is M }.(A). This process is continued until a subspace is obtained which has the dimension of the algebraic multiplicity of the eigenvalue A. In this process, one also has obtained a basis for the generalized eigenspace M }.(A). Having done this for each eigenvalue A of the matrix A, one has a basis for On. In terms of this basis the element xo can be expanded uniquely, which determines the xO,j and therefore the solution x(t) by formula (4.10). The general solution of equation (4.1) is obtained by replacing x O, j ~n (4.10) by an arbitrary linear combination of the basis vectors for the generalized eigenspaces M }.1(A), j = 1, 2, ... , 8. EXERCISE 4.1. Using Theorem 4.1, find the general real solution ofthe equation i = Ax with
~];
(a)
A= [ 0
(b)
A=
[~.
~l
(c)
A=G
-~];
(d)
A~[~
-1 1 1
A~[;
1 3
(e)
-1
2
~l
-1
0]
-4 . -1
Theorem 4.1 gives a complete description ·of the general form. of the solution of a linear equation with (lonstant coefficients; namely. any solution
LINEAR SYSTEMS AND LINEARIZATION
99
of (4.1) is a sum of exponential terms with coefficients which are polynomials in t. The exponents of the exponential terms involve the eigenvalues of A. The degree of the corresponding polynomial in t is no more than one less than the dimension of the generalized eigenspace of the eigenvalue. A more complete description of the number of linearly independent solutions of (4.1) of the form p(t)e)'t, where p(t) is a polynomial of a given degree is given by the Jordan canonical form: For an n X n matrix A, there is a nonsingular matrix Q such that (4.11)
Q-IAQ
= diag(OI' ... , Op);
OJ =Aj 1
+ Rj;
~~[~
1 0 0 1
0 0
... ...
0 0
0 0 0 0
fl
and Aj ,j = 1, 2, ... , p, is an eigenvalue of A. The AI, ... , Ap are not necessarily distinct and the dimension nj of OJ is not necessarily equal to the r(Aj) mentioned above. If Q, A are related by (4.11), then (4.12)
Q-1eAtQ = e[diag(C" ... ,C,,)]t = diag(eC1t, ... , eC"t), 1
eCjt
t2
t n j-l
2!
(nj -I)! t nr 2
= e).jteRJt, e14t = 0 1 0
0
(nj -2)!
0
1
The representation (4.12) gives more specific information about eAt, but depicts the same general structure of the solutions of (4.1) as Theorem 4.1. A similarity transformation is a transformation which takes a matrix A into a matrix' Q-IAQ, Q nonsingular. Similarity transformations are very important in differential equations. In fact, if Q is nonsingular and x = Qy in (4.1), then iJ =Q-IAQy. The Jordan canonical form implies that stich transformations can be used to reduce a differential equation to a very simple form. THEOREM 4.2. (i) A necessary and sufficient condition that the system (4.1) be stable is that the eigenvalues of A have real parts ~ 0 and those with zero real parts have simple elementary divisors.
100
ORDINARY· DIFFERENTIAL EQUATIONS
(ii) A necessary and sufficient condition that the system (4.1) be asymptotically stable is that all eigenvalues of A have real parts < O. If this is the case, there exist positive constants K, ex such that (4.13)
t
~o.
PROOF. Since (4.1) is autonomous, stability implies uniform stability, asymptotic stability implies uniform asymptotic stability. Theorem. 2.1 implies that stability and boundedness of (4.1) are equivalent and asymptotic stability and exponential asymptotic stability are eguivalent. (i) If all solutions of (4.1) are bounded, then there can be no eigenvalue of A with positive real parts from Lemma 4.1. Furthermore, if there are eigenvalues with zero real part and M).(A) *N(A ->.1), then Theorem 4.1 implies there is a solution with a term teAt times a nonzero constant vector. Such a solution would not be bounded. Conversely, if all eigenvah,les have nonpositive real parts and M).(A) =N(A ->.1) for those eigenvalues with zero real parts, then the general representation theorem, Theorem 4.1, implies that no powers of t occUr except when multiplied by an eAt with Re>. < O. Therefore, the solutions are bounded. (ii) This part is proved in the same manner using Theorem 4.1 and Theorem 2.1. Since (eAt)-l = e- At , formula (1.7) implies the variation of constants formula for (4.2) is
x(t) =
(4.14)
eA(t~T)x(T) +
f
t
eA(t-s)h(s) ds.
or
As a particular illustration of this relation, consider the scalar equation ii
+ u =g(t).
This equation is equivalent to the system U=v,
If x
= (u,
v = -u + g(t). v),
A= [
~l f=[~l
0
-1
then eAt = [
C?s t -smt
sin t] cos t '
and the first component of (4.14) is
u(t)
= U(T) cos(t -
T)
+ U(T) sin(t -
+ f sin(t t
T)
or
a well known formula in elementary differential equations.
s)g(s) ds,
~
LINEAR SYSTEMS AND LINEAR:U;ATION
101
111.5. Two Dimensional Linear Autonomous Systems As an application of Theorem 4.1, we classify the different behaviors of the solutions of (5.1)
x=Ax,
det A =1=0.
where a, b, c, d are real constants. If AI, A2 designate the eigenvalues of A, there are many cases to consider. Case 1. AI, A2 real, A2 < AI. Let VI, v 2 designate unit eigenvectors of A associated with the eigenvalues AI, A2, respectively. A general real solution of (5.1) is (5.2)
where Cv c2 are arbitrary real constants. For a given Cv c2 ' ci +c~ > 0, the unit tangent vector to the orbit 'Y described by (5.2) approaches a vector parallel to v 1 as t~ ""-if C1 =1= 0 and ±v2 as t~-oo if c2 =1= O. Case la. Negative roots, A2 < Al < 0 (Stable node). All solutions approach zero as t ~ 00 and the above information on asymptotic directions allows one to sketch the orbits as in Fig. 5.1. The straight lines Ll and L2 are the lines that contain the eigenvector vI and v 2, respectively. The origin is stable and called a stable node.
Figure III.D.l
Stable node (Aa
<
At 0).
102
ORDINARY DIFFERENTIAL EQUATIONS
Figure III./j.2
Unstable node (0 < A2 < A).
Case lb. Positive roots, 0 < A2 < Al (Unstable node). This is similar to Case la except for reversing the arrows and changing the lines of tangency of the curves. See Fig. 5.2. The origin is unstable and called an unstable nope. Case 1e. One positive and one negative root, A2 < 0 < Al (Saddle point). From (5.2), those orbits which lie on L2 approach zero as t ~ 00 and those which lie on Ll approach zero ast ~ - 00. All other orbits are unbounded. The zero solution is unstable and the orbits are depicted in Fig. 5.3. The origin is called a saddle point.
+
Figure III./j.a Saddle point A2
< 0 < Al .
103
LINEAR SYSTEMS AND LINEARIZATION
Case 2. Complex roots. Since A is real, we have Al = IX + i{3, A2 = i{3, IX, {3 real, {3 > O. If vI and v 2 are chosen as complex conjugate, a general solution for the real solutions of (5.1) is IX -
x(t}
= cle(Cl+i/3)tv l + cle(Cl-i/3)tv l = 2Re(cle(ClH/3)tv1},
where c1 is an arbitrary complex number and b designates the complex conjugate of a. vector b. If VI = U + iv, u, v real, then u, v are linearly independent and if Cl = ae i6 where a, 8 are real, then a general real solution of (5.1) is (5.3)
x(t}
=
2aeClt[u cos({3t
+ 8} -
v sin({3t
+ 8}].
The expression (5.3) gives all of the essential properties of the solutions. If {3t + 8 = k1T, k an integer, then the orbit of the solution crosses the line U generated by u and if {3t 8 = (2k 1}1T/2, k an integer, it crosses the line V generated by v. The components of the solution curve in the direction u and v oscillate and are 1T/2 radians out of phase. Therefore, the orbit must resemble a spiral.
+
+
Case 2a. Purely imaginary roots (Center). If the eigenvalues of A are ±i{3, a general real solution of (5.1) is x(t)
= a[u cos({3t + 8} -
v sin({3t
+ 8}],
where a, 8 are abitrary real constants and u, v are as above. The orbits are closed curves and every solution is periodic of period 2rr/{3. These curves are ellipses with center at (0,0) (see Fig. 5.4). The origin is stable and called a center.
u
Figure 1II.5.4 Center A1
=::\2,
Re A1
= 0, 1m A "# O.
104
ORDINARY DIFFERENTIAL EQUATIONS
Complex roots with negative real parts (Stable focus). If it follows from (5.3) that all solutions approach zero as t ~ 00 and the orbits are spirals. The equilibrium point is called a stable focus (see Fig. 5.5). Case 2b.
at
< 0,
v
Figure l11.S.li Stable JOCWl.
Complex roots with positive real parts (Unstable focus). If then solutions approach zero as t ~ - 00 and the equilibrium point is called an unstable focus (see Fig. 5.6). Case 2c.
at
> 0,
v
Figure 111.6.6
UnBtableJOCtuJ.
Case 3. Both eigenvalues equal (Improper node). The eigenv:alUes are real. If there are two real linearly independent eigenvectors vI, ,,2 ~sociated with the eigenvalue>. (i.e., >. has simple elementary divisors), then a general
LINEAR SYSTEMS AND LINEARIZATION
105
solution is where Cl, C2 are arbitrary real constants. The unit tangent vector to the orbit of x is constant for any Cl, C2 • Therefore, all orbits are on straight lines passing through the origin (see Fig. 5.7).
Figure III. 5.7 Stable improper node.
If there is only one linearly independent eigenvector v l of ,\, then the general solution given by Theorem 4.1 is x(t)
=
(Cl
+ C2 t)eAtvl + C2 eAtv2,
where v2 is any vector independent of VI. By direct computation, one shows that the tangent to the orbit becomes parallel to VI as·t-+ +00 and t-+ -00. A typical situation is shown in Fig. 5.8.
Figure III.5.B Stable improper node.
106
m.G.
OBDINARY DIFFEBENTIAL EQUATIONS
The Saddle Point Property
In the previous section, we have given, a rather complete characterization of the behavior of the solutions of a two dimensional linear system with constant coefficients. It is of interest to study the conditions under which the behavior of the solutions near an equilibrium point of a particular type is not " chan,ged " by " small" perturbations of the right hand side of the differential equation: By considerin,g only linear perturbations Bx with IBI small, it is easy to see that the only types of equilibrium points for two dimensional linear systems which are preserved are the focus proper node and saddle point, On the other hand, the limiting behavior at 00 is insensitive to small perturbations for all types except the center. In this section, we discuss for linear systems of arbitrary order a special type of equilibrium point which is insensitive to perturbations in the differential equation. The classification of the equilibrium point is crude in the sense that it does not take into account the" fine" structure of the trajectories but only general asymptotic properties of the trajectories. The questions discussed here are treated in a much more general context in a later chapter, but, in spite of the duplication of effort, it seems appropriate to consider the special case at this time. The equilibrium point x = 0 of (4.1) is called a saddle point of type (k) of (4.1) if the eigenvalues of the matrix A have nonzero real parts with k ~ 0 eigenvalues with positive real parts. For n = 2, this definition of a saddle point does not coincide with the one in Section 5. In fact, an equilibrium .point which is completely stable (k = 0) as well as one which is completely unstable (k = 2) is referred to in this section as a saddle point. H x=O is a saddle point of (4.1) of type (k), the space 0" can be decomposed as 0" =0'+ €a~, 0'+ = 'IT + 0", ~ = 'IT_0", where 'IT+ , 'IT_are projection operators, 'IT+ + 'IT _ = I, O~ has dimension k, O'!.. has dimension n - k, 0'+, ~ are invariant under A and ther~ are constants (6.1)
K >0,01: >0 such that (6.2)
(a)
(b)
leAt'IT+xl leAt'IT_xl
~ ~
Keatl'IT+xI, Ke-atl'IT_·xl,
t~O, t~O,
for all x {n 0". Th~se relations are immediate from the observation thab there exists a nonsingular matrix U such that U-IAU = diag(A+ , A_) "fIire A+ is a k X k matrix whose eigenvalues have positive real parts and ·A_ is an (n - k) X (n - k) matrix whose eigenvalues have negative real parts. From
LINEAR SYSTEMS AND LINEARIZATION
107
Theorem 4.2, there are constants KI > 0, ex> 0 such that JeA+tj ~ Kle at , t ~ 0 and l.e A_t J ~Kle-at, t ~ O. The first relation is obtained by replacing t by -t in the equation u = A+ u. Let O't = {x in on such that x = Uy, Y = (u, 0), u a k-vector} and Or:.. = {x in on such that x = Uy, Y = (0, v), v an (n - k)-vector}. Since U is nonsingular q Et> Or:.. = on and this defines the projection operators 7T+, 7T- above. It is dear that (6.1) and (6.2) are satisfied. The subspace q defined in (6.1) is called the unstable manifold passing through zero and Or:.. is called the stable manifold passing through zero. The orbits of solutions on the unstable manifold approach zero as t ~ - 00 and the orbits of solutions on the stable manifold approach zero as t ~ 00. Furthermore, the construction of O't, Or:.. clearly shows that the only solutions of (4.1) whose orbits remain in a given neighborhood of the origin for all t ~ 0 (t ~ 0) must have their initial values on Or:.. (q). What type of behavior is expected when a linear system with x = 0 as a saddle point is subjected to perturbations which are small near x = O? The following example is instructive. Consider the second order system
Xl
=XI,
X2 = -X2+X~,
whose general solution is
where a, b are arbitrary constants. It is easily seen that Xl(t), X2(t) ~O as t ~ 00 if and only if a = 0 and b is arbitrary. Also, Xl(t), X2(t) ~O as t ~ - 00 if and only if b = a 3J4 and a is arbitrary. If b = a 3J4, then the corresponding orbit in the phase plane is x 2 = xiJ4. The phase plane portrait is shown in Fig. 6.1. Notice that the stable and unstable manifolds near x = 0 are essentially the same as the ones for the linear system. This example motivates the following definition of a saddle point for nonlinear equations. Suppose xo, is an equilibrium point of the equation (6.3)
X=Ax+f(x),
where f is continuous on On. We say Xo is a saddle point of type (k) of (6.3) if there is a bounded, open neighborhood V of Xo and two sets Uk, Sn-k. in on, Uk () Sn-k = {:to}, of dimensions k and n - k, respectively, Uk negatively invariant, Sn-k positively invariant with respect to (6.3), such that the orbit of any solution of (6.3) which remains in V for t ~ 0 (t ~ 0) must have initial value in U k(Sn-k) and the orbit of any solution with initial value in
108
ORDINARY DIFFEREN'lliIAL EQUATIONS
F~r~
Ill. 6.1
Uk(Sn-k) approaches Xo as t _ - 00 (+ 00). The set tJk is called the unstable manifold and Sn-k the 8table manifold. If x = 0 is a saddle point of (4.1) of type (k), we say that the 8addle point property is preserved relative to a given daB8 iF of functions if, for each f in iF, the system (6.3) has an equilibrium point Xo = xo(f) which is a saddle point of type (k). Our main interest 'centers around the preservation of the saddle point property when the family iF is a family of "small" perturbations and the equilibrium point Xo = xo(f) satisfies xo(O) = O. Iff has continuous first derivatives and we measure f by specifying that the norm of f in a bounded-, neighborhood V of x = 0 as
I I
If I =suplf(x)1 + sup zeY
zeY
of (x) -",- . uX
then there is no los"! in generality in assuming that f(x) = o(lx\) as Ixl-O. In fact,'there is a 8 >0 such that the matrix A + of(x)/ox as a function of x has k eigenvalues with positive real parts, n - k with negative real parts for Ixl < 8. From the implicit function theorem, the equation Ax + f(x) =0 has a unique solution Xo in the region Ixl < 8. The transformation x = Xo y yields the equation .
+
Of(xo)]. of(xo) y+f(xo+Y) -f(xo) -~ y
y= [ A+a;def
= By+g(y),
;,
where B has k eigenvalues with positive real parts, n - k with negative real parts and g(y) = o(ly\) as IYI-O.
109
LINEAR SYSTEMS AN:D LINEARIZATION
On the strength of this remark, we consider the preservation of the saddle point property for equation (6.3) for families of continuous functions f which at least satisfy f(x) = o(lxi) as Ixl ~o. LEMMA 6.1. Iff: On ~On is continuous, x = 0 is a saddle point of type (k) of (4.1), 7T+, 7T_ are the projection operators defined in (6.1), then, for any solution x(t) of (6.3) which exists and is bounded on [0, 00), there is an x_ in7T_On such that x(t) satisfies
x(t)
(6.4)
= eAtx_ +
foeA(t-S)7T_f(x(s)) ds + I t
0
e- AS7T+f(x(t
+ s)) as,
~
~ O. For any solution x(t) of (6.3) which exists and is bounded on (-00,0], there is an x+ in 7T+O n such that
for t
t
x(t) = eAtx+
(6.5)
0
+ So eA(t-S)7T+ f(x(s)) ds + L~ e- AS7T_f(x(t + s)) ds
for t ~ O. Conversely, any solution of (6.4) bounded on [0, 00) and any solution of (6.5) bounded on (-00,0] is a solution of (6.3). PROOF. Suppose x(t) is a solution of (6.3) which exists for t ~ 0 and Ix(t)1 ~ M for t ~ O. There is a constant L such that 17T+xl ~ Llxl for all x in On and, thus, 17T+X(t)1 ~ML for all t ~ O. Since f is continuous there is a constant N such that If(x(t))1 ~ N, t ~ 0. For any u in [0, 00), the solution x(t) satisfies
7T+X(t)
= eA(t-U)7T+x(a) +
f eA(t-S)7T+ f(x(s)) ds, t
tin [0, 00),
(J
since A7T+ = 7T+A, A7T- = 7T_A. Since the matrix A satisfies (6.2),
for t
~
a and, therefore, approaches zero as
u~
00. Also, for t
~
IS: eA(t-s)7T+ f(x(s)) dsl ~ K f e (t-s)I7T+f(x(s))1 ds CX
~ K LN
r
eCX(t-s)
ds
t
KLN
= - - [1 - eCX(t-U)] , (X
KLN
~--. (X
u,
110
ORDINARY DIFFERENTIAL EQUi\.TIONS
Therefore, the integral
I:
eA(t-S)7T+!(X(S)) ds exists. From the above integral
equation for 7T+X(t), this implies 7T+x(t)
=
Io
e- AS7T+!(X(t + s)) ds.
00
Since X(t)=7T+X(t)+7T_X(t), this relation and the variation of constants formula yield (6.4). Relation (6.5) is proved in a completely analogous manner. The last statement of the lemma is verified by direct computation to complete the proof. Notice that x_, x+ in (6.4), (6.5) are not the initial values of the solutions at t = 0, but are determined only after the solution is known. LEMMA 6.2. Suppose ex> 0, y> 0, K, L, M are nonnegative constants and u is a nonnegative bounded continuous solution of either the inequality
(6.6)
u(t)
~ Ke- cxt + L
t
00
f e-CX(t-s)u(s) ds + M f. o
e-YSu(t
+ s) ds,
t
~
0,
0
or the inequality (6.7)
u(t)
~ Ke rxt + L
fo
eCX(t-s)u(s) ds
+M
t
I
0
eYBu(t
+ s) ds,
t
~
O.
-00
If QdefL
(6.8)
t'
M
=-+ -y < 1, ex
then,. in either case, (6.9) PROOF. We only need to prove the lemma for u satisfying (6.6) since the transformation t -l> -t, s -l> -s reduces the discussion of (6.7) to (6.6). We first show that u(t) -l>0 as t -l> 00. If S = limt~oo u(t), then u bounded implies S is finite. If () satisfies f1 < () < 1, then S > 0 implies there is a tl ~ 0 such that u(t) ~ ()-lS for t ~ it. From (6.6), for t ~ it, we have
(6.10)
u(t)
~ Ke- cxt + Le-cxt (
eCXsu(s) ds
+ (~+ ~)()-lS.
Since the lim sup of the right hand side of (6.10) as t -l> 00 is < S, this is a contradiction. Therefore, S = 0 and u(t) -l>0 as t -l> 00. if If v(t) = sUPs;;;;t u(s), then u(t) -l> 0 as t -l> 00 implies for any t. in [0, (0), there is a it ~ t such that v(t) = v(s) = u(it) for t ~ s ~ it, v(s) < v(it) for
111
LINEAR SYSTEMS AND LINEARIZATION
s > h. Fro.m (6.6), this implies
v(t)
= u(h) ~ Ke- at • + L
Jo e-cx(t.-s)v(s) ds t
It
+ L
00
J e-CX(t,-s)v(s) ds + M J t
~ Ke- cxt ,
e-l' Sv(t'+ s) ds
0
+ L Je-CX(t,-s)v(s) ds + (3v(t), t
o
where (3 = L/cx.
+ M/y < 1. If z(t) = ecxtv(t), then h z(t) ~ (1 - (3)-lK + (1 - (3)-lL
~
t implies
t
J z(s) ds. o
Fro.m Gro.nwall's inequality, we o.btain z(t) ~ (1 - (3)-lK exp(1 - (3)-lLt and, thus, the estimate (6.9) in the lemma fo.r u(t). EXERCISE 6.1. SupPo.se a, b, care no.nnegative co.ntinuo.us functio.ns on [0, (0), u is a no.nnegative bo.unded co.ntinuo.us so.lutio.n o.f the inequality
u(t) ~ a(t) +
t
Jo b(t -
s)u(s) ds +
and a(t) ~O, b(t) ~O as t ~ 00, u(t)~O
as
t~ 00
J c(s)u(t + s) ds, 00
t ~ 0,
0
Iooo b(s) ds <
00,
Iooo c(s) ds <
00.
Pro.ve that
if
JOO b(s) ds + Joo c(s) ds < 1. o
0
If r is any subset in Cn which co.ntains zero., and Cn = 1T+Cn Ej;) 1T_cn, 1T+, 1T_ pro.jectio.n o.perato.rs with 1T+1T_ = 1T-1T+= 0, we say r is tangent to 1T_Cn at zero. if 11T+xl/I1T-xl ~O as x~O in r. Similarly, we say r is tangent to 1T+Cn if 11T-xl/I1T+xl ~O as x~O in r. THEOREM 6.1. SupPo.se TJ is a co.ntinuo.us, no.ndecreasing, no.nnegative functio.n o.n [0, (0) with TJ(O) =0 and let !l'ifz(TJ) designate the family o.f co.ntinuo.us functio.ns f: Cn ~ Cn such that
(6.11)
= 0,
'(a)
f(O)
(b)
If(x) -f(y)1 ~ TJ(u)lx-yl,
lxi, Iyl ~ u.
If x = 0 is a saddle Po.int o.f type (k) o.f (4.1), then the saddle Po.int pro.perty is preserved relative to. the family !l'e·fz(TJ). If, fo.r any fin !l'ifz(TJ), Uk = Uk(f), Sn-k =Sn-k(f) are the unstable and stable manifo.lds o.f the equilibrium Po.int x = 0 o.f (6.3), then Uk is tangent to. 1T+Cn at x = 0 and Sn-k is tangent to. 1T_Cn at x = 0, where 1T+Cn and 1T_Cn are the unstable
OltDINARY DIFFERENTIAL EQUATIONS
112
and stable manifolds of the saddle point x =0 of (4.1). Furthermore, there ar~ positive constants M, y such that (6.12)
(a)
Ix(t)1 ~ Me-vtlx(O)I,
(b)
Ix(t)1 ~Mevtlx(O)I,
t ~ 0, x(O) in 8 n- k , t~O,x(O)in Uk.
It is worthwhile to consider the following schematic representation Fig. 6.2 of Theorem 6.1. The picture for f = 0 is global whereas for f *- 0, it is local. In the diagram, we have indicated orbits of (6.3) other than the ones on Uk and 8 n -k. Actually, we do not prove that the orbits which do not intersect Uk or 8 n - k behave as shown but only assert that these orbits must leave a certain neighborhood of zero with increasing t.
Figure. III.6.2
PROOF OF THEOREM 6.1. From Lemma 6.1, for any solution x of (6.3) which is bounded on [0, (0), there is an x_ in 7T_On such that x(t) satisfies (6.4). We first discuss the existence of solutions of (6.4) on [0, (0) for any x_ in 7T _On sufficiently small. Since 7T_, 7T+are projections, there is a constant KI such that 17T+xl ~ Kllxl, 17T-xl ~ Kllxl for all x in On. Suppose K, IX are the constants given in (6.2) and 71(0'), 0' ~ 0, is the function given in (6.11). Choose so that
a
(6.13)
a
With this choice of and for any x_ in 7T_On with lx_I ~ a/2K, define 8) as the set of continuous functions x: [0, (0) _On such tht,t Ixl = sUPO;a;t
113
LINEAR SYSTEMS AND LINEARIZATION
+ f eA(t-s)1T_!(X(S» t
(6.14)
(Tx)(t) = eAtx_
o
ds +
f
0
e- A81T+!(X(t
+ S»
ds,
00
for t ~ O. Since x is in ~(x_, 8), it is easy to see that Tx is defined and continuous for t ~ 0 with [1T- Tx](O) = x_. From (6.2), (6.11), (6.13), we obtain
I(Tx)(t)1 ;:;; Ke-atlx_1 + f Ke-a(t-s) I1T_!(:»(s»1 ds + f .t
o
8
00
Ke-asl 1T+! (x(t
+ s»1 ds
0
8
<2+2 =8, fort
~ O. Thus, ITxl <8 and T: ~(x_, 8)-'.>-~(x-, 8). Furthermore, the same type of estimates yields'
I(Tx)(t) -
2KKl 1 (Ty)(t)1 ;:;; -1X-7}(8)lx -YI ;:;; 21x -YI,
for t ~ O. Thus T is a contraction on ~(x_, 8) and there is a unique fixed point x*(', x_) in ~(x_, 8) and this fixed point satisfies (6.4). Using the same estimates as above, one shows that the function x*(·, x_) is continuous in x_ and x*(·, 0) =0. However, more precise estimates of the dependence of x*(·, x_) on x_ are needed. If we let x* = x*(·, x_), x* = x*(·, x_), then, from (6.4),
+ KKl7}(8) f e-a(t-s)lx*(s) t
Ix*(t) - x*(t)1 ;:;; Ke-atJx_ - x_)1
+ KKl7}(8)
t' o
o
e-aslx*(t
x*(s)1 ds
+ s) -x*(t + s»1 ds
for t ~ O. We may now apply Lemma 6.2 to this relation. In Lemma 6.2, let y = IX, M = L = KKl7}(8). If u(t) = Ix*(t) - x*(t)l, 8 satisfies (6.13) and appropriate identification of constants are made in Lemma 6.2, then (6.15)
Ix*(t, x_) - x*(t, x-)I ;:;; 2K( exp -
~) Ix- -
x-I,
t ~ O.
Smce x*(·, 0) = 0, relation (6.15) implies these solutions satisfy a relation of the form (6.12a) and approach zero exponentially at t -'.>- 00. Let B6/2K denote the open ball of radius 8/2K in On with center at the origin. Let S1l-k designate the initial values of all those solutions of (6.3) which
114
ORDINARY DIFFERENTIAL EQUATIONS
remain inside B6 for t ~ 0 and have 1T-X(O) in B 6/ 2K . From the above proof, S:_k ={x: x = x*(O, x_), x_ in (1T_On) ('\ B 6 / 2K }. Let g(x_) = x*(O, x_), x_ in (1T_on) ('\ B 6 /2K. The function g is a continuous map of (1T_On) ('\ B6/2K onto S=-k and is given by o (6.16) g(x_) = x_ e-A81T+ f(x*(s, x_» ds.
+I
eo
From (6.2), (6.11), (6.13), (6.15), we have
Ig(x-) - g(x-) I ~
Ix- -x-I - Ioeo Ke-asKl7](8)lx*(s, x_) -
x*(s,
x_»1 ds
~ Ix- -x-I [1 1 ~ "2 Ix- -x-I,
for all x_, x- in (1T_On) ('\ B 6 /2K. Therefore g is one-to-one. Since g-l = 1T_ is continuous it follows that g is a homeomorphism. This shows that S:_k is homeomorphic to the open unit ball in On-k and, in particular, has dimension n - k. However, S:_k may not be positively invariant. If we' extend S:_k to a set Sn-k by adding to it all of the positive orbits of solutions with initial values in S:_k' then Sn_k is positively invariant and also home?morphic to the open unit ball in On-k from the uniqueness of solutiohs of the equation. The set Sn-k coincides with S:_k when x in Sn~k implies 11T-xl < 8/2K. From (6.14), (6.15) and the fact that x*(·, 0) = 0, we also obtain
11T+X*(O, x_)1
~ KKI foeo e- as7](lx*(s, x_>I>lx*(s, x_)I.ds ~ KKI Ioeo e-cX87](2Klx_D2Klx~1 ds 2K2Kl
= -<X- 7](2Klx-Dlx-l· Consequently 11T+X*(O, x_)I/lx_I-+O as Ix-I-+O in Sn-k which shows that Sn-k is tangent to 1T_On at x = O. Using relation (6.5), one constructs the set Uk in a completely analogous manner. This completes the proof of the theorem. In the proo(ofTheorem 6.1, it was actually shown that the m~ping g taking 1T_On ('\ B6/2K into S:_k is Lipschitz continuous [see relati9hs (6.15) and (6.16)]. Since the solutions of (6.3) also depend Lipschitz continuously on the initial data if (6.11) is satisfieQ., it follows that the stable m,anifold
LINEAR SYSTEMS AND
115
~EARIZA.TION
8 n -" and also the unstable manifold Uk are Lipschitz continuous; that is, 8 n-,,(U,,) is homeomorphic to the unit ball in on-k(Ok) by a mapping which is Lipschitz continuous. It is also clear from the proof of Theorem 6.1 that the Lipschitz condition of the type specified in (6.11b) was unnecessary. One could have assumed only that If(x) - f(y)j ~ ylx - yj,
where y satisfies (6.13) with 7](8) replaced by y. Of coUrse, the assertion of tangency in the theorem may not hold in this more general situation. An even weaker version of Theorem 6.1 can be proved for functions which may not be lipschitzian but satisfy
I-'Ixl for x in some neighborhood V of x = and I-' satisfies (6.13) with 1](a) replaced by 1-'. In fact, let 8 be chosen such that Ixl < 8 implies x in V and let ~(x_, 8) If(x)1 ~
(6.17)
°
be defined as before as all continuous functions taking [0, (0) into On which are bounded by 8, but use the topology of uniform convergence on compact subsets of [0, (0). If the mapping T is defined as in (6.14), one can show that T: ~(x_, 8) ~ ~(x_, 8) in the same manner as before. For any e > 0, choose T so large that t
j(Tx)(t) - (Ty)(t)1
~ KKI SO e-a(t;-S)lf(x(s))
- f(y(s))1 ds
+ KKI s: e-aslf(x(t + s)) -
f(y(t
+ s))1 ds
+ e/2 for any x, y in ~(x_, 8). Therefore, given any compact set G in [0, (0), one can always choose a 8> such that Ix(t) - y(t)1 < 8 on G implies I(Tx)(t) -(Ty)(t)1 < e on G. This implies T is continuous on ~(x_, 8) in the topology of uniform convergence on compact sets. Since (Tx)(t) is a solution of the equation z = Az + f(x(t)), and ITxl ~ 8 for all x in ~(x_, 8), it follows that Id(Tx)(t)jdtl is uniformly bounded and T is a completely continuous map of ~(x_, 8) into ~(x_, 8). Using the Schauder-Tychonov theorem, one can assert the existence of a fixed point x*( ., x_) of T in ~(x_, 8). Furthermore, since x*(·, x_) satisfies (6.4), it follows that
°
Ix*(t, x-)l
~ Ke-atlx_1
t
+ KKll-' So e-a(t-s)lx*(s, x_)1 ds
+ KKll-' fo'l' e-aslx*(t + 8, x_)1 ds.
116
ORDINARY DIFFERENTIAL EQUATIONS
Again, using Lemma 6.2, one obtains (6.18)
t
~
O.
Also, exactly as in the proof of Theorem 6.1, one obtains (6.19)
* 2K2Kl 117+X (0, x_)1 ~ - - fLlx-l· (X
Thus, all solutions x(t) of (6.3) such that Ix(t)1 ~ S, t ~ 0 and 117-X(0)1 ~ Sj2K approach zero as t ---+ 00 exponentially and in fact satisfy (6.19) for an appropriate x_. If we designate S* as the set of initial values of such solutions and extend S* to a set S by adding to it all of the positive orbits of solutions with initial values in S*, then it is reasonable to call S a stable manifold of (6.3). Iff satisfies the stronger condition (6.20)
f(x)
=
o(lxl> as
Ixl---+O,
then there is an.7](u) continuous for U ~ 0, 7](0) = 0, such that If(x)1 for Ixl ~ u. The estimate (6.19) can then be improved to (6.21)
~
7](u)lxl
2K2Kl j17+X*(0, x_)1 ~ - - 7](2Klx-l>lx-l· (X
Relation (6.21) shows that S is tangent to 17_On at x = O. In the same manner, one obtains an unstable manifold U which is tangent to 17+On at x = O. One will still have the property S fl U = {O}, but cannot assert that S has dimension n - k and U has dimension k. Iff satisfies the condition (6.17), the estimate (6.19) shows that the stable manifold must lie in a region containing 17_0, the region being bounded by two planes which approach 17-0 as fL---+O. The same remark applies to the unstable manifold. If f satisfies (6.20), then the tangency of the stable and unstable manifolds of (6.3) to the stable and unstable manifolds of (4.1) at x = 0 implies the following: For any neighborhood N of (17_On) intersected with the ball of radius one with center at the origin, there is a neighborhood V of x = 0 such that the stable manifold of (6.3) relative to the neighborhood V lies in N. The same remark applies to the unstable manifold. An important consequence of this remark is the following corollary, the first part of which is a special case of Theorem 2.4. 6.1. "Suppose f: On ---+.On is continuous and f(x) = o4{xl> as Ixl---+O. If the eigenvalues of A have negative real parts, then the'$~lution x = 0 of (6.3) is uniformly asymptotically stable. If one eigenvalue of A has a positive real part, then the solution x=o is unstable.
COROLLARY
117
LINEAR SYSTEMS AND LINEARIZATION
The next example due to C. Olech shows that the dimension of the stable and unst:tble manifolds may increase under perturbations which are continuous and o(ixi) as ixi-')-O, but which ii.re not differentiable at x = 0. Let ~(x), -00 < x
°
°
+
°
(6.22) X2
= X2 -
I/J(~(XI)' X2 - ~(XI))WXI)
+ xlf(XI)].
Under the above hypotheses on ~, I/J, we see that the perturbation is o(Jxi) as ixi-')-O, x = (Xl, X2). Also, X2 = 0, Xl e-ta, and X2 = ~(XI), Xl = e-ta, where a is arbitrary, are solutions of (6.22). These solutions approach zero as t -,)- 00, the corresponding orbits belong to the stable manifold of X = 0, and obviously these orbits are distinct. The fan near X = consisting of the orbits of the solutions whose initial values are between the curve X2 = and X2 = ~(XI) in Fig. 6.3 must also belong to the stable manifold of the solution X = of (6.22). In fact, one easily shows that the perturbation being o(ixi) as ixi-')-O implies there is a cone enclosing the positive xl-axis so that near x = the tangent vector to any orbit cannot be perpendicular to the xl-axis. This immediately yields the result.
=
°
Figure III.6.3
111.7 J Linear Periodic Systems
I
Consider the homeogeneous linear periodic system (7.1)
X =A(t)x,
A(t
+ T) = A(t),
°
° °
118
ORDINARY DIFFERENTIAL EQUATIONS
where A(t) is a continuous l n X n real or complex matrix function of t. Our first objective in this section is to give a complete characterization of the general structure of the solutions of (7.1). LEMMA 7.1. If 0 is an n X n matrix with det 0:;6 0, then there is a matrix B such that 0 = eB . PROOF. If P is nonsingular and there is a matrix B such that 0 = eB , then p-lOP =eP-'BP. Therefore, we may assume that 0 is in Jordan canonical form; that is, 0= diag(Ol, ... , Op), OJ =Aj1+ Rf ,
R,
~ [~ ~ ~ ~I :::
By hypothesis, each Aj:;6 0, j = 1,2; ... ; p. To prove the lemma, it is sufficient to show that each OJ can be written as OJ = eBJ. Therefore, we drop the subscripts and suppose 0 = AI R where A :;6 0 and R is a matrix of the same type as the Rf above; in particular, Rk = 0 for all k ~ m, for some integer m. Since A:;6 0, (j = A(1 + RIA). Let
+
B ~ (log A)1 +8, m-l
(-R)f
8=-j~1 )N' The matrix 8 is the matrix power series obtained by taking the power series for log(l+t) near t=O and replacing t by RIA. Since Rk=O for k~m, there is no problem of convergence. On the other hand, one can show directly by substitution in the power series for eB that 0 = eB . The lemma 'is proved. EXERCISE 7.1. For any real matrix D, det D:;6 0, show there is a real matrix B such that eB = D2. If 0 is a real matrix in Lemma 7.1 and there is a real matrix B such that eB = 0, must there exist a real matrix D such that 0=D2?
~~L_T~H_E_O~R_E_M~7_.1_'....Io(_Fl_o_)Q_lu_e.. t)..1 Every fundamental matrix solutionX(t) of (7.1)hastheform d:-C1:\:MI,\::di) J AU:+i\:::Mi:/ Vt-
r>o
X(t) = P(t)e Bt
(7.2)
where P(t), Bare n
X
n matrices, P(t
+ T) =
P(t) for all t, and Bisaco!lstant.
t
This assumption is for simplicity only. The theory is valid for A(t) which are periodic and Lebesgue integrable if (7.1) is required to be satisfied except for a set of Lebesgue measure zero. No changes in proofs are required. 1
119
LINEAR SYSTEMS AND LINEARIZATION
Suppose X(t) is a fundamental matrix solution of (7.1). Then is also a fundamental matrix solution since A(t) is periodic of period T. Therefore, there is a nonsingular matrix C such that PROOF.
X(t
+ T)
X(t
+ T) = X(t)C.
From Lemma 7.1, there is a matrix B such that C = eBT . For this matrix B, let p(t) = X(t)e- Bt . Then P(t
+ T) = X(t + T)e-B(t+T) = X(t)eBTe-B(t+T) = P(t),
and the theorem is proved. COROLLARY 7.1. There exists a nonsingular periodic transformation of variables which transforms (7.1) into an equation with constant coefficients. PROOF. Suppose P(t), B are defined by (7.2) and let x The equation for y is
'!i = Since P
= Xe- Bt , it
foilows that
=
P(t)y in (7.1).
P-l(AP - p)y.
P = AP -
P B and this proves the result.
EXERCISE 7.2. Prove that B in the representation (7.2) can always be chosen to be a real matrix if A(t) in (7.1) is real and' it is only required that
P(t
+ 2T) = P(t).
Theorem 7.1 states that any solution of (7.1) is a linear combination of functions of the form eAtp(t) where p(t) is a polynomial in t with coefficients which are periodic in t of the same period as the period of the coefficients in the differential e uation. monodrom matrix of (7.1) is a nonsingular matrix C associated with a fundamental matrix solution X ,t of (7.1) throu h the relation X t T = X t C. The eigenvalues p of a monodromy matrix are called th characteristic multi liers of (7.1) and any A such that p = eAT is called a characteristic ex anent if (7.1). Notice that the characteristic exponents are not uniquely defined, but the multipliers are. The real parts of the characteristic exponents are uniquely defined and we can always choose the ex onents A as the eigenvalues of B, where B is any matrix so that C = eBT . The characteristic multipliers do pot depend upon the particular monodromy matrix chosen; that is, the particular fundamental solution used to define the monodromy matrix. In fact, if X(t) is a fundamental matrix solution, X(t T) = X(t)C and Y(t) is any other fundamental matrix solution, then there is a nonsingular matrix D such that Y(t) = X(t)D. Therefore, Y(t T) = X(t T)D = X(t)CD = Y(t)D-ICD and the monodromy matrix for Y(t) is D-ICD. On the other hand, matrices which are similar have the same eigenvalues. We shall usually use the term monodromy matrix for X(T) where X(t), X(O) = I is a fundamental matrix solution of (7.1).
+
+
+
120
ORDINARY DIFFERENTIAL EQUATIONS
I LEMMA 7.2] A complex number A is a characteristic exponent of '(7.1) if and only if there is a nontrivial solution of (7.1) of the form eAtp(t) where p(t T) = p(t). In particular, there is a periodic solution of (7.1) of period T (or 2Tbut not T) if and only if there is multiplier = + 1 (or -1)
+
PROOF. If eAtp(t),p(t + T) = p(t) ~ 0, satisfies (7.1), Theorem 7.limplies there is an Xo ~ 0 such that eAtp(t) = P(t)eBtxo. Thus, P(t)eBt[e BT - eAT1]xo =0 and, thus, det(e BT -e AT1) =0. Conversely, if there is a A such that det(e BT - eAT1) = 0, then choose Xo ~ 0 such that (e BT - eAT1)xo = O. One can choose the representation (7.2) so that A is actually an eigenvalue of B. Then eBtxo = eAtxo for all t and P(t)eBtxo = P(t)xo eAt is the desired solution. The last assertion is obvious. LEMMA 7.3, If Pi = eAJT, j = 1,2, ... , n, are the characteristic multipliers of (7.1), then
ni=1Pj =exp(JoT tr A(s) ds),
(7.3)
fAi== .: fT tr A(s) ds (mod 27Ti).
;=1
PROOF.
ToT
Suppose 0 is the monodromy matrix for the matrix solution
X(t), X(O) = 1, of (7.1). Then Lemma 1.5 implies
det 0 = det X(T) = exp(JoT tr A(s)
ds)).
The statemeu.ts of the lemma now follow immediately from the definitions of characteristic multipliers and exponents. I T"'En... .,.,... 7 ') I .(i) A necessary and sufficient condition that the system (7.1) be uniformly stable is that the characteristic multipliers of (7.1) have modulii ~1 (the characteristic exponents have real parts ~O) and the ones with modulii =1 (the characteristic exponents with real parts =0) have simple elementary divisors. (ii) A necessary and sufficient condition that the system (7.1) ~e uniformly asymptotically stable is that all characteristic multipliers of (7.1) have modulii <1 (all characteristic exponents have real parts < 0). If this is the case and X(t) is a matrix solution of (7.1), there exist'K>O, oc>O Sl;lch that IX(t)X-l(S)1 ~ Ke-a(t-B), t ~ s. PROOF. The' proof is essentially the same as the proof of Theo,m 4.2 if one uses Coroll 7.1. , ;:7 At first glance, it might appear that linear periodic equations share the same simplicity as linear equations with constant coefficients. However, there
121
LINEAR SYSTEMS AND LINEARIZATION
is a very important difference-the characteristic exponents are defined only after the solutions of (7.1) are known and there is no obvious relation between the characteristic ex onents and the matrix At. The following example illustrates that the eigenvalues of the matrix A(t) cannot be used to determine the asymptotic behavior of the solutions. Example 7.1.
+ ~ cos 2 t
1-
3 . -1 - 2' sm t cos t
-1
=-1 (7.4)
A(t)
If
(Markus and Yamabe [1]).
=
~ c0s t sin t], 3
[
+ 2' sin 2 t
then the eigenvalues Al(t), A2(t) of A(t) are Al(t) = [ -1 +iJ7]j4, A2(t) = X1(t) and, in particular,the real parts of the eigenvalues have negative real parts. On the other hand, one can verify directly that the vect.or (-cos t, sin t) exp
G)
is a solution of (7.1) with A(t) given in (7.4) and this solution is unbounded as t-+ 00. One of the characteristic multipliers is en:. The other multiplier is e- 2 n: since 7.3 im lies that the roduct of the multi liers is e-n:. The proplem of determining the characteristic multipliers or exponents of linear periodic systems is an extremely difficult. one. Except for scalar second order equations and, more generally, Hamiltonian and canonical systems, very little is known at all. Even the equations of the form x = Ax e2 are extremely difficult and exhibit very striking behavior. This shall be illustrated in a later chapter by means of examples.
+
m.s·1 Hill's Equation I In this section, we give a rather detailed discussion of the stability properties of the Hill equation, (8.1)
I y+
(a
+ cp(t))y = 0,
cp(t + 17) = cp(t),
I
where a is constant and cp is assumed real and continuous. Actually, there is no change in the theory if cp is integrable and bounded, but in such a situation, we have to sa the e uation is satisfied almost ever h~re. The ultimate goal is to characterize the values of the arameter a for which there is stability of the equation The previous section implies this is
122
ORDINARY DIFFERENTIAL EQUATIONS
equivalent to determining the qualitative structure of the characteristic multipliers of (8.1) as a function of a. Equation (8.1) is equivalent to the system (8.2)
:i: = [0(/1)
+ A(t)]x,
A(t) =
O(a) =
[-~(t) ~], [-~ ~l
Suppose
(8.3)
X(t)def [YI(t) li1(t)
Ys(t)] ys(t) ~
X(O) =1,
is the principal ma.trix solution of l8.2) at t - OJThe characteristic multipliers of (8.2) are the eigenvalues of the matrix X(1T); that is, the roots of the equation det(X(1T) -
Dl) =
o.
Since tr [O(a) + A(t)] = 0, (7.3) implies that the characteristic multipliers are the roots of the equation (8.4)
r-------------------~
ps -2B(a)p+ 1 =0,
= tr X(1T) = YI(1T) + YS(1T)
2B(a)
....:~~.:I.I.I..u-=~~..,::;:,~~:;:...::~~.L.refined above. From Chapter I, the
function B a is entite in the arameter a. In view of Lemma 7.2 and the fact that the multipliers PI, P2 of (8.4) satisfYPIP2 = 1, the equation (8.1) can be stable only if Ipil = Ip21 = 1. The next lemma. shows this can never be the case if a is Qomplex. LEMMA 8.1. If a is complex, 1m a =F 0, equation (8.1) is unstable and no characteristic multiplier of (8.4) has modulus 1. PROOF. AB remarked above, equation (8.1) can be stable only if both characteristic multipliers have modulii one. Therefore, the lemma is proved if we show no characteristic multiplier has modulus one if a is complex. If a characteristic multiplier has modulus one, then Lemma 7.2 implies there must be a solution of (8.1) of the form eUtp(t) where A is real and p(t + 1T) = p(t) ¢ o. If eUtp(t) = u + iv, a = IX + i{J, u, v, IX, {J, real, then
+ [IX + cfo(t)]u = ii +[IX + cfo(t)]v =
it
{Jv, -{Ju.
" This implies itv - iiu =
(J(u 2 + V2)
LlNEAR SYSTEMS AND LINEARIZATION
123
and, thus, upon integrating,
It [u 2(s) + v2(s)l ds = f3 ItIp(s)12 ds def = f3 o
u(t)v(t) -ti(t)u(t)
+ c,
0
where c is a constant. Since the right hand side is bounded for all t and p is periodic of period 'TT this gives a contradiction unless either f3 = 0 or p = O. This proves the lemma. LEMMA 8.2. The equation B2(a) = 1 can have only real solutions. B(a) = 1 (or -1) is equivalent to the statement that there is a periodic solution of (8.1) of period 'TT (or 2'TT). If a is real, then B2(a) < 1 implies all solutions of (8.1) are bounded and quasiperiodic on (-00, (0). If a is real and B2(a) > 1, there is an unbounded solution of (8.1).
+J
The roots PI, P2 of (8.4) are PI = B(a) B2(a) -1, P2 = B(a) If B2(a) = 1, then PI = P2 = B(a) = ±1 and Lemma 7.2 yields the statement concerning periodic solutions. This implies B2(a) = 1 can have only real solutions from Lemma 8.1. If a is real, then B(a) is real and B2(a) < 1 is equivalent to PI = P2, PI =F P2, IPII = 1. Lemma 7.2 implies the existence of two linearly independent solutions which are quasiperiodic. Thus, every solution is quasiperiodic. If B2(a) > 1, then one characteristic multiplier is > 1 and one is < 1. Theorem 7.2 completes the proof of the lemma. Lemmas 8.1 and 8.2 imply that system (8.1) can never be asymptotically stable. In the remainder of the discussion, the parameter a is taken to be real. An interval 1 will be called an a-interval of stability (instability) of (8.1) iffor each a in 1, equation (8.1) is stable (unstable). Lemma 8.2 implies that the transition from an a-interval of stability to an a-interval of instability can only occur at those values of a for which B2(a) =1. Therefore, the basic problem is to find those values of a for which B2(a) = 1 and discuss the behavior of the function B(a) in a neighborhood of such values. Theorem 8.1 below gives a qualitative description of the manner in which the a-stability and a-instability intervals of (8.1) are situated on the real line. The following lemmas lead to a proof of this theorem. PROOF.
- J B2(a) -1.
LEMMA 8.3. If a + t/>(t) ~ 0 for all t in [0, 'TT], then there is an unbounded solution of (8.l). If, in addition, a + t/>(t) is not identically zero, then B(a) > 1. As before, let YI(t) be the solution of (8.1) with YI(O) = 1, 1h(0) = 0, and let I/J(t) = -(a t/>(t)) ~ O. From (8.1), PROOF.
(8.5)
+
(rh(t))2 = 2
~ l/J(s)YI(s)'!h(s) ds
for all t~ O. Since YI(O) = 1, ih(O) = I/J(O)YI(O) ~ O. Thus, rh(t) ~ 0 on an interval 0 ~ t ~ "I. If rh(t) = 0 for all t ~ 0, then YI(t) = 1 is a solution of
124
ORDINARY DIFFERENTIAL EQUATIONS
(S.I) which implies a + ,p(t) == 0 for all t and conversely. In this case, there are unbounded solutions since y(t) = t is a solution. Suppose a + ,p(t) is not identically zero and let 7J be such that 1h(t) = 0 for 0 ~ t ~ 7J and, for any T > 0, there is a t in (7J, 7J + T) such that 1h(t) =1= O. Such an 7J always exists. One can choose T so small that Yl(S) > 0 for 0 ~ S ~ 7J+ T. Since I/I(s) ~ 0, it follows from (S.5)that 1h(t) > 0 on (7J, 7J + T). The right hand side of (S.5) is a nondecreasing function of t and, therefore, 1h(t) > 0 for all t > 7J and 1h(t) is monotone increasing for t ~ 7J. Finally Yl(t) = I
+ 1o 1it(S) da ~ 1+1
1h(s) da
~+T
~ I + 1it(7J + T)(t -7J -
T)
fort ~ 7J + T, T > O. Since 1h(7J + T) > 0, this shows Yl(t) is unbounded and proves the first part of the lemma. To prove the second part of the lemma, we first recall the fact that we have just proved that lit(t) ~ 0 for all t and, also, Yl(1T) > 1. Now, consider the solution Y2(t) of (S.I) with Y2(0) = 0, Y2(0) =1. The function Y2(t) will satisfy , (Y2(t»2
=
1
+ 211/1(S)Y2(S)Y2(S) da. o
Since I/I(s) ~O and is not identically zero, it follows that Y2(1T) > 1. Therefore, 2B(a) = Yl(1T) Y2(1T) >2 and the lemma is proved. Lemma S.3 shows that a +,p(t) must take on some positive values if the solutions of (S.I) are to remain bounded. Since ,p is bounded, there is an a* such that a*+,p(t) <0. Thus B(a) >1 and the equation (S.I) is unstable" for -00
+
+
LEMMA S.4. The functions B(a) -I, B(a) 1 have an infinite number of' real zeros {aD < al < a2"'}, {at < a: < ...}, respectively, and ak , a: approach +00 as k_oo. PROOF. We actually show that B(a) is an entire function of order 1/2, that is, B(a) is an entire function and, for any e > 0, IB(a)1 exp[ial~+1/2] _ 0 aslal_oo and, fore
. X(t)
= 1+
1 o
[O(a) +A(s)]X(s) da.
125
LINEAR SYSTEMS AND LINEARIZATION
Denote the right hand side of this integral equation by (TX)(t) and define the sequence {X(k)} of functions by X(O) =1, X(k+l) = TX(k), k =0,1, .... Each function X(k) is an entire function of a. Suppose a belongs to a compact set V and t is in a compact set U. Exactly as in the proof of the power serie!. representation of eAt (see Section 4), one shows the sequence X(k)(t) = X(k)(t, a) converges uniformly to X(t) =X(t, a) for t in U, a in V. Therefore, X(t, a) is an entire function of a and B(a) is an entire function of a. To show the order relation, let w 2 = a and consider the variation of constants formula for (8.2) treating A(t)x as the forcing function. If Yl, Y2 are defined as above, then t
(8.6)
fo w-
Yl(t, a)
= cos wt -
Y2(t, a)
= w- 1 sin wt -
1
sin w(t - s)cp(S)Yl(S, a) ds,
fo w- sin t
Since Iw- 1 sin wtl
~elwlt,
1
Icos wtl ~elwlt, t
w(t-s)cp(S)Y2(S, a) ds. ~
0, if Iwl
t
IYJ(t, a)1
for 0
~ t ~
1T,
~ e1w1t +K fo e'W'(t-B)IYJ(s, a)1 ds,
j
~
1, it follows that
= 1, 2,
where K is a bound on cp. If z(t) =e-,w,tIYi(t, a)l, then z(t)
~
t
1
+ K fo z(s) ds.
Gronwall's inequality implies z(t) ~ eKt ~ eK1T., 0 ~ t ~ 1T, and, therefore, 1T, j =1, 2. Since Y2(t, a) satisfies
IYi(t, a)1 ~ eK1T.elwlt, 0 ~ t ~
t
Y2(t, a) =cos wt -
fo cos w{t -
s)cp(S)Y2(S, a) ds,
we have the existence of a constant L such that IY2(t, a)1 ~Lelwlt, 0 ~ t ~1T. Therefore, the order of B(a) is ~ 1/2. To prove the order is ~ 1/2, it is no loss in generality to assume that cp(t) ~ -1 for all t. In fact, we can always replace cp by cp -M -1, a by a + M + 1, where M is a bound on Icp(t)l. Also, choose a < 0 so that = ilL with IL > o. For a < 0, cp(t) ~ -1, it was shown in the proof of Lemma 8.3 that Yl(t, a) ::>;1, Y2(t, a) > 1 for t > o. This fact, -cp(t) ~ 1 and (8.6) imply that
Ja
Yl(t, a)
fo cosh ~ f cosh o ~
t
IL(t -S)Yl(S, a) ds
t
sinh ILt
=---
IL(t - s) ds
126
ORDINARY DIFFERENTIAL EQUATIONS
Thus, Yl(t, a) ~ (e llt -1)/21-'2 + 1 for all t in [0, 7T]. Since Y2(7T, a) ~ 0, it follows that B(a) is of order at least 1/2. This completes· the proof of the lemma. LEMMA 8.5. If b is a root of B(a) = 1 such that B'(o) ~dB(b)/db ::;; 0 then B;(a) < 0 for b < a< c* provided B(a) > -Ion this interval. lib* is a root of B(a) = -1 such that B'(b*)~ 0, then B'(a) >0 for b*
Let X(t, a) = X(t) = 1 for all t, and, thus,
be defined as in (8.3). From Lemma l.5,
PROOF.
det X(t)
~2(t)
X-l(t) = [
-Yi(t)
-Y2(t)]. Yl(t)
Also, from Theorem 1.3.3 and the variation of constants formula for a linear system, oX(t, a)
--...:....:........:. = X(t, a)
Oa
ft X-l(S, a) OC(a) - X(s, a) ds, oa
0
where O(a) is the matrix in (8.2). Thus, OC(a) oa
= [
0 00]. -1
In the same way, 02X(t, a) = 2X(t, a) oa 2
ft X-l(t, a) oO(a) oX(s, a) ds. oa
0
oa
From the definition of B(a), these relations imply 2B'(a) = (01: - $)
(8.7)
r
f" yi ds + f"
YIY2 ds - {1
ds
r
+ Ii y~ ds,
0 0 0
.B "
(a)
= 01:
f"o
Y2 -°Yl ds - {1
Oa
f" Yl -
°Yl
0
Oa
Y2 -°Y2 ds - {1.
Ii
0
Oa
f" Yl -
Oy2 ds,
0
Oa
where for simplicity in notation 01: = OI:(a) = Yl(7T, a), {1 = {1(a) = Y2(7T, a), Ii = Ii(a) = Yl(7T, a), $ = $(a) = Y2(7T, a). Using the fact that det X(t) = 1 for all t, we have (8.8)
4(B2 -:-1)
=
(01:
+ $)2 -
4(01:$ - {11i)
= (01: -
$)2
+ 41i{1.
Suppose b is such that B(b) = 1 and B'(b) ;:;; o. We wish to pro.J~ there is as> 0 such that B'(a) < 0 for b < a < b + S. If B'(b) < 0, it is clear that isuch a [) exists. Suppose B '(b) = 0, B(b) = l. For this value of b, Relation (8.8)
127
LINEAR SYSTEMS AND LINEARIZATION implies 4a{3 = -(a -
i3)2 . From (8.7), we have 'I
2i:xB (b)
1 = f1'[, o' aY2 + 2" (a -
,]2 = °
(3)Yl'
° °
and thus, aY2 (s) + (a - ~)Yl (s)/2 = for ~ s ~ ir. Since Yl, Y2 are linearly independent, we have a = 0, a = ~. Since B'(b) = 0, relation (8.7) implies (3 = 0. The fact that a + ~ = 2B(b) = 2 implies a = ~ = 1. Also, direct evaluation in (8.7) and an integration by parts yields
B"(b)
=
U:
Yl Y2 ds
r-s:
s:
yi ds y~ ds.
°
The Schwarz inequality implies B"(b) < since the functions Yl, Yz are linearly independent. Therefore, there must be a i} such that B'(a) < for
°
b
°
Suppose now there is a c* such. that B'(a) < for b < a < c* and B'(c*) = with B(c*) > -1. Then B2(C*) -1 < and (8.8) implies ci(c*)fJ(c*) < and, in particular, ci(c*) 0. One easily shows from (8.7) and (8.8) that
° °
*
*
°
*
for any a such that ci(a) 0. Since B2(C*) < 1, ci(c*) 0, it is clear that B'(c*) 0, which is a contradiction. This proves the lemma for the case B(b) = 1. The case B(b) = -1 is treated in essentially the same manner to complete the proof of the lemma. In the middle part of the proof of this lemma, the following relationship was proved.
*
°
LEMMA,8.6 If for a particular b, B2(b) = 1, B'(b) = 0, then B"(b) < if B(b) = 1, and B"(b) > if B(b) = -1. In particular, a root of B2(b) = 1 can be at most double. A necessary and sufficient condition for a double root at b is
°
Yl(1T, b)
=
Y2(1T, b)
= 1 (or -1),
LEMMA 8.7. If ao is the smallest root of the equation B2(a) ao is simple and B'(ao) < 0.
= 1, then
PROOF. From Lemma 8.3, B(a) > 1 for a < ao. If ao were a double root of B(a) = 1, then Lemma 8.6 would imply it is a maximum, which is impossible. This proves the lemma. By combining the information in the above lemmas we obtain
ORDINARY DIFFERENTIAL EQUATIONS
128
at
THEOREM 8.1. There exist two sequences {aD < a l at ~ ...} of real numbers, ale' at -+ 00 as k -+ 00,
~
a2 ~ ...}, {at
~
~
ao < at ~ at
< al
~
a2 < at ~ at
< as ~ a4 < ... ,
such that equation (8.1) has a periodic solution of least period 7T (or 27T) if and only if a = ale for some k = 0,1,2, ... (or at for some k = 1,2, ... ). The equation (8.1) is stable in the intervals (a 2 , at),
and unstable in the
in~ervals
(at, at),
(-00, ao],
The equation (8.1) is stable at a 2Hl or a 2H2 (or atHl or a 2H2 ) if and orily if a 2Hl =a2H2 (or atHl = atle+2), k ~ 0. Equation (8.1) is always unstable if a is complex. PROOF. We remark first of all that (8.1) is unstable if a is complex (Lemma 8.1). Lemma 8.4. implies the existence of the two infinite sequences {ale}' {at}· Lemma 8.3 implies a o =F -00. Lemma 8.3 and 8.7 imply that the first zero of B2(a) = I is ao, it is simple and (...:. 00, ao] is an interval of in· stability. Lemma 8.5 implies ao < at and Lemmas 8.5 and 8.2 imply (a o' at) is an interval of stability. If the equation (8.1) is stable at at, then B(aj = - I would have a double root at at. Lemma 8.6 would imply B(a) has a minimum at at and then Lemma 8.5 implies at =at. If at
d2y dT2
+ (u2 + 2q cos WT)Y =0,
LINEAR SYSTEMS AND LINEARIZATION
129
B(a)
Figure III.S.l
where. a ~ 0, q, ware real constants. If we let equivalent to (8.10)
y
W'T
= 2t,
this equation is
+( (~r +~; cos 2t)Y =0,
which is a special case of (8.1) with a = (2afw)2, cp = (8qfw 2) cos 2t. Let us investigate this equation for q near 0. The equation for the characteristic multipliers is p2 - 2Bp
+1 =
0,
where B = B(a, q, w) is a continuous function of a, q, w- 1 . For q = 0, a principal matrix solution of (8.10) is COS
x=
2a t w
~ sin 2a t] 2a w
[
2a 2a - -;- sin -;;; t
2a cos -;- t
Therefore, B(a, 0, w) = cos 27Tafw and B2(a, 0, w) < 1 if 2a;j:. kw [or a ;j:. k 2], k = 0, 1, 2, .... Thus. for any a and w for which 2a;j:. kw, there is a q=q(a,w) such that B2(a,q,w) <1 and the equation (8.9) is stable (Lemma 8.2). These stability regions are shown in the (a 2 , q)-plane in Fig. 8.2. One can give a very geometric proof of this stability result in the
130
ORDINARY DIFFERENTIAL EQUATIONS q
~--- __
0"2
(~ Figure III.8.2
following way. If PI> P2 are characteristic multipliers of (8.9), then Pll, P2 1 are also characteristic multipliers since PIP2 = 1. Also, since a, q, ware real PI, P2 are multipliers. Therefore, if PI -:1= P2, IPII = IP21 = 1 for q =0, then IPII = IP21 =1, PI -:l=P2 for q sufficiently small since the multipliers are continuous functions of q. To determine whether or not (8.9) is stable or unstable for a 2 = (kw/2)2, k an integer, is extremely difficult. Analytical methods applicable to this problem will be discussed in a later chapter. For equation (8.9), in any neighborhood of any of the points (0, [kw/2]2), k an integer, it can actually be shown there are values of (a 2 , q) such that (8.9) is unstable. There are a large number of results in the literature which are concerned with the estimation of the stability regions in Theorem 8.1 in terms of the function a cp(t) in (8.1). We give a theorem of Borg [1] on the first stability region which generalizes a result of Liapunov [1].
+
I THEOREM 8.2.1 If p(t + 7T) = tinuous, real and 7T
p(t)
r o
°
:t for all t,
Ip(t)1 dt
f: p(t) dt ~ 0, p(t) is con-
~ 4,
then all solutions of the equation (8.11)
ii
+ p(t)u =
0,
are bounded on (-00, (0). PROOF. It is only necessary to show that no characteristic multiplier of (8.11) is real. If a characteristic multiplie.r P of (8.11) is reltl, then there is a real solutioq u(t) ;f= 0, u(t 7T) = pu(t) for all t. Either u(t) -:1= for all t or it has infinitely many zeros with two consecutive zeros a, b, ~ b - a ~ 7T. In the first case u~) = pu(O), U(7T) = pu(O) and U(7T)/U(7T) = u(O)/u(OkSince ii/u p = 0, an integration by parts yields k<
°°
+
+
fo"uu
f
2 (t) " - d t + p(t) dt =0, 2 (t) 0
131
LINEAR SYSTEMS AND LINEARIZATION
which is impossible from the hypothesis on p. In the second case, we may assume that u(t) >0 on a < t < b. Let u(c) =maxa
-7T ; ; f0.
4"
Ip(t)1 dt ;;;;
f ~ dt bl··(t)1
a lu(t)1
1 >1- fb lii(t)1 dt;;;; -Iu(oc) -u(,B)I, u(c) a
u(c)
for any oc, f3 in (a, b). From the mean value theorem, there exist oc, f3 such that u(oc) = u(c)j(c - a), -u(f3) = u(c)j(b - c). Therefore, 4
1
1
4
b-a
4
->--+--= ~--~-, ir c-a b-c (c-a)(b-c)-b-a-7T
+
since 4X'1j ~ (x y)2 for all x, y. This contradiction shows that the characteristic multipliers are complex and proves the result.· EXERCISE
8.1.
Consider the equation ii.+ [a - 2qS(t)]u = 0,
(8.12)
where S(t)=+1 if -7Tj2~t<0, =-1 if 0~t<7Tj2, S(t+7T)=S(t) for all t. Show that every neighborhood of a point (ao, 0), ao > 0, in the (a, q) plane where cos = ±1 contains points for which (8.12) has ll'hbounded solutions. Hint: If a> 12ql, r2 = a + 2q, 8 2 = a - 2q, one can show that A = (pI + p2)j2 is given by
7TFo
1[
7T]
A= (8 + r)2 cos 7 - (8T +.r). - (8 - r)2 cos -2 (8 - r) • 4r8 2
m.9. Reciprocal Systems Consider the system (9.1)
i = A(t)x, A(t
+ T) =
A(t), -
00
00,
where A(t) is a continuous real or complex n X n matrix and T > 0 is a constant. Following Lyapunov, system (9.1) is calledlreciprocallif for every characteristic multiplier p of (9.1) the number p-I is also a characteristic multiplier. If A is real, the characteristic multipliers occur in complex conjugate pairs and (9.1) will be reciprocal if p a characteristic multiplier implies p-I is a characteristic multiplier. Let d designate the Banach space
ORDINARY DIFFERENTIAL EQUATIONS
132
of continuous real or complex valued n X n matrix functions A(t), -00 0 such that the system (9.2)
i
= B(t)x,
is stable on (-00, (0) for all Bin f4 with IA - BI the set of A in d for which (9.1) is reciprocal.
< 8. We let rJld designate
LEMMA 9.1. If A is in rJld and po = po(A), Ipol = 1, is a simple characteristic multiplier of (9.1), then there is a 80 > 0 such that the system (9.2) has a simple characteristic multiplier po(B), Ipo(B)1 = 1, for every B in rJld with IA -BI <8 0 • PROOF. Suppose A in rJld and po = po(A) is a simple characteristic multiplier of (9.1), Ipol = 1. Formula (1.U) implies that the matrix solution XA(t); XA(O) = I, of (9.1) is a continuous function of A in d. In particular, the characteristic multipliers of (9.1) are continuous functions of A in d. Therefore, there is a disk De(po), e > 0, in the complex plane of radius e and center po and a 81 > 0 such that (9.2) has exactly one characteristic multiplier po(B) in De(po) for all B in rJld, IA - BI < 81 . Since (9.2) is reciprocal, p-1(B), is also a characteristic multiplier. But, Po1(B) = po(B)/lpo(B)12 =1= po(B) unless Ipo(B)1 = 1. On: the other hand, the hypothesis IPo(A)1 = 1 implies Po1(A) = Po(A) and by continuity of Po(A) in A, we can find a 80 < 81 such that Po1(B), Po(B) belong to De(Po) if IA - BI <8 0 , Bin rJld. This implies Ipo(B)1 = 1 for IA -BI <8 0 , Bin rJld, and proves the lemma. THEOREM 9.1. If A is in rJld and all of the characteristic multipliers of (9.1}are distinct and have unit modulii, thenA is strongly stable relative to rJld. PROOF. This is immediate from Lemma 9.1 and the Floquet representation of the solutions of a periodic system. LEMMA 9.2.
If A in d is real and there exists an n X n no~ingular
If A is not continuous but only Lebesgue integrable, then the results below are valid with IAI = J5IA{s)l ds.
2
LINEAR SYSTEMS AND LINEARIZATION
133
matrix D such that either (i)
DA(t)
=
-A( -t)D
(ii)
DA(t)
=
-A'(t)D,
or (A' is the transpose of A)
then A is in P/td. In (i) the principal matrix solution X(t) satisfies X-l( -t)DX(t) = D and in (ii) it satisfies X'(t)DX(t) = D for all t. PROOF.
Let X(t), X(O) = I, be a matrix solution of (9.1). If Y(t),
= Yo is an n X n matrix solution of the adjoint equation if = -yA(t), then Y(t)X(t) = Yo for all t. Case i. If DA(t) = -A( -t)D, then Y(t) = X-l( -t)D satisfies the adjoint equation. In fact, Y(t) = _X-l( -t)D =X-l( -t)A( -t)D = _X-l( -t)DA(t) = - Y(t)A(t). Therefore, X-l( -t)DX(t) = D which implies X(t) is similar to X( -t) for all t. If X(t) = P(t)e Bt , then P(O) = I and X(t) similar to X( -t) implies the roots of det(e BT - pI) = 0 and det(e- BT -,u) Y(O)
= 0 are the same. Obviously, these roots are the reciprocals of each other and this proves case (i). Case ii. If DA(t) = -A '(t)D, then Y(t) = X'(t)D is a solution of the adjoint equation. In fact, Y=X'D=X'A'D=-X'DA=-YA. Thus, X'(t)DX(t) = D for all t and X'(t) is similar to X-l(t) for all t. The remainder of the argument proceeds as in case (i). By far the most important reciprocal systems are[Hamiltonian systems' namely, the systems
Ex =H(t)x,
(9.3) where H'
=
H is a real2k
X
2k matrix of period T, E=[
0
I k ],
-h 0
and Ik is the k X k unit matrixJ Since E2 = -12k. E' = -E system (9.3) is a special case of system (9.1) with A = -EH and EA =H =H' =A'E' = -A 'E. Thus, (9.3) is reciprocal since this is a special case of case (ii) of Theorem 9.2 with D = E. Furthermore, the matrix solution X(t), X(O) = I of (9.3) satisfies" (9.4)
X'(t)EX(t)
= E.
The set of all matrices which satisfy (9.4) is called the real symplectic group or sometimes such a matrix is called E-orthogonal. A general class of complex reciprocal systems consists of the canonical systems, (9.5)
Jx=H(t)x,
134
ORDINARY DIFFERENTIAL EQUA'f!ONS
where H is Hermitian (i.e. H* H) and J -
(9.6)
.[11'
-~ 0
= H,
where H* is the conjugate transpose of
0]
i=J=i,
-Iq'
To prove this system is reciprocal, let X(t), X(O) = I, be a principal matrix solution of (9.5). Then A = -JH is the coefficient matrix in (9.5) and
a,
dt X*(t)JX(t)
=
-X*(t)H*(t)J*JX(t) - X*(t)J2H(t)X(t)
and X*(t)JX(t) is a constant. Since X(O)
= 0,
= I, we have
X*(t)JX(t) = J,
(9.7)
for all t. Thus X(T) is similar to X*-l(T) and the result follows immediately. Matrices which satisfy (9.7) are called J.unitary. Notice that J-unitary mat· rices are nonsingular. System (9.5) includes as a special case system (9.3) for the following reason. Any two skew Hermitian matrices A, B with the same eigenvalues with the same multiplicity are unitarily equivalent; that is, if A* == -A, B* = -B, then there is a matrix U such that U*U = UU* = I and U* AU = B. If, in addition, A and B are real, there is a real unitary (ortho. gonal) matrix U such that U'AU =B. Since E is skew Hermitian with the eigenvalue i of multiplicity k there is a unitary U such that UEU* = J, where J is given in (9.6) with p = q = k. If we let x = U*y then (9.3) is transformed into (9;5) with H replaced by UHU*. A matrix U which accom· plishes this is
U =~
(9.8)
.j2
[ lie -ille]. -ille·
lie
A special case of (9.5) is the second order system
.u+QU+ P(t)u=O,
(9.9)
whereu is a k.vector, Q is a constant matrix, Q* fact, system (9.9) can be written·in the form
= -Q,
P*(t)
= P(t).
In
Ki=H(t)x, where
K
=
[-Qlie
-lie] 0'
H(t) =-= [P(t)
o
0]. lie,
There is a nonsingular matrix P such that PKP* = J and, therefore, the trans· formation x = P*y reduces (9.9) to a special case of (9.5)~
LINEAR SYSTEMS AND LINEARIZATION
135
Case (i) in Lemma 9.2 expresses some even and oddness properties of the coefficient matrix A(t). To illustrate, consider the second order matrix system
u + P(t)u =0,
(9.10)
where P(t..f- T) = P(t) is a k x k continuous real matrix. This is equivalent to the system of order 2k,
i
(9.11)
== A(t)x,
A(t) =
[ ° °Ik] . -P(t)
If P =(P1k), j, k =1, 2 is partitioned so that P u is an r X r matrix and P2 2 is an 8 X 8 matrix with P 1k (t) = (-l)J+kPjk(---:t), then (9.11) is reciprocal. In fact, case (i) of Lemma 9.2 is satisfied with D =diag(Ir, -Is, -Ir, Is). An even more special reciprocal system is the system
u+Fu=O,
(9.12)
F = diag(O"i, ... , O"~), The matrix F is periodic of any period. For any period T > 0, the characteristic multipliers of (9.12) are P21-1 = P21, P2j-l = eil1jT , j = 1,2, ... , n. These multipliers are distinct if and,only if (9.13)
20"1 :;E
°
(mod w),
0"1
± O"k:;EO (mod w),
where T = 27T/W. Consequently, if (9.13) is satisfied, Theorem 9.1 implies there is a S >'0 such that all solutions of (9.2) are bounded in (-00,00) provided that B is in P/t.r;l and
IB-AI <S, and F is defined in (9.12). In particular, if (t) = (t + T) is real, symmetric or satisfies the even and oddness conditions above and (9.13) is satisfied, there is an eo > 0 such that all solutions of the system (9.14)
u + (F + e«P(t»u =
°
are bounded in (- 00, 00) for all Iel ~ eo. Compare this result with the one at the end of Section 8 for a single second order equation. If some of the conditions (9.13) are not satisfied, it is very difficult to determine whether there is an eo > 0 such that solutions of (9.14) are bounded for lei ~ eo· For Hamiltonian systems, some general results are available (see Section 10) but, for the other cases, only special equations have been discussed. Iterative schemes for reaching a decision are available and will be discussed in a subsequent chapter.
136
ORDINARY DIFFERENTIAL EQUATIONS
I:t'system (9.14) is not a reciprocal system, it can be shown by example (see Chapter VIII) that even when (9.13) is satisfied, solutions of (9.14) may be unbounded for any £ =1= o. EXERCISE 9.l. Suppose B(t) is an integrable T-periodic matrix such that the characteristic multipliers of x = B(t)x are distinct and have unit modulii. If A(t) is an integrable T-periodic matrix such that x = A(t)x is
fo
a
a
reciprocal, then there is a > 0 such that T IA(8) - B(8)1 ds < implies the equation x=A(t)x i~ stable on (-00, 00). Hint: Use the continuity of the fundamental matrix solution of x = A(t)x in A which is implied by formula (l.ll).
m.l0. Canonical Systems
As in Section 9, let d be the Banach space of n X n complex integrable matrix: functions of period T with IAI= ffIA(t)ldt. Let CCd be the subspace consisting of those matrices ofthe·form -JH, H* =H and (10.1)
J =i
[~p. -~J.
If A belongs to CCd, then the aBBOciated periodic differential system is a canonical system
(10.2)"
Jx =H(t)x,
H*=H,
Our main objective in this section is to give necessary and sufficient conditions in order that system (10.2) be strongly stable on (- 00, 00) relative to CC.r4. We have seen in Section 9 that a canonical system (10.2) is reciprocal and, therefore, stable on (- 00, 00) if and only if all characteristic multipliers of (10.2) are on the unit circle and have simple elementary divisors; or, equivalently, that all eigenvalues of the monodromy matrix S have simple elementary divisors and modulii 1. This latter statement is equivalent to saying there is a nonsingular matrix Usuch that Vi real.
It was also shown in Section 9 that the monodromy matrix S of a principal matrix solution of (10.2) is J-unitary; that is,
(10.3)
S*JS=J.
Since the stability properties of (10.2) depend only upon the eigenvalues of S, we use the following terminology: A J-unitary matrix Sis 8table if all
LINEAR SYSTEMS AND LINEARIZATION
137
eigenvalues have modulii 1 and simple elementary divisors. A J-unitary matrix 8 is strongly stable if there is a 8 > 0 such that every J-unitary matrix R for which IR -81 < 8 is stable. Preliminary to the discussion of stability, we introduce the following terminology. For any x, y in en, define the bilinear form
<x, y) = i- 1y*Jx.
<
For Hamiltonian systems, the expression x, y) is related to the Lagrange bracket. In fact, if U is given in (9.8) and x = U*u, y* = v*U, then for all x =1= 0 in V. Two vectors x, yare called J-orthogonal if <x, y) = 0. If <x, y) = for all y, then Jx = which implies x = 0. This immediately implies that 8 is J-unitary if and only if <8x, 8y) = <x, y) for all x, y. It is immediate from the definition of J that the vectors e1 = (1, 0, ... ,0), e2 = (0, 1, ... ,0), ... , en = (0,0, ... , 1) satisfy
°
°
°
(10.4)
<ei, ek ) =0, <ei, ei ) = 1, <ei, ei ) = -1,
°
j =1= k,
1 ~j ~p, p <j ~ n.
LEMMA 1O.l. Eigenvectors associated with distinct eigenvalues of a stable J-unitary matrix are J-orthogonal. The eigenvectors of a stable J-unitary matrix span en.
If x, yare eigenvectors associated with A, /1-, respectively, A=1= /1- and IAI = 1, then <x, y) = <8x, 8y) = >"p,<x, y). If >..p, =1= i, then <x, y) = O. Since the system is assumed stable p,-1 = /1- and A=1= /1- imply >..p, = AI/1- =1= l. This proves <x, y) = 0. The proof that the eigenvectors of a stable J-unitary matrix span the space is now supplied in exactly the same way that one proves the corresponding result for unitary matrices in linear algebra. PROOF.
LEMMA 10.2. A nonnegative eigenspace V of a stable J-unitary matrix is positive. A nonpositive eigenspace of a stable J - unitary matrix is negative. PROOF. Suppose V is nonnegative. If x is in V and <x, x) for any y in V and any complex number >..,
= 0,
o ~ ..<x, y). have <x, y) = 0 for otherwise>" could be chosen in such a
then
We must manner as to make this expression negative. Therefore, <x, y) = 0 for all y in V. This
138
ORDINARY DIFFERENTIAL EQUATIONS
fact together with Lemma 10.1 imply (x,y> = 0 for all y. . Thus, x = O. A similar argument applies when V is nonpositive and proves the lemma. TmcOREM 10.1. A matrixB is a stable J.unitary matrix if and only if there is a J.unitary matrix: U such that
U-IBU =diag(efl\
(10.5)
••• ,
efl'a),
where each VJ is real. PROOF. Let G = diag(ef!'l, ... , ef'a). If B = UGU-l and U is J.unitary, it is clear that B is stable and J.unitary. Conversely, suppose B is a stable J.unitary matrix and A . ef ,!' is an eigenvalue of multiplicity r. One can choose a J·orthonormal basis vI, ... , vr for the eigenspace V}, so that (vJ, v k ) = 0, j k, = ±1 if j = k; j, k = 1,2, ... , r. For if not, there would be an eigenvector v such that is J.orthogonal to V},. This is impossible from Lemma 10.1 since it would imply v is J. orthogonal to the whole space 0". Some of the vi may have (vi, vi) = +1 and some may have this expression equal -1. From Lemma 10.1, we c,!-n choose a J·orthonormal basis u 1 , .•. , u" of eigenvectors for the whole space and we can order these vectors in such a way that (uJ, uk) =0, j *k, (uJ, uJ) = 1, j ~ p' and = -1 for p' <j ~ n. But the law o~ inertia for Hermitian forms and (10.4) imply that p' = p. If U =·(u1 , ••• , u"), then (10.5) is satisfied. Furthermore, for the vectors el in (10.4), (Ue l , Ue k ) = (uJ, uk) = (ei, ek ) for all j, k. Thus, (Ux, Uy) = (x, y) for all x, y and U is J.unitary. This proves the theorem. It was shown in the proof of the above theorem that for any stable J.unitary matrix, a complete set of J·orthonormal eigenvectors uJ and corresponding eigenvalues Ai = efl'l can' be obtained. We shall say that the eigenvalue Ai is of positive type (or negative type) if (uJ, uJ) = 1 (or -1). In the Russian literature, the terminology is first kind (or second kind). There are p eigenvalues of positive type and n - p of negative type. The following theorem asserts that a stable J.unitary matrix is strongly stable iftmd only if a multiple eigenvalue is not of both positive and negative type.
*
v
THEOREM 10.2. A stable J.unitary matrix is strongly stable if and only if each of its eigenspaces is definite (tl1at is, either positive or negative). PROOF. Suppose S is a stable J.unitary matrix and has an eigenvalue 1, whose eigenspace is not definite. Lemma 10.2 implies tqere are eigenvectors vI, v2 such that (VI, v2 ) =0, (VI, vI) = 1, (v 2 , v, = ~1.
A, IAI =
Choose v3 ,
••• ,
v", J.orthogonal to vI, v2 so that VI, ... , v" form a basis for
0". For anY' or; ~ 0, define the linear transformation R: 0" _0" by
139
LINEAR SYSTEMS AND LINEARIZATION
Rv l
=
Rv 2
=
+ (sinh oc)v2 ], 8[(sinh oc)v l + (cosh oc)v 2 ], 8[(cosh oc)v l
k=3, ... ,n. One easily verifies that 1 for any oc > o. Thus, R is not stable on (-00, (0). Sinc~ R approaches 8 as oc ~O, this implies 8 is not strongly stable. Conversely, suppose 8 is a stable J-unitary matrix whose eigenspaces are definite. Let Ak, k = 1,2, ... , r, be the distinct eigenvalues of 8 with multiplicity nk and let V k of dimension nk be the corresponding eigenspaces. Then
+
and we let Pk denote the corresponding projection operators of On onto V k; that is, for any x in On, P k X is in V k and Pk x = x if x is in V k. These projection operators satisfy Pk Pi = 0, j oF k, Pz = P k , and can be defined by the formula (10.6)
where Yk is a positively oriented circumference of a circle with center Ak and radius so small that it contains no other eigenvalue of 8. If R is any n X n matrix with 1R - 81 sufficiently small, then each of circumferences Yk encloses exactly nk eigenvalues (counted with their multiplicities) of R. Therefore, the integrals
define projection operators Qk of On onto subspaces W k of dimension nk such that the W k are invariant under R and On = W 1 EB W 2 EB"· .• EB W r . The subspace W k is the algebraic sum of the generalized eigenspaces of R associated witli the eigenvalues of R enclosed by Yk . Also, this formula shows that Qk is a continuous function of Rand IQk - Pkl ~O as R ~8. Our next objective is to show that <x, x) is definite on each Wk. For x in Wk , <x, x) =
+ «Qk -
+ 2Re
Pk)x, (Qk - Pk)X)
Pk)x).
ORDINARY DIFFERENTIAL EQUATIONS
140
From the definition of <x, y>, we have I<x, y>1 ~ IJI . Ixl . Iyl for all x, yin On. Also, since S is definite on eigenspaces, there is an oc > 0 such that I<x, x>1 ~ oclxl 2 for all x in Vk and each k = 1, 2, ... ,r. Therefore, for x in Wk, I<x, x>1 ~OCIPkXI2 -IJI'IQk - Pkl 2 'l x l2 -2IJI'jPkl'IQk - Pkl21 x l 2 ~ [oc(I-IQk - Pkl)2 -IJI'IQk - Pkl 2 - 2I J I'IPkl'IQk - PklJl x l2 • Consequently, if IS - RI is small enough, the continuity of the projections Qk in R implies that k
= 1,2, ... , r.
Since each Wk is invariant under R, for any integer m, it follows that IR mxl 2 ~ 2oc-1 11
= 2oc-1 1<x, x>1
~ 2oc- 1 IJI'lxI 2 ,
for all x in Wk, k = 1,2, ... , r, if Ris J-unitary. For any x in On we have x = QIX Qr x and, thus,
+ ... +
It is easy to soo this implies all eigenvalues must have simple elementary divisors and modulus 1. Thus, R is stable which in turn implies S is strongly stable. This proves the theorem.
THEOREM 10.3. System (10.2) is strongly stable on (-00, (0) if and .• only if the monodromy matrix is strongly stable. PROOF. It was remarked at the beginning of this section that (10.2) is stable on (- 00, (0) if and only if the monodromy matrix is stable. Since the solutions of (10.2) depend continuously upon A in .!ii, then (10.2) is strongly stable on (-00, (0) if the monodromy matrix is strongly stable. Suppose now that the monodromy matrix S of the solution X(t), X(O) = I, of (10.2) is stable and not strongly stable. In the proof of Theorem 10.2, it was shown that any neighborhood of S contains a J-unitary matrix R which has an eigenvalue with modulus >1. Since S-IR is nonsingular, Lemma 7.1 implies there is a matrix F such that S-IR = eF. For IR -SI sufficiently small, one can show directly from the power series representation of eF or from the implicit function theorem that F can be chosen to be 'a continuous function of R 'which vanishes for R =S. Furthermore, S-IR being J-unitary iinplies e-F =J-1eFoJ=e r ' FOJ and we can choos~F so that _F=J-IF*J. Therefore, F can be written as F = TJ-IG .where G is Hermitian and T is the period. If we define Y(t) = X(t)etr'G, then Y(O) = I, Y(T) = R, and Y(t) is a
LINEAR SYSTEMS AND LINEARIZATION
141
fundamental matrix solution of the canonical system Ji
= L(t)x,
where L =H + X*-1GX- 1. However, L(t) may not be periodic of period T. On the other hand, we can alter L(t) by a symmetric perturbation L 1(t) such that L(t) + L1(t) is periodic of period T, foT IL1(t)1 dt < S for any preassigned S. If we let Y 1(t), Y 1(0) = I, be the fundamental matrix solution with L replaced by L + L1, then formula (1.11) implies there is a constant X such that
IY 1 (T) ~ Y(T)I
~
SX.
Therefore, for S sufficiently small the monodromy ma~rix Y 1(T) will have an eigenvalue with modulus >1 and (10.2) is not strongly stable. This proves the theorem. If the eigenvalues elllJ of S in the representation (10.5) are ordered so that the first p are of positive and the remaining n - p are of negative type, then Theorem 10.2 states that a stable J-unitary matrix S is strongly stable if and only if (10.7)
Vj
:;E Vk (mod 21T),
If S is the monodromy matrix of a stable canonical system (10.2) and the eigenvalue~ of S are denoted by eI6JT , j = 1,2, ... , n,· and ordered in the same way as above, then Theorem 10.3 implies that the canonical system is strongly stable on (-00, 00) if and only if
(10.8) EXERCISE
·10.1.
Show that inequalities (10.8) are equivalent to the
inequalities (10.9)
(Jj
+ (Jk;i: 0 (modw),
j, Ie = 1, 2, ... , n,
for system (9.12), and, thus, (9.12) is strongly stable if and only if (10.9) is satisfied. Compare these inequalities with (9.13). Theorem 10.3 answers many of the questions concerned with the qualitative properties of the stability of canonical systems. It states very clearly the critical positions of the characteristic multipliers for which one can always find a canonical system that is unstable and yet arbitrarily near to the original one. One of the basic remaining problems is the determination of an efficient procedure to obtain these critical positions of the multipliers. A method for doing this for equations which contain a small parameter is given in Chapter VIII.
142
m.ll.
ORDINARY DIFFERENTIAL EQUATIONS
Remarks and Suggestions for Further Study
The reader interested in more results for the stability' of perturbed linear systems may consult Bellman [1], Cesari [1], Coppel [1]. In the case of linear systems with· constant or periodic coefficients, the stability properties of the linear system were completely determined by the characteristic exponents of the equation. Furthermore, if all characteristic exponents have negative real parts, then the linear equation is uniformly asymptotically stable. Consequently, the linear system can be subjected to a perturbation of the type given by.relation (2.9) and the property of being uniformly asymptotically stable is preserved for the perturbed system. If the coefficients of the matrix A(t) in (1.3) are bounded, then every solution of (1.3) is bounded by an exponential function. Therefore, it is possible to associate with each solution x of (1.3) a number A = Az defined by .
1
A.=hm sup-Ioglx(t)l. t-+ 00 t This number A was called by Lyapunov [1] the characteristic number of the solution x. If A < 0, then Ix(t)l- 0 as t - 00. If Az <: 0 for all solu~ions x of (1.3), then all solutions approach zero exponentially. On the other hand, the example of Perron for A(t) as in (2.14) has the property that perturbations satisfying (2.9) can lead to unbounded solutions in contrast to' the above remarks for periodic systems. There is an extensive theory of the characteristic numbers of Lyapunov and their application to stability (see Cesari [1], Malkin [1], Nemitskii and Stepanov [1], Lillo [1]). In the discussion of the preservation of the saddle point property in Theorem 6.1, we were only concerned with the orbits on the stable and unstable manifolds. Of course, one could consider the following problem: Suppose x =0 is a saddle point of (4.1) andJ: Rn_Rn has continuous first derivatives such thatf(O) =0, 8f(0)/()x =0. Does there exist a neighborhood V of x = 0 such that for any f of the above class, there is a transformation h which takes the trajectories of (4.1) onto the trajectories of the perturbed equation (6.3) in the neighborhood V? This problem has a long and interesting history. The reader may consult Poincare [1] and Lyapunov [1] for the analytic case, Sternberg [1], Chen [1] for the case where h may have a finite number of derivatives and Hartman [1] for the general case. The problePl posed in the previous paragraph in a local neighborhood of a critical point can be made much more general. In fact, one call; say two n-dimensional differential equations ;:<, (11.1)
i =f(x),
LINEAR SYSTEMS AND LINEARIZATION
(11.2)
143
if =g(y),
are equivalent in a region G of Rn if there is a homeomorphism h which takes the trajectories of (ILl) onto the trajectories of (11.2) in G. Suppose the vector fields j, g belong to some topological space 31. A system (11.1) is said to be structurally stable if there is a neighborhood NU) of j in 31 su('~ that (11.1) and (11.2) are equivalent for every g in N(J). The study of structural stability and equivalent systems of differential equations is at present one of the most exciting topics in differential equations. All of the concepts are equally meaningful for vector fields on arbitrary n-dimensional manifolds. The reader is referred to the basic paper of Peixoto [1] for two-dimensional
systems and the paper of Smale [1] for the general problem. See also the book of Netecki [1] . The presentation of the stability theory of Hill's equation given in Section 8 relies very heavily upon the book of Magnus and Winkler [1], but does not begin to' indicate the tremendous number of results that are available for this equation. The precise determination of the a-intervals of stability and instability for a given function cp(t) is extremely important in the applications. Each function cp(t) defines a class of functions depending upon a and the particular aj, a; of Theorem 8.1 yield special periodic functions of period 7T or 27T. It is important to discuss the expansion of arbitrary functions in terms of these special periodic functions in the same way that one develops a function in a Fourier series. For ~ more complete discussion as well as many references, see Arscott [1], Cesari [1], Magnus and Winkler [1], McLachlan [1]. The presentation in Section 10 follows the thesis of Howe [1] and should be considered only as introduction to the theory of stability of Hamiltonian and canonical systems. The topological characteristics of the class of strongly stable systems have been discussed as well as the regions of stability and instability for given equations. Computational methods also have been devised for equations with a small parameter. The reader is referred to Gelfand and Lidskii [1], Yakubovich [1, 2], Krein [1], Diliberto [2], Coppel and Howe [1] for results as well as further references.
CHAPTER IV Perturbations of Noncritical Linear Systems
It is convenient to have Definition 1. If A(t) is an n X n eontinuous matrix function on ( - 00, (0) and ~ is a given class of functions which contains the zero function, the homogeneous system (1)
X =A(t)x,
is said to be noncritical with respect to q; if the only solution of (1) which belongs to q; is the solution x =0. Otherwise, system (1) is said to be critical with respect to ~. Throughout the following, P-6(-00, (0) ={f: (-00, oo)-+C n , f continuous and bounded} and for any fin P-6(-00, (0), If I =sup-OOd
x = A(t)x +f(t, x),
which belong' to one of the classes PA( -00, (0), df!J or f!JT for 'the case in which the matrix A is in f!JT, system (1) is noncritical with respect to the class under discussion and f is "small" in a sense to be made precise later. The presentation will proceed as follows.' The nonhomogeneous linear system is studied first in great detail. It is shown that system (1) being noncritical with respect to ~, one of the classes P-6(-00, (0), d~lor f!JT, implies that the nonhomogeneous linear equation has a unique solution in ~ which depends linearly and continuously upon the forcing function in ~. 144
PERTURBATIONS OF NONCRITICAL LINEAR SYSTEMS
145
For f in (2) "small," the contraction principle yields the existence of a solution of (2). By extending the saddle point property of Chapter III, Section 6, to nonautonomous equations, one obtains the stability properties of the solution of (2). Some generalizations of the results are given in.8ection 4 together with some elementary properties of Duffing's equation with large forcing and large damping in Section 5. At the end of Section 1, there is also a stability result for the case when A is not in f1JJT. LEMMA 1. (a) System (1) with A in f1JJT is noncritical with respect to 81( -00, (0) [or df1JJ] if and only if the characteristic exponents of (1) have
nonzero real parts. (b) System (1) with A inf1JJ T is noncritical with respect to f1JJTif and only if I - X(T) is nonsingular, when X(t), X(O) = I, is a fundamental matrix solution of (1). PROOF. (a) If the characteristic exponents of (1) are assumed to have nonzero real parts, then the Floquet representation of the solutions implies that the only solution of (1) in 81( -00, (0) is x = 0 and (1) is noncritical with respect to 81( -00, (0) and, therefore, also with respect to df1JJ. Conversely, if there is no solution x =f= 0 of (1) in 81( -00, (0), then the Floquet representation implies there cannot be a characteristic exponent A of (1) with A= i(} since equation (1) would then have a nonzero solution etOtp(t),
p(t
+ T) = p(t).
(b) With X(t) defined as in the lemma, the general solution of'(I) is X(t)xo, where Xo is an arbitrary constant vector. System (1) has a nonzero periodic solution of period T if and only if there is an Xo =f= 0 such that [X (t T) - X (t) ]xo = 0 for all t. Since A is assumed to belong to f1JJ T, this is equivalent to the existence of an Xo =f= 0 such that [X(T) -l]xo = 0; that is, X(T) - I is singular. This proves the lemma.
+
Remark I. If A in equation (1) is a constant, then (1) is noncritical with respect to 81( -00, (0) or df1JJ if and only if all eigenvalues of A have nonzero real parts. The equation (1) is noncritical with respect to f1JJT if and only if all eigenvalues A of A satisfy AT *- 0 (mod 27Ti).; or equivalently, the purely imaginary eigenvalues iw of A satisfy w *- 0 (mod 27T/T).
IV.I. Nonhomog.meous Linear Systems Basic to any discussion of problems concerned with perturbed linear systems is a complete understanding of the nonhomogeneous linear system (l.l)
x = A (t)x + f(t),
where f is a given function in some specified class.
ORDINARY DIFFERENTIAL EQUATIONS
146
Let us recall that the equation adjoint to (1) is if = -yA(t) and X-l(t) is a fundamental matrix solution of this equation if X(t) is a fundamental matrix solutIon of (1). Since A belongs to flJ'T, it follows that the adjoint equation has a nontrivial solution y with y' in flJ'T (' is the transpose of y) if and only if y(O)[X-l(T)-I] =0; that is, if and only if X-l(T)-1 is singular. Equation (1) has a solution xin flJ'T if and only if[X(T) - l]x(O) = O. Since the matrices X-l(T) - 1 and X(T) - 1 are the same except for multiplication by a nonsingular matrix, it follows that the dimensions of the set of Yo such that Yo[X-l(T) -1] = oand the set ofxo such that [X(T) -I]xo = 0 are the same. Therefore, the adjoint equation and equation (1) always have the same number of linearly independent T-periodic solutions. LEMMA 1.1. (Fredholm's alternative). If A is in flJ'T and / is a given element of flJ'T, then equation (1.1) has a solution in flJ'T if and only if T
(1.2)
fo y(t)/ (t) dt
= 0,
for all solutions y of the adjoint equation
if =
(1.3)
-'-yA(t),
such that y' is in flJ'T. If (1.2) is satisfied, then system (1.1) has an r-parameter family of solutions in flJ'T, where r is the number of linearly independent solutions of (1) in flJ'T' PROOF. Since A and / belong to flJ'T, x(t) is a solution of (1.1) in flJ'T if and only if x(O) =x(T). If X(t, T), X(T, T) =1. is a matrix solution of (1), and x(O) = xo, then x(t)
Since X(t, T)X(T, 8) lent to
= X(t, O)xo + fX(t, 8)/(8) ds.
= X(t, 8) for all t,
o
T,
8, the condition x(T)
Jo
= Xo is equiva-
T
(1.4)
[X-l(T, 0) -I]xo =
X-l(8, 0)/(8) ds.
If B =X-l(T, 0) -I, b = foTX-1(8, 0)/(8) ds, then (1.4) is equivalent to Bxo = b. From elementary linear algebra, this matrix equation has a solution if and only if ab =0 for all row vectors a such that aB =0. Since X-l(t, 0) is a principal matrix of (1.3), the set of vectors a for which aB = 0 coiflcides with the set of initial values ofthose solutions y of (1.3) which are T-~iodic; that is, the solution y(t) =aX-l(t, 0) is a T-periodic solution of (1.3) if and only if aB = O. Using the definition of b, we obtain the first part of the lemma.
147
PERTURBATIONS OF NONCRITICAL LINEAR SYSTEMS
If (1.2) is satisfied and cp(t) is a solution of (1.1) in f!J'T, then any other solution in f!J'T must be given by x = Z + cp, where z is a solution of (I) in f!J'T. This proves the lemma. EXERCISE 1.1. What is the analogue of Lemma 1.1 for the case in which A is in f!J'T and f is in df!J'?
Example I.i.
Consider the second order scalar equation
If Xl =
w>o.
it + u = cos wt,
(1.5) U,
an equivalent system is
(1.5)'
Xl
=X2,
X2 =
-Xl
+ cos wt,
w>o.
The adjoint equation is YI = Y2, Y2 = -YI which has the general solution YI = a cos t + b sin t, Y2 = -a sin t +b cos t where a, b are arbitrary constants. If w -=1= I, there are no nontrivial solutions of the adjoint equation of least period T = 2rr/w. Thus, (1.2) is satisfied. Since the homogeneous equation has no nontrivial periodic solutions, the equation (1.5) has a unique periodic solution of period 27T/W. If w = I, then every solution of the adjoint equation has period 27T. On the other hand relation (1.2) is not satisfied since
fo y(t)f(t) dt f 211
211
=
(-a sin t cos t + b cos 2 t) dt = 7Tb
0
and this is not zero unless b = O. Therefore, the equation (1.5) has no solution of period 27T and, in fact, all solutions are unbounded since the general solution is U = a sin t + b cos t + (t sin t)f2. Example 1.2.
Consider the system (1.1) with -I
A=~ =~ [
(1.6)
2
o 2
One can easily show that (1.7)
I [ 1 + cos t - sin t eAt =.:: I - cos t - sin t 2 - I + cost -Slllt .
2 sin t 2 cos t 2 sin t
t
t]
-I + cos + sin -I+cost-sint. I + cos t + sin t
Since the rows of e- At are solutions of the adjoint equation, it follows that all solutions of the adjoint equation have period 27T, the same period as the forcing function f, and the solutions are linear combinations of the rows of e- At . To check the orthogonality condition (1.2), we need to verify that (211
Jo
.
e-Atf(t) dt = O. One easily finds that e-Atf(t) = z sin t, z = (-I, -I, I)
ORDINARY DIFFERENTIAL EQUATIONS
148
and the orthogonality condition is satisfied. Therefore, system (1.1) with A and I given in (1.6) has a periodic solution of period 217, and, in fact, every solution has period 217. EXERCISE 1.2. In example 1.2, is there a unique solution of period 217 which is orthogonal over [0,217] to all of the 217-periodic solutions of the homogeneous equation-which is orthogonal to all of the 217-periodic solutions of the adjoint equation? How does one obtain such a solution? EXERCISE 1.3. Show that every solution of (1.1) is unbounded 'if relation (1.2) is not satisfied and A, I are in !?I'T. THEOREM 1.1. Suppose A is. in !?I'T and ~ is one of the classes 00), .r;I!?I' or !?I'T. The nonhomogeneous equation (1.1) has a solution :ttl in ~ for every I in ~ if and only if system (I) is noncritical with respect to~. Furthermore, if system (I) is noncritical with resI*''''1; to ~,then :tt'l is the only solution of (1.1) in ~ and is linear and continuous in I; that is, :tt'(al bg) = a:tt'l b:tt'g for all a, b in C, I, g in ~ and there is a constant K such that ~(-oo,
+
+
1:tt'/1 ~ Kill,
(1.8) for
all/in~.
Finally, if ~ =.r;I!?I', then m[:tt'f] c m[/, A].
PROOF. Case 1. ~ = !?I'T . If I belongs to !?I'T , then Lemma 1.1 implies that equation (1.1) has a solution in !?I'T if and only if I is orthogonal in the sense of (1.2) to all T-periodic solutions of (1.3). But i = A(t)x and 1i = -yA(t) have the same number of linearly independentT-periodic solutions. Therefore, equation (1.1) has a solution in !?I'T for every I in !?I'T if and only if equation (I) has no nontrivial solutions in!?l'T; that is, (I) is noncritical with respect to !?I'T. If system (I) is noncritical with respect to !?I'T, then Lemma 1.1 implies system (1.1) has a unique solution :tt'l in !?I'T for every I in !?I'T. It is clear from the uniqueness that :tt' is a linear mapping of !?I'T into !?I'T. If X(t, 7'), X(7', 7') = I, i~ the principal matrix solution of (I), then ~he function :tt'l can be written explicitly as (1.9)
(:tt'f)(t)
=
fo [X-let + T,t) _l]-IX(t, t +8)/(t + 8)ds. T
The kernel function in this expression is known as .the Green' 8 lu~ion for the boundary value problem x(O) = x(T) for (1.1). The expJicit cOnW'Iltations for obtaining (1.9) proceed as follows. . If we let (:tt'/)(O) = aiel, then
PERTURBATIONS OF NONCRITICAL LINEAR SYSTEMS
149
+ f X(t, s)/(s) ds. t
(%/)(t) = X(t, O)xo
Since (%f)(t have
+ T) =
[l-X(t
o
(%/)(t), and X(t,s) = X(t,r)X(r,s) for any t,r,s, we
+ T,t)]X(t, O)xo = -[l-X(t+ T,t)!] fX(t, s)/(s) ds o
, ft+T
+X(t+T,t)
X(t,s)/(s)ds.
t
Multiplication by [/ - X(t
+ T,i)] -1
and a substitution in the formula for
(%f)(t) yields (1.9).
Formula (1.9) obviously implies there is a constant K such that (1.8) is satisfied. In fact, K can be chosen as T-IK =
sup
I[X-l(t + T,t) _1]-lX(t, t
0;;;;8, t;;;;T
+ s)1 .
This proves Case 1. Case 2. ~ = fJI( -00, oo}. Let X(t) be a fundamental matrix solution of (1). The Floquet representation implies X(t) = P(t)e Bt where P(t + T) = P(t) and B is a constant. Furthermore, the transformation x = P(t)y applied to (1.1) yields
..
(1.10)
iJ = By + P-l(t)/(t)~ By + g(t),
where g is in fJI(-oo, (0) (or dfffJ) if/is in f!l(-oo, (0) (or dfffJ). Since P(t) is nonsingular, it is therefore sufficient for the first part of the theorem to show that (1.lO) has a solution in fJI(-oo, (0) for every g in §B(-oo, (0) if and only if iJ = By is noncritical with respect to fJI( -00, (0); that is, no eigenvalues of B have zero real parts. By a similarity transformation, we may assume that
where all eigenvalues of B+(Bo)(B-) have positive (zero) (negative) real parts: If y = (u, v, w), g = (g+, go, g-) are partitioned so that block matrix multiplication will be compatible with the partitioning of B, then (1.lO) is equivalent to the system (1.11)
(a)
u=B+u+g+,
(b)
v=
(c)
w=B_w+g_.
Bo v
+ go ,
150
ORDINARY DIFFERENTIAL EQUATIONS
There are positive constants K, ex so that (1.12)
(a) (b)
leB+tl ~ Ke at , leBetl ~ Ke- at ,
t ~ 0, t ~ o.
Equations (LIla), (1.Ilc) have unique bounded solutions on (-00, 00) given, respectively, by Q
(a)
(1.l3)
(%+ U+)(t)
= f·
eC- B+8U+(t
+ 8) ds,
co
(b)
(% - U-)(t) =
fo
e- B - 8u_(t + 8) ds.
-co
One can either verify this directly or apply Lemma III.6.1. These remarks show that if if = By is noncritical·with respect to P1J( -00, 00) then there is a unique solution %/ of (1.10) in P1J( -00, 00). Furthermore, using (1.12) we see that 1%+ u+1 ~ (K/ex) lu+l, 1%- u-I ~ (K/ex) lu-I· Therefore, %/ satisfies (1.8). If the matrix Bhas any eigenvalues with zero real parts; that is, the vector v in (1.Ilb) is not zero dimensional, we show there is a Uo in P1J( -00, 00) such that all solutions of (1.10) are unbounded in (-00,00). This will complete the proof of Case 2. Without loss in generality, we may assume that Bo = diag(Bol, ... , BOB), where BOj = iWj I + R j and R j has only zero as an eigenvalue. It is enough to consider only one of the matrices BOJ since iWj may be eliminated by the multiplicative transformation exp( iWj t). Thus, we consider the equation
x=Rx+U, where R has only zero as an eigenvalue and U is in f!l( -00, 00). For any row vector a,
ai=aRx+au,
*
for all t. If a 0 is chosen so that aR = 0 and U= a*, where a* is the conjugate transpose of a, then ax = lal 2 > 0 which implies ax(t) -'>- 00 for every solution x(t). Therefore, every solution is unbounded as t -'>- 00. This completes the proof of case 2. Notice that the U chosen in (1.10) to make all solutions unbounded if B has eigenvalues with zero real parts was actually periodic and, therefore, / = P( t)g is an almost periodic (quasiperiodic) function which makes all solutions of (1.1) unbounded if (1) has purely imaginary characteristic expoRents. . Case 3. !!) = d&'. The previous remark implies that (1.1),.las a solution in P1J( - 00, 00) for all / in d&' if and only if no characteristic' exponents of (1) have zero real parts or, equivalently, (1) is noncritical with respect
PERTURBATIONS OF NONCRITICAL LINEAR SYSTEMS
151
to df!J'. If (1) is noncritical with respect to df!J', then the unique solution X"J in !J6( - 00, 00) is ~ periodic transformation of the functions given in (1.13) where g';, g_ are almost periodic with their modules contained in m[f, A] Therefore, it remains only to show that the functions in (1.13) are in df!J' and their modules are in m[f, A]. We make use of Definition 1 of the Appendix. Suppose 01.' = {~} is a sequence in R. Since g is almost periodic there is a subsequence 01. = {OI.n,} of 01.' such that {get + 0I.n)} converges uniformly on (-00,00). Since .Jf+ is a continuous linear operator on f?O(-oo,oo), this implies {(ff+g+)(t + 0I.n)}, converges uniformly on (-00,00). TlJ.us, Definition 1 of the Appendix implies X"+g+ is almost periodic. Theorem 8 of the Appendix implies m[X"+g+] C mfi+] C m[/,A]. The same argument applies toX"_g_ to complete the proof of Case 3 and the theorem.
EXERCISE 1.4. Let X" be the operator defined in Theorem 1.1. Prove or disprove the relation m[X"f] =m[f, A] for every J in df!J'. Theorem 1.1 clearly illustrates that requiring a certain behavior for some solutions of the nonhomogeneous equation (1.1) for forcing functions J in a large class of solutions imposes strong conditions on the homogeneous equation (1). For the case in which A does not belong to Y'T, similar conclusions can be drawn but the analysis becomes more difficult. Preliminary to the statement of the simplest result of this type are some lemmas of independent interest.
°
LEMMA 1.1. If A(t) is a continuous n X n matrix, IA(t)1 <M, ~ t < 00, and X(t, T), X(T, T) = I, is the principal matrix solution of (1), then there is a ~ > such that
°
1 IX(t, 8) - X(t, T)I ~ '2IX(t, T)I '
for all t, T, 8 ~'o and 18 - TI
~ ~.
PROOF. Since X(t, 8) = X(t, T)X(T, 8) for all t, T, 8 in [0, (0), it is sufficient to show there is a ~ > such that IX( T, 8) - II ~ 1/2 for T, 8 ~ 0, IT - 81 ~ ~. Since
°
X(T, 8) - I
=
{A(u)[X(u, 8) -I] du 8
+ {A(u) du, 8
ORDINARY DIFFERENTIAL EQUATIONS
l52
an application of Gronwall's inequality yields IX(T, s) - II ~ MIT for all T, s. If 0 is such that Moe M6 ~ 1/2, the lemma is proved. LEMMA 1.2.
-
sl eM11'-sl
With A and X(t, T) as in Lemma 1.1, the condition
f~ IX(t, s)1 ds < e, a constant, for all t ~ 0 implies IX(t, s)1 uniformly bounded for 0
~
s ~ t < 00; that is, the equation (1) is uniformly stable for to
~
O.
PROOF .. Since X(t, s)X(s, t) = I for all s, t, it follows that X(t, s) as a function of s is a fundamental matrix solution of the adjoint equation. Therefore,
X(t, s) = I
+f
t
X(t, g)A(g)dg s
for all t, s. In particular, for 0 <s ~t, the hypotheses of the theorem imply IX(t, s)l < 1 Me. The uniform stability follows from Theorem III.2.1. This proves the lemma.
+
THEOREM 1.2. If A(t) is a continuous n X n matrix, IA(t)1 < M, 00, then every solution of (1.1) is bounded on [0, 00) for every continuous f bounded on [0, 00) if and only if system (1) is uniformly asymptotically stable.
o~ t <
PROOF.
We first prove that if every solution of (1.1) is bounded on
[0,00) for every continuous f bounded on [0,00), then f~IX(t, s)1 ds constant, for.t
~
<e,
a
O. The solution of (1.1) with x(O) = 0 is given by
foX(t, s)f(s) ds. t
x(t)
=
Let as'[0, 00) be the class offunctionsj: [0, 00) _On,f continuous and bounded and let If I =supo~koolf(t)l· For any fixed t in [0,00), consider the mapping T t : as'[0, 00) -as'[0, 00) given by
(TtI)(a)
rX(a, s)f(s) ds, =
{
0
foX(t, s)f(s) ds, t
t
~
a<
00.
For each fixed t, T t is a continuous linear map. Furthermore, by hypothesis, for eachf in as'[0, 00), there is a constant N such that ITtl I ~ N, 0 ~ t < 00. Consequently, the. principl~ of uniform boundedness implies there. is a constant K such that ITtfl ~ K.lfl for all t in [0,00), f in as'[0, c~). In particular, C
I{X(t, S)f(S)dS\ ~ K If I,
o~ t <
00.
PERTURBATIONS OF NONCRITICAL LINEAR SYSTEMS
153
If X = (Xjk), let fI' k(8) be the function which is the sign of Xjk(t, s), j, k = 1,2, ... , n, for a fixed t. Choose a sequence of uniformly bounded continuous functions f/.'rk(8) which approach fl' k(8) pointwise almost everywhere on [0, t]. Choose the norm of a vector x = (Xl, ... , xn) to be Ixi = maxj IXj~. For any given j, k, let ft(8), ft. r(8) be the n-vectors with all components zero exc~pt the kth which is fl' k(8), ft: :(8), respectively. Then
f X(t, 8)ft. ,(8) ds f X(t, 8)ft(8) ds. t
lim r--?
<Xl
t
=
0
0
From the above definition of the norm of a vector and the fact that t
fo X(t, s)it. ,(8) d8 I ~ K 1ft. rl ~ K I ,
o ~ t < 00,
it follows that
~ I{X(t, 8)f;(8) ds I~ K I , for all t ~ 0 andj, k =:= 1,2, ... , n. Since all norms in en are equivalent, this {IXjk(t, 8)1 d8
clearly implies there is a constant c such that f~ IX(t, 8)1 ds
< c, 0 ~ t <
00.
From Lemma 1.2, this relation implies X(8, 7') is uniformly bounded by a constant N for 0 ~ 7' ~ 8 < 00 and, thus, equation (1) is uniformly stable for to ~ O. Therefore,
If: for all t
~ 7'.
X(t, 8)X(8, 7') d8
I~
fIX(t, 8)1 . IX(s, 7')1 d8
~ Nc,
But the first expression in this inequality is equal to
(t - 7') IX(t, 7')1 . Therefore, IX(t, 7')1 approaches zero as t - ? 00 uniformly and equation (1) is uniformly asymptotically stable. Conversely, suppose system (1) is uniformly asymptotically stable for to ~ O. From Theorem III.2.1, there are positive constants K, oc such that
IX(t, 7')1
~
Ke-a(t-T),
o ~ 7'~ t < 00.
The general solution of (1.1) is t
x(t)
= X(t, O)x(O) + fo X(t, 8)f(s) ds,
t ~ O.
If f is in 81[0, 00), then Ix(t)1 ~ Klx(O)1 + If I . (K/oc) for all t ~ 0 and thus every solution of (1.1) is in 81[0, 00). Th:is proves the theorem.
Theorem 1.2 is due to Perron [1] and was the first general statement dealing with the determination of the behavior of the solutions of a homogeneous equation by observing the behavior of the solutions ofa nonhomo-
154
ORDINARY DIFFERENTIAL EQUATIONS
geneous equation for forcing functions in a certain given class of functions. Investigations along this line have continued to this day with the most significant recent contributions being made by Masseraand Schaffer [1]. For a better appreciation of the problem, we.rephrase it in another way. Let (BI, !i) be two Banach spaces of functions mapping [a, (0) into On where a may be finite or infinite. The pair (BI, !i) is said to be admis8ible for equation (1.1) if for every fin !i), there is at least one solution x of (l.I) in BI. Theorem l.1 states that (BI(-oo, (0), BI(-oo, (0», (.III!?;, .III!?;), (!?;p, !?;p) are admissible for equatio~ (l.I) if and only if equation (1) is noncritical with respect to BI(-oo. (0), .III!?;, !?;p, respectively. Theorem l.2 shows that (BI[O, (0), BI[O, (0» is admissible "for (l.I) if equation (1) is uniformly asymptotically stable. The further investigation of such admissible pairs is extremely interesting and the reader is referred to Coppel [1, Chap. V], Hartman [1, Chap. 13] and Massera and Schaffer [1], Antosiewicz [2].
IV.2. Weakly Nonlinear Equations-Noncritical Case Throughout this section, it will be asSumed that A is a continuous n X n matrix in !?;p, O(p, a) = {x in On, e in 0': Ixl ~ P, lei ~a}, "1(p, a), M(a), P ~ 0, a ~ 0, are continuous functions which are nondecreasing in both variables, "1(0,0) = 0, ¥(O) = 0, and !l'ifi("1' M) = {q: R X O(po, eo)On: q continuous, Iq(t, 0, e)1 ~ M(lej), Iq(t, x, e) - q(t, y, e)1 ~ "1(p, a)lx - yl, for all (t,x,e), (t,y,e) in R X n(p,a), 0 ~ P ~ Po, 0 < a < eo and q(t,x,e) is continuous in x, e uniformly for t in R}. If q is in !l'ifi("1, M), then automatically q(., x, e) is in BI( -00, (0). In fact, Iq(t, x, e)1 ~ "1(p, a) Ixl M(lej) for (t, x, e) in R X O(p, a). A function q will be said to be in.III!?; n!l'ijt("1,M) if q is in!l'ifi("1,M) and, for each fIxed e; q(t,x,e) is almost periodic in t uniformly with respect to x for x in compact sets. A function q will be said to be in !?;p n !l'''·fi(''1, M) if q is in !l'ifi("1, M) and q(t T, x, e) = q(t, x, e) for all (t, x, e) in R X O(po, eo). A function q will clearly be in !l'ifi("1' M) for some "1,·M if q(t, x, e)_O and CJq(t, x, e)/ox _0 as x -0, e -0 uniformly in t. In this section, results are given concerning the existence of bounded, almost periodic and periodic solutions of the nonlinear "equation
+
+
i
= A(t)x + q(t, x, e),
where q is in !l'ifi("1' M) and the homogeneous equation (1) is noncritical. More specifically, we prove
i
THEOREM 2.l. Suppose!i) is one of the classes BI(-oo, oo»)lit!?; or !?;p. If q is in !'J n !l'ifi("1, M) and· system (1) is noncritical with "respect to !i), then there are constants PI >0, el >0 and a function x*(t, e) con-
PERTlJ,RBATIONS OF NONCRITICAL LINEAR SYSTEMS
155
tinuous in t, e for -00
(2.2)
w = !Ix = :£q(., x(·), e),
where:£ is the operator uniquely defined by Theorem 1.1. From Theorem 1.1, !I: P}p, -rP}. Furthermore, the fixed points of !I in P}p, coincide with the solutions of (2.1) which are in P}p,. We now use the contraction principle to show that !I has a unique fixed point in P}p, for PI and lei sufficiently small.
Since q is in !l'ift(T), M), (2.3)
Iq(t, x, e)1 ~ Iq(t, x, e) - q(t, 0, e)1 + Iq(t, 0, e)1 ~ T)(pI, ell Ixl + M(eI),
for (t, x, e) in R X Q(pl' ell. Let K be the constant defined in (1.8) and choose PI ~ po, el ~ eO positive and so small that K[T)(pl, el)pl + M(el)]
< pl·
For this choice of PI, el, it follows from relations (1.8), (2.2), (2.3) and the fact that q is in !l'ift(T), M) that I!lxl ~ Klq(·, x(·), e)1 ~ K[T)(pl, el)IXI +M(eI)]
<8I x -yl, for all x, yin P}p" 0 ~ lei ~ el and 8 < 1 is a positive constant. Therefore, !I is a uniform··(Jontraction on P}p, for lei ~el and has a unique fixed point x*(t,e) in fPp ,. Since q(t,x,e) is continuous in x,e uniformly in t, x*(t,e) is continuousin t,e, -00 < t < 00, 0 ~ lei ~ eI· For e = 0, x = 0 is obviously a solution of (2.1) and, therefore, x*(·, 0) =0. To prove the last statement of the theorem, suppose 0/ == {~} is a sequence in (-00,00). Let M be a compact set containing {x*(t, e), te(-oo, 00), lei ~ ed. Theorem 11 of the Appendix implies there is a subsequenc~ a = {C\!n} of a' such that {q(t + C\!n,x,e)} converges uniformly for te(-oo,oo),xeM and each fixed e. Since x*(t,e) = Jt"q(', x*(·, e), e)
156
ORDINARY DIFFERENTIAL EQUATIONS
and Jf'is continuous and linear on 14(-00,00), this implies {x*(t verges uniformly for tin (-00,00) for each fIxed e. The fact that
+ an, e)} con-
m[x*(·,e)1· C m[q,A]
follows from Theorem 8 of the Appendix. Theorem 2.1 has interesting implications for the equation i = A(t)x
(2.4)
+ b(t) +eh(t, x, e),
where e is a scalar, h belongs to fjJi? j I (fI, M), b is in fjJ and system (1) is noncritical with respect to fjJ. Of course, as before fjJ is one of the classes 14( - 00, (0), .JII[7jl or [7jlT. Under these hypotheses, Theorem 1.1 implies that the equation i = A(t)x
has a unique solution :f(b in
if =
fI).
+ b(t),
If I:f(bl
A(t)y +eh(t, y
< po and x =
y +:f(b in (2.4), then
+ :f(b, e) ~ A(t)y + q(t, y, e).
The function q is in !l'".jt('Y/, M) with 'Y/(p, 0') = O''Y/l(p, 0'), where 'Y/l(p, 0') is a continuous function of p, 0', since q(t, y, e) approaches zero a;s e -+0 at least as fast as a linear function in e. Therefore, Theorem 2.1 implies the existence of a solution x*(t, e, b) of (2.4), x*(" e, b) in fI), x*(·. 0, b) =:f(b and this is the only solution of (2.4) which is in fI) and at the same time in a pl-neighborh90d of :f(b.
IV.3. The General Saddle Point Property Theorem 2.1 of the previous section asserts the existence of a distinguished solution x*(·, e) of (2.1) in a class fI) of functions. What are the stability properties of the solution x*(·, e)? If the real parts of the characteristic exponents of the linear system (1) have nonzero real parts, one would expect the stability properties of x*(·, e) to be the same as those of the solution x = 0 of (1). The purpose of this section is to prove this is actually the case and also ~o discuss some of the geometrical properties of the solutions of (2.1) near x*(·, e) in the same manner as the saddle point property was treated in Section III. 6. The proofs will follow the ones in Section III.6. very closely but are slightly more complicated due to tp.e fact that the differential equations depend explicitly upon t. ;i To redl).ce the equations to a simpler form, let (3.1)
where x*(·, e) is the function given in Theorem 2.1. If x is a solution of
PERTURBATIONS OF NONCRITICAL LINEAR SYSTEMS
157
equation (2.1), then y is a solution of
iJ = A(t)y + p(t, y, e),
(3.2)
where p(t, y, e) = q(t, x*(t, e) + y, e) - q(t, x*(t, e), e). Consequently, if q is in .!l'ijZ(1], M), thenp is in .!l'ijZ(1], 0). To simplify the equations even further letX(t) = P (t)e Bt , P(t + T) = P(t), B a constant matrix, be a fundamental matrix solution of (1) and let y = P(t)z. If y is a solution of (3.2), then z is a solution of
z=Bz+!(t,z, e),
(3.3)
where !(t, z, e) = P-l(t)p(t, P(t)z, e).,Finally, we can assert that if ~ is one of the classes f!4( -00, (0), d&' or fiT, and q is in ~ n .!l'ijZ(1], M), then! is in ~ n .!l'ijZ(1], 0). Any assertions made about system (3.3) yield implications for system (2.1) which are easily traced through the above transformations. The remainder of the discussion centers around (3.3) under the hypothesis that! is in .!l'ijZ(1], 0) and the eigenvalues of the matrix B have nonzero real parts, k with positive real parts and n - k with negative real parts. As in Sect on III.6, the space Cn can be decomposed as
cn =C't
(3.4)
$C~,
C't = 1T+Cn, C~ = 1T_cn,. where 1T +, 1T _ are projection operators, C't ' ~. have dimensions k, n - k, respectively, are invariant under B and there are positive constants K, IX such that (3.5)
(a)
le Bt1T+zl
(b)
le Bt1T_zl ~ Ke- at 11T-zl '
~
Kecxtl1T+zl,
t
~
0,
t
~
o.
For any a in (-00, (0), let z(t, a, ZC1, e) designate the solution of (3.3) satisfying z(a, a, ZC1, e) = ZC1, let K designate the constant in (3.5) and, for any 8 > 0, define (3.6)
Also, let
(a)
~(a, 8, e) = {ZC1 in Cn; 11T-ZC11 < 2~' Iz(t, a, ZC1, e)1
<
8, t
~ a},
(b)
U(a, 8, e) = {ZC1 in Cn; 11T+ZC11
< 2~' Iz(t, a, ZC1, e)1 <
8, t
~ a}.
s: designate {z in Cn; Izl < pl.
THEOREM 3.1. If! is in .!l'ijZ(1], 0) and the eigenvalues of B have nonzero real parts, then there are 8 > 0, el > 0, f3 > 0 such that for any a in (-00, 00), 0 ~ lei ~ el, the mapping 1T_ is a homeomorphism of S(a, 8, e) onto (1T_cn) n B~12K' S(a, 8,0) is tange~t to 1T_Cn at zero and {'.l 7\
t>
fT.
158
ORDINARY DIFFERENTIAL EQUATIONS
for any zG in 8(a, 8, e). The mapping 7T+ is a homeomorphism of U(a, 8, e) onto (7T+cn) (") B;/2K' U(a! 8,0) is tangent to 7T+cn at zero and (3.8)
Iz(t, a, zG,
e)i
~
2K 17T+ZG Ie(j(t-G),
t ~ a,
for any zG in U(a, 8, e). Furthermore, if g(., a, e): (7T_Cn) m/2K -?8(a, 8, e) is the inverse of the homeomorphism 7T-, then g(z_, a, e) is lipschitzianin z_ with lipschitz constant 2K. Iff is in dfJ> (") !eijt('Y}, 0), then g(z_, a, e) is almost periodic in a with module contained in m[f]. Iff is in fJ>T (") !eijt('Y}, 0), then g(z_, a, e) is periodic in a of period T. The same conclusions hold for the inverse of the homeomorphism 7T+ of U(a, 8, e) onto (7T+cn) (") B;/;K'
n
Before proving this theorem, let us make some remarks about its geometric meaning and some of its implications. The accompanying Fig. 3.1
\
\ \ I
\ (0-) )(8(0-,0,£)
V
I
8(0,£)
Figure IV.3.1
may be useful in visualizing the following remarks. For any a E ( - 00, 00), the set {a} X 8(a, 8, e) is a subset of En+!. For any 7 in [a; t], z(t, a, zG, e) = z(t, 7, z(7,a, zG, e), e), and, therefore, any solution of (3.3) with initial value in {a} X 8(a, 8, e) must cross {7} X 8(7,8, e) for any 7 ~ 0 for which 7T_Z(7, a, zo-, e) has norm less than 8J2K. Since this may not occur for ~l 7 ~ a, the set consisting of the union of the {o-} X 8(0-, 8, e) for all 0- in,f~oo, 00) may not be an integral manifold in the sense that the trajectory of any solution with initial value (7, ZT) on the manifold remains on the manifold for all t ~ 7. On the other hand, this set can be extended to an integral
PERTURBATIONS OF NONCRITICAL LINEAR SYSTEMS
159
manifold by extending the set {a} X S(a, 8, e) to a set {a} X S*(a, 8, e) where S*(a, 8, e) =S{a, 0, 8) U {z:z =z(a, T, ZT, e), (T, ZT) in {T} X S(T, 8, e) for some T ~a}. The set S(8, e) consisting of the union of the {a} X S*(a, 8, e) is then an integral manifold of (3.3) and is rightfully termed a 8table integral manifo1il since all solutions on this manifold approach zero as t _ 00 and these are the only solutions which lie in a certain neighborhood of zero for increasing' time. In the same way one defines the corresponding sets U*(a, 8, e) = U{,u,o, &} U {z: z = z(a, T, ZT, e), (T, ZT) iri'U(T, 8, e) for some T ~ a} and an unstable integral manifo1il U(8, e), The sets S(8, e) and U(8, e) are hypersurfaces homeomorphi(j to R X B~-k and R X .B~, respectively, andS(o, e) n U(o, e) is the t-axis. If k ~ 1 ,the above remarks imply the solution z = 0 of (3.3) is unstable for lei ~ e1 and any a in (-00, 00). From (3,7), if k = 0, the solution z = 0 of (3.3) is uniformly a8ymptotically 8table for lei ~ e1 and to ~ a for any a in (-00, (0). If B andf are real in (3.3), then the sets S(a, 0, e), Uta, 0, e) defined by taking only real initial values are in Rn, If system (3.2) is real, then the decomposition X(t) 0;= P(t)e Bt of a fundamental matrix solution of (1) may not have P(t), B real if it is required that P(t + T) = P(t) for all T (see Section III.7). the other hand, this decomposition can be chosen to be real if it is only required that P(t + 2T) = P(t) for all t. In such a case, system (3.3) will be real, but f belongs to f!iJ2T n fi'ijt(T], 0) if P is in f!J'T n fi'ijt(T], 0). This implies tha~ the sets S(a, 0, e), Uta, 0, e) will be periodic in a of period 2T rather than T. It is not asserted in the statement of the theorem that S(a, 0, e) is tangent to TT_Cn at zero. This may not be true since f(t, y, e) could contain a term which is linear in y and yet approaches zero when e _0. Theorem 3.1 is clearly a strong generalization of Theorem III.6.1 due to the fact that the perturbation term f(t, z, e) may depend explicitly upon t.
On
PROOF OF THEOREM 3.1.. Since this proof is so similar to the proof of Theorem III.6.1, it is not necessary to give the details but only indicate the differences. In the same way as in the proof of Lemma 6.1, one easily shows that, for anY,§!olution z(t) of (3.3) which is bounded on [a, (0), there must exist a z_ in en.. such that (3.9)
z(t) =eB(t-a)z_ +
f eB(t-S)TT_f(8, Z(8), t
e) dB
a
. 0
+f
e-BsTT+ f(t
+ 8, z(t + 8), e) d8,
t ~ a,
co
and for any solution z(t) of (3.3) 'which is bounded on (-00, a], there is a z+ in C't such that z(t) satisfies the relation
160
ORDINARY DIFFERENTIAL EQUATIONS
f eB(t-B)7T+!(8, Z(8), e) ds t
(3.10)
z(t) =eB(t-U)Z++
+f
o
" e- B'7T_!(t + 8, z(t + 8), e) ds,
t ~ a.
-00
We first discuss the existence of solutions of (3.9) on [a, 00) for any z_ in 7T_On. There is a constant K1 such that 17T+zl ~ K11zl, 17T-zl ~ K11z1 for any z ill On. If K, ex are the constants given in (3.5), and 7J is the lipschitz constant of !(t, z, e) with respect to z, choose S, e1, so that (3.11)
With this choice of S, e1 and for any z_ in 7T_on with Iz-I ~ S/2K, define f§(17, z_, S) as the set of continuous functions z: [a, 00) -+On such that Izl = supu:;t <00 Iz(t)1 ~ Sand 7T-Z(17) = z_. f§(u, z....:, S) is a closed bounded subset of the Banach space of all bounded continuous functions taking [a, 00) into On with the uniform topology. For any z in f§(17, z_, S) define!Tz by
+ f eB (t-B)7T_ !(8, Z(8), e) ds t
. (3.12)
(!Tz)(t) = eB(t-u)z_
"
o
+ f e- B'7T+!(t + 8, z(t + 8), e) ds,
t ~ a.
00
Exactly as in the proof of Theorem 111.6.1, one uses the contraction principle to show that !T has a unique fixed point z*(o, a, z_, e) for lei ~e1, z*(·, 17,0, e) =0, z*(t, a, z_, e) depends continuously upon t, a, z_, e and (3.13)
IZ*(t,17, z_, e) - z*(t, a, L, e)1
~ 2K [exp - ex(~ -
a)] Iz- -z_1
~ a. From the definition of 8(17, S, e) and the above construction of z*(o, a, z_, e), it follows that
for t
(3.14)
8(17, S, e) ={z: z = Z*(17, a, z-;-, e), z_ in (7T_On) n
Bd/2k
for lei ~ e1. Relation (3.14), (3.13) and the fact" that z*(·, a, 0, e) = 0 implies relation (3.7) with f3 = ex/2. If we let g(z_, a, e) = Z*(17, a, z_, e), then o (3.15) g(z_, a, e} = z_ e':' B '7T+!(17 8, Z*(17 8, a, z_, e), e) ds.
+f
+
+
00
As in the proof of Theorem 111.6.1,
" Ig(z_. a, e) I -g(z_, a, e)1 N
~
Iz- 2
LI
161
PERTURBATIONS OF NONCRITICAL LINEAR SYSTEMS
for any CT in (-00, (0) and lei ~el. This shows that g(', CT, e) is a one-to-one map of (7I'_cn) 1"'1 B"/2K into S(CT, e). Since the inverse of this map is 71'and therefore is continuous, it follows that it is a homeomorphism. The following estimate is also easy to obtain:
a,
*
17I'+Z (CT, CT, Z_, e)1
~
4K2Kl
- - 1](2K Iz-I, leI) Iz-I. (X
Since 71'-'-Z*(CT, CT, z_, e) = z_,. it follows from the properties of 1] that S(CT, S, 0) is tangent to 71'_Cn at zero. Relation (3.13) implies that the set S(CT, S, 0) is a lipschitzian manifold. NowsupposeJ.is in df!J 1"'1 fl'ijt(1], 0). We now prove the representation g(z_, CT, e) of S(CT, e) is almost. periodic in CT with module contained in m[f]. From (3.15), it will be necessary to estimate Z*(CT s, CT, z_, e) as a function of CT and s. To simply the notation, let z(t, CT) = z*(t, CT, z_; e),!(t, z) = !(t, z, e). The equation for Z(CT t, CT) becomes
a
+
+
foeB(t-8)7I'_!(CT+S,Z(CT+S, CT))ds t
(3.16)
Z(CT+t, CT) =eBtz_+ +
foe- Bs7l'+!(CT+t+S,Z(CT+t+S,CT))ds. CD
The objective is to show that Z(CT + t, CT) is almost periodic in CT if! is in df!J 1"'1 fl'ijt(1], 0). Suppose 01.' = {~} is a sequence in (-oo,co). There is a subsequence 01. = {an} of 01.' such that {J(t + an,z)} converges uniformly for te(-co, co), Izl S8. For any n,m, let
'Yn,m = sup-.. <s
un m·(t, a) SKK 1 ft e- Ci(t-s)17(8, el)U n m(s, a) ,
-
0
+KKI
'
oo
fo e-
2KKI
CiS
17(8,el)Unm (t+s,a)+--'Ym,
'
01.
Using the ineqllalities (3.11), one easily observes that Un met, a) S 4KKl'Ym/0i. for all m. Since 'Y m -+ 0 as m -+ co, it follows that un,m (t, a) -+ 0 m -+ co and, therefore, z(a + t, a) is almost periodic in a with module contained in the module of f. In particular, geL, a, e) = z*(a, a,L, e) is almost periodic with module contained in m[fl. If! is in fiT 1"'1 fl'ijt(1], 0), then g(z_, CT T, e) = g(z_, CT, e) for all CT, since one sees directly from the uniqueness of the solution of (3.16) that Z(CT t T, CT T) = Z(CT t, CT) for all CT, t.
as
+
+ +
+
+
162
ORDINARY DIFFERENTIAL EQUATIONS
In the same manner as above, one uses (3.10) to prove the assertions on U(a, 8, e) to complete the proof of the theorem.
Asimple example iliustrating the above results is the forced van der Pol equation (3-.17) X2 =
-Xl
+ k(l -
xi)X2
+ eg(t),
where k =F 0, e are real parameters and g is in .!!If!). If X = is of the form (2.1) with
q(t, x, e) =
[ -kXrX2
(Xl,
X2), this system
~ eg(t)].
Since k =F 0, the eigenvalues of A have nonzero real parts and q is in .!!IfjJ n fRije(TJ, M) with TJ(p) ~Kp2 for some constant K and M = e SUPtlg(t)l. Therefore, Theorem 2.1 implies there are PI> 0, el > 0 such that system (3.17) has an almost periodic solution x*(t, e) with m[x*(" e)] c m[g], x*(·, 0) = 0 and this solution is unique in the pI-neighborhood of Xl = X2 = O. Since the eigenvalues of A have negative real parts if k < 0 and positive real parts if k > 0, Theorem 3.1 asserts that the solution x*(t, e) is asymptotically stable if k <0 and unstable if k >0. For e = 0, we have seen in Theorem 11.1.6 that the van der Pol has a unique asymptotically orbitally stable limit cycle for any k > O. This implies geometrically that the cylinder generated by the limit cycle in (Xl, X2, t)space is asymptotically stable. Therefore it is intuitively clear (and could be made very precise). that for e small there must be a neighborhood of this cylinder in which solutions of (3.17) enter and never leave. This" stable" neighborhood. is certainly more interesting than the almost periodic x*(·, e) determined above since x*(·, e) is unstable for k > O. A discussion of what happens in this neighborhood is much more difficult than the above analysis for almost periodic solutions and will be treated in Chapter VII.
1V.4. More General Systems Tlie first interesting modifications of the results of the previous sections are obtained by considering the parameter e as appearing in the system in a different manner than in (3.1). Consider the system (4.1)
eX =A(t)x,
where e > 0 isa real parameter and A belongs to fjJ T . In general, it is almost impossible to determine necessary and sufficient conditions on the matrix A
PERTURBATIONS OF NONCRITICAL LINEAR SYSTEMS
163
which insure that system (4.1) is noncritical with respect to one of the classes or &>T for all e in some interval 0 < e ~ eo. On the other hand, .if A is a constant matrix, the following result is true.
81(-00, (0), d&>
LEMMA 4.1. If A is a constant matrix and eo > 0 is given, then system (4.1) is noncritical with respect to 81( -00, (0) (or d~) (or &>T) for 0 < e ~ eo if and only if the eigenvalues of A have nonzero real parts. PROOF. The assertions concerning 81( -00, (0) or d&> are obvious from Lemma I since the eigenvalues of Ale are lie times the eigenvalues of A. Also, from Lemma I, (4.1) is noncritical with respect to .&>T if and only if det[J - exp(Ale)T] -=1= 0; that is, if and only if ATle -=1= 2k1Ti, k =±1, ±2, ... for all eigenvalues A of A. If Re A-=1= 0 for all A, this relation is satisfied. If there is a A = iw, w real, then these relations become e -=1= £l!TI2k1T, k =0, ± I, . . .. It is clear these relations cannot be satisfied for all e in an interval (0, eo]. This completes the proof of the lemma. LEMMA 4.2.
If A is a constant matrix,.@ is one of the classes8l( -00, (0), < e ~ eo,
d &> or &> T , and system (4.1) is noncritical. with respect to .@ for 0
then the system (4.2)
ex=Ax+f(t),
fin .@,
has a unique solution X" e f in f$, 0 < e ~ eo, X" e: .@ -+.@ is a continuous linear map and there is a K > 0 (independent of e) such that IX" e fl ~ Kill for 0 <e ~ eo. PROOF. The hypothesis and Lemma 4.1" imply the eigenvalues of A have nonzero real parts. If t = e7, Y(T) =x(eT), U(T) =1(8T), and x is a solution of (4.2); then
where U E 81(- 00, (0). Theorem 1.1 implies the existence of a unique X"u E 81( -00, (0) satisfying this equation, with lX"ul ~ X" lui .. Since lui = If I , it follows that X" e f
eX = Ax
+I(t, x, e),
164
ORDINARY DIFFERENTIAL EQUATIONS
where f satisfies the same conditions as stated in those theorems. This result is stated in detail for further reference. Suppose the eigenvalues of A have nonzero real parts, let 17+, 17_ be the projection operators taking On onto the invariant subsp~ces of A corresponding to the eigenvalues with positive, negative real parts, respectively, and let K, ex be positive constants so that (4.4)
t le Atl7+xl ~ Ke rlt ,
~
0,
t ~ O.
For any a.in (-00,00), let x(t, a,xa, e) designate the solution· of (4.3) satisfying x(a, a, xii, e) = x a and, for any S >0 and any function cp;.R --+on, let (4.5)
(a) S(cp,
a~ S, e) = {x = x a -
cp(a) in On; 117-[Xa - cp(a)]1 Ix(t, a, x a, e) - cp(t)1
(b)
U(cp, a, S, e)
= {x = x a -
< 2~
< S,
cp(a) in On; 117+[Xa - cp(a)]1 Ix(t, a, x a, e) - cp(t)1
t
~ a},
< 2~
< S,
t
~ a},
In words, the set S(cp, a, S, e) is the set of initial values at a of those solutions of (4.3) which have their 17- projections in a S/2K neighborhood of l7-cp(a) and remain in a S-neighborhood of the curve cp(t) for all t ~ a. A similar statement concerns U(cp, a, S, e) for 17+ and t ~ a. If B~ designates the set {x in On; Ixl < p}, then the following proposition holds. THEOREM 4.1. Suppose ~ is one of the classes PA( -00, (0), dE!J or E!JT. If f is in ~ n fRi/t(TJ, M) and the eigen,v!l'lues of A have nonzero real parts, then there are S > 0, e1 > 0, fJ > 0 and 11 function' x*(t, e) continuous in t, e for. -00
o ~ e ~ e1. Furthermore, the mapping 17- is a homeomorphism of S(x*(o, e), a, 8, e) onto (l7_on) n B;/2K and (4.6)
Ix(t, a, x a, e)-x*(y, e)1 ~2KI17_[Xa_X*(a, e)]le-E-I~(t-a),
.
-
t~a,
for any :l!a -x*(a, e) in S(x*(o, e), a, 8, e), 0 < e ~ e1. The map~ng 17+ is a homeomorphism of U(x*(o, e), a, S, e) onto (l7+on) n B~2K' Y' (4.7)
Ix(t, a, x a, e) - x*(t,
e)1 ~ 2K 117+[Xa - x*(a, e)]l eE-I~(t-a),
for any x a - x*(a, e) in U(x*( 0, e), a, 8, e), 0 < e ~ e1.
t ~ a,
PERTURBATIONS OF NONCRITICAL LINEAR SYSTEMS
165
The dependence of the stable and unstable manifolds of u*(·, e) upon (J is exactly the same as ip. the statement of Theorem 3.1. The only part of the proof of the above theorem which is nqt exactly the same as the proofs of Theorem 2.1 and 3.1 is the fact that the statements are asserted to be true on the closed interval 0 ~ e ~ el rather than the interval 0 < e ~ el. The proof for the interval 0 < e ~ el is the same as before and from the proof itself one observes that x*(·, e) --?O as e --?O. If one defines x*(·, 0) = 0, it will obviously be a continuous function on o ~ e ~ el. LEMMA 4.3. system
If A is constant matrix and eO > 0 is given, then the
i=eAx,
(4.8)
is noncritical with respect to 2J( -00, (0) or d&" for 0 < e ~ eO if and only if the eigenvalues of A have nonzero real parts. There is an eO > 0 such that system (4.8) is noncritical with respect to &"T for 0 < e ~ eO if and only if det A =1=-0. PROOF. The first part is a restatement of Lemma 1. Also, Lemma 1 implies (4.8) is noncritical with respect to &"T if and only if ewT =I=- 2k7r for all k = 0, ±1, ... and all real w such that A = iw is an eigenvalue of A. If w = 0 is not an eigenvalue of A, there is always an eo> 0 such that these inequalities are satisfied for 0 < e ~ eO. If w = 0 is an eigenvalue, there is never such an eO and the lemma is proved. EXERCISE 4.2. For what values of e in [0, (0) is system (4.8) noncritical with respect to &"T? For what complex values of e is system (4.8) noncritical with respect to &" T ? LEMMA 4.4. If A is a constant matrix, q) isoneoftheclasses~( -00, (0), d&" or &"T, and system (4.8) is noncritical with respect to q) for 0 < e ~ eo, then the system
i = e(Ax + I(t)),·
(4.9)
I in q),
has a unique solution :% e I in q), 0 < e ~ eo, :% e : q) --? q) is a continuous linear map and there is a K > 0 (independent of e) such that 1:% e II ~ K III for 0 < e ~ eo . . If, in addition,
I
is in d&" (or &"T) and
fl
is in d&" (or &"T) then
1:%e!I--?O as e--?O .. PROOF. Since (4.8) is assumed to be noncritical with respect to q), Theorem 1.1 implies the existence of the continuous linear operators :%e , o < e ~ eo. The existence of a K as specified in the lemma for the case when q) is ~( -::- 00, (0) or d &" is verified exactly as in the proof of Lemma 4.2. If
ORDINARY DIFFERENTIAL EQUATIONS
166
!1) = rY'T, then formula (1.9) yields :£ e for the special case (4.9) as
fo e[e- tAT -lrle-eA8j(t+s) ds, T
(4.10)
(:£el)(t) =
0 < e ~ eo·
Since [e-eAT _1]-1 is a continuous function of e on 0 < e ~ eo, :£ e1 is continuous for 0 < e ~ eo. We now show that :£el defined by (4.10) has a limit as e ~O. Since system (4.8) is noncritical with respect to rY'T, Lemma 4.3 implies A is nonsinguJar. Also, lime .... o+ e[e-eAT - l r 1 = -(AT)-I. This shows :£ e has a uniform bound on (0, eo] and completes the first part of the lemma .. To prove the last part of the lemma, let x
=
y+ef
1 and y will
satisfy
the equation
An application of the first part of the lemma to this equation shows that the unique solution of dflJ' (or flJ'T) approaches zero as e~O and the lemma is proved. The condition that flbe in drY' for lin dflJ' is equivalent to saying that the integral of 1 is bounded and in particular, implies (4.11)
M[f]
def
=
1 t
lim t .... co
Itl(s) ds = O. 0
In Chapter V, we show that (4.11) is sufficient to draw the same conclusion as in the last part of Lemma 4.4. EXERCISE 4.3. Discuss the manner in which the solutions of (4.9) approach the solutions of the equation x = O. Can you give any reason in the almost periodic or periodic case besides the one discussed in Lemma 4.4 for why all solutions approach the zero solution of x = 0 if f 1 is bounded? Lemma -4.4 and the same proofs as used in Theorems 2.1 and 3.1 yield an immediate extension of those results to the system (4.12)
x = e(Ax +l(t, x, e)),
where 1 satisfies the same conditions I\-S stated in those theorems. This result is. the basis for the method of averaging given in the next chapter and is therefore stated in detail-{or further reference. The o~rators 17+, 17-, constants K, <X and sets S(cp, G, S, e), U(cp, G, S, e) are as~U:ined to be the same as the ones given in (4.4) and (4.5). THEOREM 4.2. Suppose!1) is one of the classes!Jl( -co, co), dflJ' or flJ'T and system (4.8) is noncritical with respect to !1) for 0 < e ~ eo. If 1 is in
PERTURBATIONS OF NONCRITICAL LINEAR SYSTEMS
167
fl (') !l'ift(TJ, M), then there are S > 0, el > 0, f3 > 0 and a function x*(t, e) continuous in t, e for -oo
Furthermore, if the eigenvalues of A have nonzero real parts, then the mapping 7T_ is a homeomorphism of S(x*(·, e), a, S, e) onto (7T_Cn) (') B;/2K and (4.13)
ell ;;;; 2K 17T-[xU - x*(a, e]l e-el1(t-u) , t~ a, S(x*(', e), a, S, e), 0 < e ;;;; el. The mapping 7T+ is
Ix(t, a, xu, e) - x*(t,
for all XU - x*(a, e) in homeomorphism of U(x*(·, e), a, S, e) onto (7T+cn) (') B:;2K and (4.14)
Ix(t, a, xu, e) - x*(t,
ell ;;;; 2K 17T+[xU -
x*(a,
e)]1 eel1(t-u) ,
a
t ;;;; a,
for any xU-x*(a, e) in U(x*(" e), a, S, e), 0 < e;;;; el. The dependence of the stable and unstable manifolds of x*(·, e) upon a is exactly the same as in the statement of Theorem 3.1. Using the remark following Lemma 4.4, one easily sees that the conclusions of Theorem 4.2 remain valid in the periodic and almost periodic case for thesystehl
x = e(Ax + h(t) + !(t, x, e)),
(4.15)
f
provided that this bounded. Extensions of Theorem 4.2 to the case where f= f(t,x,e,/J.) and the vector e' = (e,/J.) is small are also easily given. EXERCISE
(4.16)
4.4.
Consider the equation
x = Ax + h(wt) +! (wt, x, e),
where w is a large parameter, the eigenvalues of A have negative real parts, h is in r!J T, foT h( s) ds = 0, and! is in fJJ T (') !l'i ft( TJ, M). Assuming Theorem
4.2 is valid when! depends upon more than one small parameter, prove there is an WI> 0, el > 0 such that (4.16) has an asymptotically stable periodic solution x*(', Wi e) in fJJT/w for w ~ WI, 0 < lei;;;; el and x*(·, w, e)-+O as w -+ 00, e -+ O. In particular, consider the forced van der Pol equation XI =X2, X2 = -XI
r
+ k(1 -
Xi}X2
+ eg(wt),
where k =1= 0, g is in fJJT and g(s) is bounded. Using Lemmas 1, 4.2 and 4.4 and the proof of Theorem 2.1, one can prove the following general result for systems which are coupled versions of the above types.
ORDINARY DIFFERENTIAL EQUATIONS
168
THEOREM 4.3. Suppose fl) is one of the classes &I( -00, (0), df!jJ or E!JT, u is an n-vector, u = (x, y, z) where x, y, z are nl-, n2-, na-vectors respectively, f = f (t, u, 8) = (X, Y, Z) is in fl) (\ fi'ijt('Y}, M), 8 = (I-', v), I-' a real scalar, B is an n2 X n2 matrix in E!JT, and A, 0 are constant nl X nl, na X na matrices, respectively, such that the system
x=
(4.17)
I-'Ax,
if =
B(t)y,
I-'Z =Oz,
is noncritical with respect to fl) for 0 < I-' ;'£ 1-'0. Then there are constants PI >0,1-'1 >0, VI >0 and a function u*(t, 8) continuous in t, 8 for -00 < t < 00, 0 < I-' ;'£ 1-'1. 0;'£ Ivi ;'£ VI, u*(t, 0) = 0, U*(·,8) in fl), lu*(·,8)1 < PI, such that u*(t, 8) is a solution of the equations
x=
(4.18)
if = I-'Z =
+ X(t, x, y, Z, 8)] l!(t)y + Y(t, x, y, Z,.8), Oz + Z(t, x, y, Z, 8), I-'[Ax
and is the only solution of (4.18) in fl) which has norm < pl. If fl) = dE!J, then m[u*(·, 8)] c m[j, B], 0 <1-';'£ 1-'1,0;,£ Ivi ;'£ VI· The stability properties of the solution u*(·, 8) of (4.18) are discussed exactly in the same manner as in Section 3. The stable and unstable manifolds of the solution u*(·, 8) can be characterized as in Theorems 4.1 and 4.2. In particular, if all eigenvalues of A, 0 have negative real parts and all characteristic exponents of if = B(t)y have neg~tive real parts, then the solution u*(·, 8) is exponentially asymptotically stable. If at least one of the eigenvalues of A or 0 or the characteristic exponents of if = B(t)y have a positive real part, then the solution u*(·, 8) is unstable. As mentioned earlier, it is difficult to remove the restriction in (4.18) that A be constant. On the other hand, it is possible to allow C to be a function of t, say 0 = O(t), provided either that the eigenvalues '\(t) of O(t) have real parts bounded away from zero or, more generally, that O(t) =diag(D(t), E(t»
where the eigenvalues of both D(t) and E(t) have real parts (not necessarily of the same sign) bounded away from zero. For references .concerning problems in the spirit of this section, see Hale [6].
IV.5. The Duffing Equation with Large Damping and Large Forcin/:,:
Consider the equation
ii + cif + y +.8by a=
(5.1)
where c > 0, b,
8,
B and
V
B cos vt,
> 0 are constants. The equivalent second order
PERTURBATIO.NS OF NONCRITICAL LINEAR SYSTEMS
169
system is (5.2)
if =Z, Z = -y - cz - eby3 + B cos vt.
The methods of this chapter and some elementary facts about quadratic forms will be used to prove there is an ro > 0 such that for any r ~ ro , there is an eo =. eo(r) > 0 such that, for lei < eo(r), system (5.2) has a unique periodic solution of period 21T/V in the disk B~ with center zero and radius r. This solution is uniformly asymptotically stable and any solution of (5.2) with initial value inB~, must approach this periodic solution as t~ 00. Consider the linear nonhomogeneous system (5.3)
if =z, Z = -y - cz + B cos vt.
Since the eigenvalues of the coefficient matrix of the homogeneous equation
if =z,
(5.4)
Z= -y-cz, have negative real parts, Theorem 1.1 implies there is a unique periodic solution of (5.3) of period 27T/V. If this solution is designated by yO(t), zO(t), then the transformation of variables y = yO(t) + u, z = zO(t) + v applied to (5.2) yields the equivalent system (5.5)
'Ii. 'Ii
= v + eh(t, u, v), = -u -cv +e/2(t, u, v),
where h, /2 are periodic in t of period 27T/V, continuous in t, u, v and continuously differentiable in u, v. Actually,h==O,/2 = -b(Yo(t) + u)3, but it is convenient for notational purposes to consider the more general system (5.5). Since c > 0, Theorem 2.1 implies there is a PI > 0 and e1 > 0 such that system (5.5) has a unique periodic solution (u*(t, e), v*(t, e)), lei ~e1 of period 27T/V in the disk B;l' this periodic solution is uniformly asymptotically stable, and u*(t, 0) = 0 = v*(t, 0). To prove the above mentioned result for (5.2), it is sufficient to show that, for any two disks B; and B;l' r1 < r,there exists an e2 > 0 such that, for Ie I < €2, any solution of (5.5) with initial value in must eventually enter and remain in the ball In fact, suppose ro is such that the periOdic solution (y*(t, e), z*(t., e)) of (5.2) given by y*(t, e) = yO(t) + u*(t, e), z*(t, e) = zO(t) + v*(t, e), lies in B;. for 0 ~ t ~ 21T/V and lei ~ e1· Choose r~ < pl. For any r ~ ro, the choice· eo(r) = min(e2 , e1) gives the desired result. We now prove the assertion about (5.5). Since the linear system (5.4) is asymptotically stable, it is intuitively clear that there should be a family
B;
B;,.
170
ORDINARY DIFFERENTIAL EQUATIONS
of ellipses encircling the origin so that the solutions of (5.4) cross the bound. aries of these ellipses from the outside to the inside with increasing time. If this is the case, then for any given ellipse one can choose e small so that the solutions of (5.5) also cross the boundary of this ellipse in the same direction as for the linear system (5.4). These ideas are now made precise by actually constructing such a family of ellipses. Consider the quadratic form c2+2 2
(5.6)
V(u, v)
= - - u 2 + 2uv + - v 2. C
C
This quadratic form is positive definite and therefore the level curves of this function are ellipses with center at the origin. The derivative of V along the solutions of (5.5) is (5.7)
V(u, v)
=
-2(u2
2 ] + v2) + 2e [ C2+2 -c-. uil + u/2 + vil + ~ v/2 .
For any positive r, rl, rl < r, choose positive constants CI > .C2 so that the region U contained between the curves V(u, v) = CI and V(u, v) = C2 contains the region between the boundaries of the disks (see Fig. 5.1). Suppose
B;, B;,
Figure IV.5.1
min(u 2 + v 2) = 0: for u, v ranging over the set V(u, v) = C2. Then 0: > 0 and one can find an e2 > 0 such that the right hand side of (5.7) is less than -0:/2 for 0 ~ lei ~ e2, - OCI < t < OCI and all (u, v) in U. For any (u O, v O) in U, the solution u(t), v(t), u(O) = u O, v(O) = v O, of (5.5) remains in the interior of the ellipse V(u, v) = CI and satisfies V(u(t), v(t))
~
t
V(u(O), v(O))
+ foV(U(S), v(s)) ds
~ V(u(O), v(O)) -
0:
2 t.
.
Since V is PQsitive in U, and the solption cannot reach the ellipse V(Ui~ = CI, there must be a to > 0 such that (u(t), v(t)) remains in the interio~ of the ellipse V(u, v) = C2 for all t ~ to. This proves the result.
171
PERTURBATIONS OF NONCRITICAL CASES
IV.6. Remarks and Extensions The technique presented in this chapter is applicable to many other types of problems. To illustrate this, let us give a brief abstract ·summary of the basic ideas. Suppose I is an interval in R n , A(t) is an n X n matrix funCtion continuous on I and §i, ~ are given Banach spaces of continuous n-vector functions on I. For every IE~, suppose the equation (6.1)
X=A(t)x + f(t)
has a unique solution:l{'f in -9; and :I{'~\~ ~ lb is a continuous linear operator. At the end of section 1, we said (lb, .@).was admissible if, for every fE .@, there is at least one solution of Eq. (6.1) in lb. Here, we are requiring uniqueness of the solution in lb~ In this case, we say (lb,~) is strongly admissible. Let rJ'I (I X R n ,Rn) = U:I X R n ~ R n ,f(t, x) continuous and continuously differentiable in x}. Let-
If 11 =sup{lf(t,x)1
+ laf(t,x)/axl,(t,x)inIXR n }.
If ¢ is in lb, let us suppose the function f(', ¢(.)) is in problem of the existence of solutions in lb of the e-quation (6.2)
~
and consider the
X=A(t)x + f(t,x)
If there exists a solution x of Eq. (6.2) in lb, then x must satisfy the equation
(6.3)
G(x,f)CWx -:I{'F(x,f) =
°
where:l{': !!). ~ lb is the continuous linear operator defined above and (6.4)
F:lbXrJ'I(IXRn,Rn)~ !!)
F(x,f)(t) = f(t, x (t)) ,
tEl.
Let us suppose the function F(x,f) is continuous together with its Frechet derivative. To solve Eq:' (6.3) for If 11 small, one can use the contraction mapping principle to obtain a generalization of Theorem 2.1. Or, one can use the Implicit Function Theorem in Banach spaces observing that G(O,O) = 0, aG(O,O)/ax = I, where G is defined in Eq. (6.3). The results may then be summarized as THEOREM 6.1. Suppose (lb,!!)) is strongly admissible for system (6.1) and the function F dermed in Relation (6.4) is continuous together with its Frechet derivative. Then there is a 0 > 0, 1/ > 0, and a unique function
172
ORDINARY DIFFERENTIAL EQUATIONS
x*:{fE ,?/l(J X R n ,Rn): 1/11
< o}+'~
such that x*(f) is continuous together with its derivative in I, x*(O) = O,x*(f) satisfies Eq. (6.2) and is the only solution of Eq. (6.2) in ~ with norm less than 1). Theorem 6.1 includes Theorem 2.1 in the case where I(t,x) has a continuous first derivative in x. We give a few examples of other applications of Theorem 6.1. Suppose J = [0,1], M,N are n X n constant matrices, A(t) is an n X n continuous matrix on J, I is a continuous n-vector function on J and consider the boundary value problem
x=A(t)x + I(t) ,
(6.5)
Mx(O)
tin J,
+ Nx(l) = 0
If X(t) is the fundamental matrix solution of the homogeneous equation, X(O) = J, then
x(t)
= X(t)x(O) +
f; X(t)X- 1(s)/(s)ds
satisfies the boundary conditions if and only if
[M + NX{1)] x(O) = -X(l) fo1 X-I (s)f(s)ds If/(s) = -X(s)X-1(l)b for a given vector b inR n , then
-X{1) fo1 X-I (s) I(s) ds
=b
This implies that Eq. (6.7) has a unique solution for every continuous function Ion J if and only if [M + NX{1)] -1 exists. If this inverse exists, then the boundary value problem (6.5)has a unique solution .:I[/given by
(~f)(t) = -X(t)[M + NX{1)] -1 X(1)fo1 X- 1(s)f(s)ds (6.6)
+
f:
X (t)x- 1(s)/(s).
These results are ~ummarized in the following: LEMMA 6.1. Let f$';
{I: [0,1] -+Rn,fcont.},1/1
= suptl/(t)l,
[1'= {x E f$ :Mx(O) + Nx(l) = O}.
PERTURBATIONS OF NONCRITICAL CASES
173
If X(t), X(O) = J, is a fundamental matnx solution of.x = AV)x, then (9J,~) is strongly admissible for the equation
x = A(t)x + let) if and only if [M + NX(1)] -1 exists. With this lemma, one can obtain the existence of a solution of the boundary value problem
x:;: A (t)x Mx(O)
+ let, x)
+ Nx(1) = 0
if the matrix [M + NX(l)] -1 exists and 1[(t,x)l, 13[(t,x)/3xl are small. As another illustration, let us briefly indicate how the saddle point property may be obtained in this way. Consider the equation
x =Ax + let)
(6.7)
where the eigenvalues of A have nonzero real parts and [is in ~dg9J( [0, 00», the space of continuous bounded n·vector functions on [0,00). We know Eq. (6.7) has at least one solution in ~ for every [in!P. However, the equation
x=Ax
(6.8)
has a finite dimensional subspace ~ of solutions which are in !P and so the solutions in !P of Eq. (6.7) are not unique unless g'fs = {OJ. The pair (!P,'fP;) is admissible but generally not strongly admissible. On the other hand, we can define 9J ~ !P so that (9J,fP) is strongly admissible. In fact, let trs :!P -+ g'fs be a continuous projection operator and define 9J = (J - 7rs )!P. The pair (9J,!P) is then strongly admissible and there is a continuous linear operator %:@ -+ r9J sllch that%[ is the unique solution of Eq. (6.7) with 7r/Y[ = O. The operator .Jf; of sourse, depends on the projection 7rs , but one can clearly choose trs so that%[ consists of the sum of the two integrals in Formula (6.4) of Chapter III. With this definition of .Jr; every solution of Eq. (6.7) in !P is given by
(6.9)
x
= 7rs e4 'xo + %f
Let us now, consider the nonlinear problem
x=Ax+ [(x)
(6.10)
where [(0) =' 0, 3[(0)/3 x = 0 and try to prove the existence of a saddle point at x = O. In particular, let us obtain the stable manifold. We know from Eq. (6.9) that every solution in -!P must satisfy
x
= 1TseA 'xo +%[(x)
If y = x - 1Ts~' xo, then y E 9J and we have
174
ORDINARY DIFFERENTIAL EQUATIONS
(6.11) One can now use the Implicit Function Theorem to determine y*(f, xo) in as a continuously differentiable function of !, xo, y*(O,O) = 0, G( y*(f, xo),f, xo) = O. This shows there is a family of solutions for each !, which are bounded and remain in a neighborhood of zero. The dimension of the family is the dimension of the family 7TseA • Xo; that is, the dimension of the stable manifold of = A x. To show these solutions approach zero, one can put more restrictions on the space ftJ and repeat tlJe same argument. This idea can be generalized to the abstract case where (~,~ is admissible and not strongly admissible (see, for example, Antosiewicz f2], Hartman [1]). ~
x
CHAPTER V Simple Oscillatory Phenomena and the Method of Averaging
In Chapter II, it was shown how the Poincare-Bendixson theory could be used to determine the existence and stability properties of periodic orbits of autonomous two dimensional systems. For systems of higher dimension and even for nonautonomous two dimensional systems, the methods of Chapter II are of no assistance in the discussion of the existence of periodic or almost periodic solutions. In Chapter IV, the existence of periodic and almost periodic solutions were discussed for systems of differential equations which were perturbations of linear systems of arbitrary order. On the other hand, it was assumed that the trivial solution was the only solution of the linear system which belonged to the class of desired solutions; that is, the system was noncritical. Applications are very common for which the linear part of the system contains nontrivial periodic solutions and the system is critical with respect to some class of forcing functions. The methods of Chapter IV are not applicable directly to such problems. However, there may be appropriate transformations of yariabIes which bring a critical system into the framework of Chapter IV. This is precisely the basis for the method of averaging discussed in Section 3 below. The method of averaging is a general method for determining sufficient conditions for the existence and stability of periodic and almost periodic solutions of a class of nonlinear vector differential equations which contain a small parameter. It is possible to discuss the existence of periodic solutions without invoking the results of Chapter IV. In fact, a much more general theory is given in Chapter VIII for the periodic case. Before discussirig the method of averaging, general properties of conservative systems and simple nonconservative systems are treated in Sections 1 and 2. The remaining sections of the chapter are devoted to specific applications.
175
176
ORDINARY DIFFERENTIAL EQUATIONS
V.I. Conservative Systems Suppose f: Rn ~ Rn is continuous and for any Xo in Rn the system
x =f(x),
(1.1)
has a unique solution x(t) = x(t, xo) with x(O, xo) = xo. A function E: Dc Rn~R is said to be anlintegral of (l.I)1on a region Dc Rn if E is continuous together with its first partial derivatives, E is not constant on any open set in D, and E(x(t) = constant along the solutions of (l.I).ISince E is assumed to have continuous first derivatives, the last ro ert is e uivalent to [oE(x(t»/ox]f(x(t» = o. S stem 1.1 is said to be nservative if it has an integral E on Rn.\The orbits in Rn of a conservative system must therefore lie on level curves of the integral E. Suppose x = (Xl, ... , xn), E is an integralof (l.I) on D and xo in D is such that oE(xO)/ox n =I- O. Let c = E(xO). From the implicit function theorem, it follows that the equation E(x) = c can be solved for Xn as a function x! of the Xk, k < n, xO and x in a sufficiently small neighborhood U of x O. Since E is assumed to be a first integral, E(x(t» = c if x(t) is the solution of (1.1.) with _x(O) = x O. Substituting x! in (l.I) results in a system of (n -1) equations for the determination of the solution of (l.I) through x O• The dimension of (l.I) is therefore decreased by one on U. In particular, if n = 2, the existence of an integral reduces the solution of the equation to a quadrature. In the following, the notation introduced in Chapter I is employed by letting y+ = y+(x), y- = y-(x) denote respectively the positive, negative orbit of (l.I) through x, w(y+) and ex(y-) denote the w- and ex-limit sets, respectively, of the orbit y+, y-. LEMMA l.l. Suppose E is an integral of (l.I) on an open set D containing an equilibrium point xO of (l.I). There is no neighborhood U of xO for which xO belongs to one of the sets w(y+(x» or ex(y-(x» for all x in U. PROOF. If there are a sequence of {tn}, tn ~ 00 as n ~ 00, and an xl in D such that x(tn, xl) ~ xO as -n ~ 00, then continuity of E implies E(x(t, Xl» = E(xO) for all t. The same statement is true for tn ~ - 00. Therefore, if there were a neighborhood U of xO such that xO belongs to one of the sets w(y+(x» or ex(y-(x» for aU x in U, then E would be constant on U. Since this is contrary to the definition of an integral, the lemma is proved.
Lemma 1.1 implies in particular that an equilibrium point of a cOllservative system can never be asymptotically stable. -, c. LEMMA l.2.
Suppose E is an integral of (l.I) in a bounded, open neigh-
SIMPLE OSCILLATORY PHENOMENA, METHOD OF AVERAGING borhood D of an equilibrium point x = 0 of (1.1). If E(O) = 0 and E(x) x#-O in D, then x = 0 is a stable equilibrium point.
177
> 0 for
for x#-O in D, then for any e > 0, O
If E(O)
= 0,
E(x)
>0
«~minlxl=£.XinDE(x»O. Choose
a.
LEMMA 1.3. For n = 2, all orbits in a neighborhood of a stable isolated equilibrium point of a conservative system must be periodic orbits and the equilibrium point is a center. PROOF. In this proof, a bar over a set denotes closure. Suppose x = 0 is an isolated stable equilibrium point of (1.1) for n = 2. Then there are neighborhoods U, V, of zero such that P\ {O} is equilibrium point free and for every x in U, the positive orbit y+ = y+(x) through x belongs to P. From the Poincare-Bendixson theory, the w-limit set w(y+(x» of y+(x) must be either {O} or a periodic orbit. Lemma 1.1 implies w(y+(x» cannot be {O} for 'every x in U. If there is an x in U such that r = w(y+(x» = y+(x)\y+(x) is a periodic orbit, then r is asymptotically stable from either the inside or outside and, thus, there is an open set on which the integral E is constant. This ccntradiction shows that any trajectory which does nat approach zero is a .• periodic orbit. Since every periodic orbit obviously has zero in its interior, this proves the lemma. A very important class of conservative systems are Hamiltonian systems with n degrees of freedom. If q = (ql, ... , qn) are the generalized position coordinates of n- particles and p = (PI, ... , Pn) are the generalized momentum, H(p, q) = T(p) V(q) where T is the kinetic energy and V is the potential energy, then the equations of motion are
+
oH op
q=-,
(1.2)
.
oH
p=--.
oq
This is a special case of system (1.1) with x = (q, pl. The Hamiltonian H(p, q) is an integral of the system (1.2). The kinetic energy T(p) is assumed always with T(O) = 0, oT(O)/op = O. If oT(p)/op #- 0 for p #- 0, the positive if p extreme points qO of V(q) therefore yield the equilibrium points (qO, 0) of (1.2). For Hamiltonian systems, we can 'prove
*.0
LEMMA 1.4. Suppose q = 0 is an extreme point of the potential energy V(q) with V(O) = O. If zero is locally an absolute minimum of V(q), then (0,0) is a stable point of equilibrium of (1.2): The equilibrium point (0,0) is unstable if zero is not a minimum of V(q) if T(P) and V(q) have the form T(P) = T2(P) + T3(P), V(q) = Vk(q) + Vk+ l(q) where T2 is a positive definite quadratic
176
ORDINARY DIFFERENTIAL EQUATIONS
V.I. ConServative Systems
Suppose f: Rn _ Rn is continuous and for any Xo in Rn the system (1.1)
i=f(x),
has a unique solution x t = x t, Xo with x(O, xo) = xo. A function E: Dc: Rn _ R is said to be an inte raJ, 0 LIon a region D c: Rn if E is contino uous together with its first partial derivatives, E is not constant on an onset in D, and Ext = constant alon the solutions of 1.1 . Since E is assumed to have continuous first derivatives, the last ro ert is e uivalent to [8E(x(t»/8x]f(x(t» = O. S stem l.l is said to be ervative if it has an integral E on Rn.IThe orbits in Rn of a conservative system must therefore lie on level curves of the integral E. Suppose x = (Xl, ... , xn), E is an integral of (1.1) on D and xO in D is such that 8E(xO)/8xn =1= O. Let c = E(xO). From the implicit function theorem, it follows that the equation E(x) = c can be solved for Xn as a function x! of the Xk, k < n, xO and x in a sufficiently small neighborhood U of x O. Since E is assumed to be a first integral, E(x(t» = c if x(t) is the solution of (1.1.) with . x(O) = x O. Substituting x! in (1.1) results in a system of (n -I) equations for the determination of the solution of (1.1) through xO• The dimension of (1.1) is therefore decreased by one on U. In particular, if n = 2, the existence of an integral reduces the solution of the equation to a quadrature. In the following, the notation introduced in Chapter I is employed by letting y+ = y+(x), y- = y-(x) denote respectively the positive, negative orbit of (1.1) through x, w(y+) and oc(y-) denote the w- and oc-limit sets, respectively, ofthe orbit y+, y-. LEMMA 1.1. Suppose E is an integral of (1.1) on an open set D contain· ing an equilibrium point xO of (1.1). There is no neighborhood U of xO for which xO belongs to one of the sets w(y+(x» or oc(y-(x» for all x in U. PROOF. If there are a sequence of {tn}, tn- 00 as n_ 00, and an Xl in D such that x(tn, xl) _xo as ·n_ 00, then continuity of E implies E(x(t, xl» = E(xO) for all t. The same statement is true for tn _ - 00. There· fore, if there were a neighborhood U of xO such that xO belongs to one of the sets w(y+(x» or oc(y-(x» for aU x in U, then E would be constant on U. Since this is contrary to the definition of an integral, the lemma is proved.
Lemma 1.1 implies in particular that an equilibrium point of a co~erva. tive system can never be asymptotically stable. ~ , LEMMA 1.2.
Suppose E is an integral of (1.1) in a bounded, open neigh.
SIMPLE OSCILLATORY PHENOMENA, METHOD OF AVERAGING borhood D of an equilibrium point x = 0 of (1.1). If E(O) = 0 and E(x) x =f= 0 in D, then x = 0 is a stable equilibrium point.
177
> 0 for
for x * 0 in D, then for any e > 0, IX~minlxl=e.xinDE(x»O. Choose O<S<e so that {x:lxl~S}cD, maxlxl ~8 E(x) < IX. Since E is an integral this implies Ix(t, xO)1 < e, t ~ 0, if Ixol < S. Thus, x = 0 is stable and the lemma is proved. PROOF.
If E(O) = 0, E(x)
>0
LEMMA 1.3. For n = 2, all orbits in a neighborhood of a stable isolated equilibrium point of a conservative system must be periodic orbits and the equilibrium point is a center. PROOF. In this proof, a bar over a set denotes closure. Suppose x = 0 is an isolated stable equilibrium point of (1.1) for n = 2. Then there are neighborhoods U, V, of zero such that F\ {O} is equilibrium point free and for every x in U, the positive orbit y+ = y+(x) through x belongs to F. From the Poincare-Bendixson theory, the w-limit set w(y+(x» of y+(x) must be either {O} or a periodic orbit. Lemma 1.1 implies w(y+(x» cannot be {O} for 'every x in U. If there is an x in U such that r = w(y+(x» = y+(x)\y+(x) is a periodic orbit, then r is asymptotically stable from either the inside or outside and, thus, there is an open set on which the integral E is constant. This ccntradiction shows that any trajectory which does nat approach zero is a .• periodic orbit. Since every periodic orbit obviously has zero in its interior, this proves the lemma. A very important class of conservative systems are Hamiltonian systems with n degrees of freedom. If q = (ql, ... , qn) are the generalized position coordinates of n- particles and p = (PI, ... , Pn) are the generalized momentum, H(p, q) = T(p) + V(q) where T is the kinetic energy and V is the potential energy, then the equations of motion are (1.2)
oH
q=-, op
oH
jJ=--. oq
This is a special case of system (1.1) with x = (q, p). The Hamiltonian H(p, q) is an integral of the system (1.2). The kinetic energy T(p) is assumed always positive if p *'0 with T(O) = 0, oT(O)/op = O. If oT(p)/op * 0 for p * 0, the extreme points qO of V(q) therefore yield the equilibrium points (qO, 0) of (1.2). For Hamiltonian systems, we can prove
°
LEMMA 1.4. Suppose q = is an extreme point of the potential energy V(q) with V(O) = 0. If zero is locally an absolute minimum of V(q), then (0,0) is a stable point of equilibrium of (1.2r The equilibrium point (O,O)is unstable if zero is not a minimum of V(q) if T(P) and V(q) have the form T(P) = T 2(P) + T3(P), V(q) = Vk(q) + Vk+ 1(q) where T2 is a positive definite quadratic
178
ORDINARY DIFFERENTIAL EQUATIONS
forIll, Vk is a homogeneous polynomial of degree k ~ 2, Tg(P) = o(lpI2), Vk+l(q) = o(lqlk) as Ipl, Iql ~ and there is a neighborhood U of(O,O) such that
°
°
nu = {(P,q) E
U:H(p,q) < O} =J:. cf>
and Vk(q) < if (p,q}is in nu. PROOF. The assertion concerning stability is an immediate consequence of Lemma 1.2. Suppose zero is not a minimum of V(q). Let W(p,q) = p'q. The point (O,O) is a bound~ point of nu. Using the fact that H is an integral of (1.2), the derivative 'W(p,q) along the solutions of (1.2); is easily seen to be W(p,q) = 2T2(P) - k Vk(q)
+ ... ,
where· •• designates terms which are o(lpI2) and o(lqlk) as Ipl,lql ~ 0. In nu, T2(P) > and Vk(q) < 0. Furthermore, one can choose the neighborhood U of (0,0) sufficiently small that W(p,q) > in nu. Since W(p,q) > 0, W(p,q) > in nu, any solution with initial value in nu must leave the set nu through the boundary of U since the boundary of nu in the interior of U consists of points where W(p,q) = 0, or H(p,q) = 0. Since (0,0) is in the boundary of nu, this proves instability and the lemma. More specific information on the nature of the integral curves for second order conservative systems can be given. Consider the second order scalar equation
°
°
°
(1.3)
ii
+ g(u) =0,
or the equivalent system
u=v,
(1.4)
v= -g(u), where g is continuous and a uniqueness theorem holds for (1.4). System (1.4) is a Hamiltonian system with the Hamiltonian function or total energy given by E(u, v) = v2J2
+ G(u)
where G(u) = IoU g(s)ds. The orbits of solutions of
(1.4) in the (u, v)-plane must lie on the level curves of the function E(u, v); that is, the curves described by E(u, v) = h, a constant. The equilibrium points of (1.4) are points (uO,O) where g(u O) = O. If G(u) has an absolute minimum at u O, then (u O, 0) is stable (Lemma 1.2) and all orbits in a neighborhood of (u O, 0) must be periodic orbits (Lemma 1.3); that is, (u O, 0) is a center. Since the solutions of E(u,v) = h, a constant, are (1.5)
v = ±J2[h - G(u)] ,
it follows that any isolated equilibrium point (U O, 0) such that u O is not a J2[h - G(u)], minimum of G must be unstable. In fact, the curves v =
+
SIMPLE. OSCILLATORY PHENOMENA, METHOD OF AVERAGING
179
v = -J2[h - G(u)] are homeomorphic images of a segment of the real line in a neighborhood of (u O, O).·If a solution starting on these curves does not leave a neighborhood of (uO, 0), then the w-limit set of the solution curve would be an equilibrium point. Since (uO, 0) is assumed isolated, this immediately gives
a contradiction. A point u O is a local absolute minimum of G(u) if g(u) < 0 for u < u O and g(u) > 0 for u > u O and u in a neighborhood of u O• The reverse inequalities apply for a local absolute maximum of G(u). If G(u) has a local absolute maximum at u O, then the equilibrium point (U O, 0) is a saddle point in the sense that the set of all solutions which remain in a small neighborhood of (u O , 0) for t ;:,;; 0 (t ~ 0) must lie on an arc passing through (u O , 0). This follows directly from formula (1.5). We can therefore state LEMMA 1.5. The stable equilibrium points of (1.4) are centers and all of the unstable equilibrium points of (1.4) are saddle points if the only extreme points of G(u) are local absolute minimum and local absolute maximum.
Particular examples illustrate how easily information about a two conservative system is obtained without any computations whatsoever. The sketch of the level curves of E are easily deduced using (1.5). dimen~ional
Example 1.1. Suppose the functionG(u) has thegraphshowninFig.1.1a with A, B, G, D being extreme points of G. The orbits of solution curves are sketched in Fig. 1.1 b, all curves of course being symmetric with respect to the u-axis. The equilibrium points corresponding to A, B, G, D are labeled as A, B, G, D on the phase plane also. The points A, G are centers, B is a saddle point and D is like the coalescence of a saddle point and a center. The curves joiping B to Band D to D in Fig. LIb are called separatrices. A separatrix is
Figure V.I.I
180
ORDINARY DIFFERENTIAL EQUATIONS
a curVe consisting of orbits of (1.3) which divides the plane into two parts and there is a neighborhood of this curve such that not all orbits in this neighborhood have the same qualitative behavior. Separatrices must therefore always pass through unstable equilibrium points . .Example 1.2. Suppose equation (1.3) is the equation for the motion of a pendulum of length l in a vacuum and u is the angle which the pendulum makes with the vertical. If g is the acceleration due to gravity, then g(u) = k 2 sin u, k 2 = gIl, G(u) = k 2(1 - cos u) and G(u) has the graph shown in Fig. 1.2a. The curves E(u, v) = 71, clearly have the form shown in Fig. 1.2b. Explain the physical meaning of each of the orbits in Fig. 1.2b.
v
tI
I (b)
Figure V.I.2
It is not difficult to determine implicit formulas for the periods of the periodic solutions of equation (1.3) for an arbitrary g(x). Any periodic solution has an orbit which must be a closed curve and conversely. From (1.5), it follows that a closed orbit must intersect the u axis at two points (a, 0), (b, 0), a < b, and must be symmetrical with respect to the u-axis. If T is the period of the periodic solution, then T = 2w where w is the time to traverse that part ofthe orbit where v> O. Since v is given by (1.5) it follows from (1.4) that 'b du (1.6) T=2{ )2(71, -G(u»' If G(u) is even in u and the periodic orbit encircles the origin, then symmetry implies a = -b and
SIMPLE OSCILLATORY PHENOMENA, METHOD OF AVERAGING T=4
(1.7)
lS1"
fb
du o J2(h - G(U)) .
For the pendulum equation, example 1.2, g(u) = k 2 sin u, G(u) = - cos u), and if u(O) = b < 7T, v(O) = 0, then h = E(u(t), v(t)) = E(u(O), v(O)) = E(b, 0) = k 2 (1 - cos b). Consequently, the period T of the periodic orbit passing through this point (b, 0) is k2 (1
b
T. = 4
fo
2
du [2k2(COS U _ cos b)]'/2 =
du
b
Ie fo
[.sm2 (b) 2
-
. (U)]
sm2
'/2·
-
2
If sin(u/2) = (sin~) sin(b/2), then
(1.S)
T=~ k
(12 0
d~ [1 - sin 2
G)
sin2
r'
~
This integral cannot be evaluated in terms of elementary functions, but if b is sufficiently small, then it is easy to find an approximate value of the period. Expanding in series, we obtain
if b is sufficiently small. To the .first approximation T = 27T/k; that is, the frequency k is approximately J(iji, which is almost the first lesson in every course in elementary mechanics. From this example and the above computation, it is seen that the period of a periodic solution of an autonomous differential equation may vary from one solution to another (contrast this fact to the linear equation).
+
EXERCISE 1.1. For g(u) = u you 3 , show that the period T(b, yo) given by formula (I.?) i~ a decreasing function of b (increasing function. of b) if yo> 0 (yo < 0). The first situation is called a hard spring and the second a soft spring. Hard implies the frequency increases with the amplitude b. Soft implies the frequency decreases with the amplitude b. In the general case, if (uO, 0) is a stable equilibrium point of (1.3), then the restoring force g(u) is said to correspond to a hard spring (soft spring) at u O if the frequency of the periodic solutions in a neighborhood of (U O, 0) is an increasing (decreasing) function of the distance of the periodic orbit from (U O, 0).
182
ORDINARY DIFFERENTIAL EQUATIONS
Example 1.3. Consider a pendulum of mass m and length l constrained to oscillate in a plane rotating with angular velocity w about a vertical line. If'lt denotes the angular deviation of the pendulum from the vertical line (see Fig. L3), the moment of centrifugal force is mw 2l 2 sin 'It cos 'It, the
mg
Figure V.l.3
moment of the force due to gravity is mgl sin 'It and the moment of inertia is 1= ml 2 . The differential equation for the motion is (L9)
Iii - mw 2l 2 sin 'It cos 'It
+ mgl sin u = o.
If I-' = mw 2l 2 /1 and A = g/w 2l, then this equation is equivalent to the system
(LI0)
u=v, 'Ii = I-'(cos 'Ii - A) sin u,
which is a special case of (1.4) with g(u) = g(u, A) = -I-'(cosu - A)sin u. The dependence of g upon Ais emphasized since the number of equilibrium points of (LI0) depends upon A. The equilibrium points of (1.10) are points (u O, 0) with g(uO, A) = 0 and are plotted in Fig. 1.4. The shaded regions correspond to g(u, A) < O. For any given A, the equilibrium points are (0,0), (77",0) and (COS-1 A,0), the last one of course appearing only if IAI < L For IAI =F 1, Lemma L5 implies the points on the curves labeled in Fig. 1.4 with black dots (circles) are stabl~ (unstable), the stable points being centers and the unstable points being saddle points. From this diagram one sees that when 0 if<. A < 1, the stable equilibrium points are not (0,0) or (77",0) but the points (cds-lA, 0). Physically, one can have A < 1 if the angular velocity w is large enough. Analyze the behavior of the equilibrium points for A= 1, A = -L
SIMPLE OSCILLATORY PHENOMENA, METHOD OF AVERAGING
183
u
Figure V.l.4
Figure V.l.5
An integral for this system is easily seen to be v I-' ="2 -2" sin2 u 2
E(u, v)
1-'>" cos u.
Suppose 0 < >.. < 1. The equilibrium points (0, 0) and (0, 7T) are saddle points
184
ORDINARY DIFFERENTIAL EQUATIONS
and the level curves of E(u, v) passing through these points are, respectively,
+ 2A(COS u -1)], p.[sin2 u + 2A(COS u + 1)].
v2 = p.[sin2 u v2 =
Both of these curves contain the points (cos-1 A, 0) in their interiors and the latter one also passes through (-TT, 0). A sketch of the orbits is given in Fig. 1.5. The two centers correspond to the two values of u for which cos u = A. Interpret the physical meaning of all of the orbits in Fig. 1.5 and also sketch the orbits in the phase plane when A > 1.
V.2. Nonconservative Second Order Equations-Limit Cycles Up to this point, three different types of oscillations which occur in second order real autonomous differential systems have been discussed. For linear systems, there can be periodic solutions if and only if the elements of the coefficient matrix assume very special values and, in such a situation, all solutions are periodic with exactly the same period. For second order conservative systems, there generally are periodic solutions. These periodic solutions occur as a member of a family of such solutions each of which is uniquely determined by the initial conditions. The period in general varies with the initial conditions but not all solutions need be periodic. In Section 1.7, artificial examples of second order systems were introduced for which there was an isolated periodic orbit to which all solutions except the equilibrium solution tend as t --+ 00. In Chapter II, as an application of the PoincareBendixson theory, it was shown that the same situation occurs for a large class of second order systems which include as a special case the van del' Pol equation. Such an isolated asymptotically stable periodic orbit was termed a limit cycle and is also sometimes referred to as a self-sustained oscillation. The phenomena exhibited by such systems is completely different from a conservative system since the periodic motion is determined by the differential equation itself in the sense that the differential equation determines a region of the plane in which the w-limit set of any positive orbit in this region is the periodic orbit. In many applications, these latter systems are more important since the qualitative behavior of the solutions is less sensitive to perturbations in the differential equation than conservative systems. To illustrate how the structure of the solutions change when a conservative system is subjected to small perturbations, consider the problem of the ordinary pendulum subje~ted to a frictional force proportional to the rate of the change of the angle u with respect to the vertical. The equation of motion is (2.1)
if
+,Bu + k 2 sin u =
0,
SIMPLE OSCILLATORY PHENOMENA, METHOD OF AVERAGING
where f1
> 0 is a
185
constant. The equivalent system is
u=v,
(2.2)
v
=
-k 2 sin u - f1v.
+
If E(u, v) = v 2/2 k 2(1 - cos u) is the energy of the system, then dE/dt along the solution of (2.2) is dE/dt = -f1v 2 ~ O. We first show that v(t) -+0 for all solutions of (2.2). Since dE/dt ~ 0, E is nonincreasing along solutions of (2.2) and v(t) is bounded. Equation (2.2) and the boundedness of sin u(t) implies v(t) is bounded. If v(t) does not approach zero as t -+ 00, then there are positive numbers e, S and a sequence tn, n = 1, 2, ... , tn -+ 00 as n -+ 00 such that the intervals In = [tn - S, tn S] are nonoverlapping and V2(t) > e for t in In, n = 1, 2, .... Let p(t) be the integer such that tn < t for all n ~ p(t). Then,
+
E(u(t), v(t)) - E(u(O), v(O))
t(dE) ds
= fo dt ~
t
- fo f1v2(s)ds
~ -2f1Sep(t).
Since p(t) -+ 00 as t -+ 00 and E(u(t), v(t)) is bounded, this contradicts the fact that v(t) does not approach zero as t -+ 00. Thus, v(t) -+0 as t -+ 00. With E(u(t), v(t)) nonincreasing and v(t) -+0 as t -+ 00, the nature of the level curves of E implies that every solution of (2.2) is bounded. Also, since the energy is nonincreasing and bounded below, it must approach a constant as t -+ 00. Since E is continuous and the limit set of any solution is invariant, the limit set must lie on a level curve of E; that is, the limit set of eachsolution must have E = 0 and, therefore, v = O. Since the limit set of each solution is invariant and must have v = 0, it follows that u is either 0, ±17", ±217", etc. All solutions of (2.2) bounded implies they have a nonempty limit set and, therefore, each solution of (2.2) must approach one of the equilibrium points (0,0), (17",0), (-17",0), etc. To understand the qualitative behavior of the soluti6ns, it re:glains only to discuss the stability properties of the equilibrium points and use the properties of the level curves of E(u, v) depicted in Fig.1.2b. The linear variational equation relative to (0, 0) is
u=v, v
=
-k 2u - f1v,
and the eigenvalues of the coefficient matrix have negative real parts. Therefore, by the theorem of Liapunov (Theorem III.2.4), the origin for the nonlinear system is asymptotically stable. The linear variational equation
186
ORDINARY DIFFERENTIAL EQUATIONS
relative to the equilibrium point (1T, 0) is
u=v, iJ = k 2u - {3v, and the eigenvalues -of the coefficient matrix are real, with one positive and one negative. Since the saddle point property is preserved (Theorem IlL6.1), this equilibrium point is unstable and only two orbits approach this point as t _ 00. Periodicity of sin u yields the stability properties of the ot4er equilibrium points. This information together with the level curves of E(u, v) allows one to sketch the approximate orbits in the phase plane. These orbits are shown in Fig. 2.1 fbrthe case {3 < 2k. v I I
Figure V.2.1
This example illustrates very clearly that a change in the differential equation of a conservative system by the introduction of a small nonconservtive term alters the qualitative behavior of the phase portrait of the solutions tremendously. It also indicates that a limit cycle will not occur by the introduction of a truly dissipative or frictional term. In such a case, energy is always taken from the system. To obtain a limit cycle in an equation, there must be a complicated transfer of energy between the system and the external forces. This is exactly the property of van der Pol's equation where the dissipative term t/YU is such that ~ depends upon u and does not have constant sign.
"
Basic problems in the theory of nonlinear oscillations in aut.lnomous systems with one degree of freedotn are to determine conditions wid~r which the differential equation has limit cycles, to determine the number of limit cycl~s and to determine approximately the characteristics (period, amplitude,
SIMPLE OSCILLATORY I'HENO)lENA, METHOD OF AVERAGING
187
shape) of the limit cycles. One useful tool in two dimensions is the PoincareBendixson theory of Section II.I. Other important methods involve perturbation techniques, but can be proved to be applicable in general only when the equation contains either a small or a large parameter. The great advantage of such methods is the fact that the dimension of the system is not important and the equations may depend explicitly upon time in a complicated manner. One general perturbation technique known as method of averaging will be described in the next section. As motivation for this theory, consider the van der Pol equation
u-8(I-u 2 )u+u=0,
(2.3)
where 8> 0 is a small para.meter. From Theorem I1.1.6, this equation has a unique limit cycle for every 8 > O. Our immediate goal is to determine the approximate amplitude and period of this solution as 8 -+0. Equation (2.3) is equivalent to the system (2.4)
u=v, v = -u + 8(1 -u2 )v. For
8
= 0, the general solution of (2.4) is u = r cos (), v = -r sin (),
(2.5)
+
where () = t 4>, 4> and r are arbitrary constants. If the periodic solution of (2.4) is a continuous function of 8, then the orbit of this solution should be close to one of the circles described by (2.5) by letting r = constant and () vary from 0 to 217. The first basic problem is to determine which constant value of r is the candidate for generating the periodic solution of (2.4) for 8 =1= O. If r, () are new coordinates, then (2.4) is transformed into the system (2.6)
(a)
() = 1
(b)
r=
+ 8(1 -
r2
cos 2
())
sin () cos (),
8(1 - r2 cos 2 ())r sin 2
().
+
For r in a compact set, 8 can be chosen so small that 1 8(1 - r2 cos 2 ()) . sin () cos () >0 and the orbits of the solutions described by (2.6) are given by the solutions of the scalar equation dr
(2.7)
d()
= 8g(r, (), 8)
where (2.8)
g(r, (), 8) =
1
+
(1 - r2 cos 2 ())r sin 2 () . ()' 8(1 - r2 cos 2 ()) sm () cos
188
ORDINARY· DIFFERENTIAL EQUA.TIONS
The problem of determining periodic solutions of van der Pol's equation for B small is equivalent to finding periodic solutions '1'*(8, B) of (2.7) of period 211 in 8. In fact, if '1'*(8, B) is. such a 211-periodic solution of (2.7) and 8*(t, B), 8*(0, B) = 0, is the solution of the equation
8 = 1 + B(1 -
(2.9)
['1'*(8, B)]2 cos 2 8) sin.(J cos 8,
then u(t) = 'I'*(8*(t, B), B) cos 8*(t, e), v(t) = -'I'*(8*(t, B), B) sin 8*(t, B) is a solution of (2.4). Let T be the unique solution of the equation 8*(T, B) = 211. Uniqueness of solutions of equation (2.9) then implies 8*(t T, B) = 8*(t, B) 2~ for all t. Therefore, u(t + T) = u(t), v(t T) = v(t) for all t and u, v is a 'periodic solution of (2.4) of period T. Conversely, if u, v is a periodic solution of (2.4) of period T, then 'I'(t T) = 'I'(t) and one can choose 8 so that 8(t T) = 8(t) 211 for all t, The functions 'I' and 8 satisfy (2.6) and, for B small, t can be expressed as a function of 8 to obtain 'I' as a periodic function of 211 in 8. Clearly, this function satisfies (2.7) .. Let us attempt to determine a solution of (2.7) of the ferm
+
+
(2.10)
+
+
+
+
'I'
= P + B'I'(I) (8, p) + B2'1'(2)(8, p) + "',
where each '1'(1)(8, p) is required to be 211-periodic in 8, and p is a constant. If this expression is substituted into (2.7) and like powers of B are equated then 01'(1)(8, p)
a8
= g(p, 8,0).
This equation will have a 211-periodic solution in 8 if and only if
I:"
g(p, 8, 0 )d8 =
O. If this expression is not zero, then it is called a secular term. If a secular term appears, then a solution of the form (2.10) is not possible for an arbitrary constant p. The constant p must be selected so that it at least satisfies the equation
I:"g(p, 8, 0)d8=0.
Having determined
p so
that this relation is
satisfied (if possible), then '1'(1)(8, p) can be computed and one can proceed to the determination of '1'(2)(8, pl. However, the same type of equation results for '1'(2)(8, p) and more secular terms may appear. One way to overcome thl;lse difficulties is to use a method divised by Poincare which consists in the following: let 'I' be expanded in a series as in (2.10) and also let p be expanded in a series in B as p = po BPI B2p2 and apply the same process as before successively determining po , Pl' P2 , ••• in such a way as to eliminate all secular terms. If the PI can be so chosen, then one obtains a periodic solution of (2.7). This method will be discussed in a much more general settjig in a later chapter. . '. We discuss in somewhat more detail another method due to Krylov and Bogoliubov which is generalized in the next section. Consider (2.10) as II>
+
+
+ ...
SIMPLE OSCILLATORY PHENOMENA, METHOD OF AVERAGING
189
transformation of variables taking r into p and try to determine r(1)(O, p), r(2)(0, p), ... and functions R(1)(p), R(2)(p), ... so that the differential equation for p is autonomous and given by dp dO = eR(1)(p)
(2.11)
+ e2R(2)(p) + ....
If such a transformation can be foqnd, then the 21T-periodic solutions of (2.7) coincide with the equilibrium points of (2.11). Also, the transient behavior of the solutions of (2.7) will be obtainable from (2.11). If (2.10) is substituted into (2.7), p is required to satisfy (2.11) and g(r, 0, e)
=
00
L ekg(k)(r, 0),
k=O
then (2.12)
R(1)(p)
+ or(l~~, p) = g(O)(p, 0),
R(2)(p)
+ or(2~:, p) = _ o;~l) (0, p)R(l)(p) + g(l)(p, 0) og(O)
+ Tr (p, O)r(1)(O, p). Since g(O)(p, 0) solution. given by
= g(p, 0, 0),
the first equation in (2.12) always has a
f 21T 1
(2.13)
R(1)(p)
=-
2"
g(O, p, 0) dO,
0
r(l)(O, p)
=
J
[g(p, 0,0) - R(1)(p)] dO,
where" f" denotes the 21T-periodic primitive of the function of mean value zero. Similarly, the second equation can be solved for R(2)(p) as the mean value ofthe right hand side and r(2)(0, p) defined in the same way as r(1)(O, p). This process actually converges for e sufficiently small, but this line of investigation will not be pursued here. Let us recapitulate what happens in the first approximation. Suppose R(l)(p), r(l)(O, p) are defined by (2.13) and consider the exact transformation of variables r
(2.14)
=
p
+ er(l)(O, p)
applied to (2.7). For e small, the equation for p is (1
+ e Or(l») op p = e[g(p + er(l), 0, e) -
g(p, 0,0)
+ R(l)(p)],
190
ORDINARY DIFFERENTIAL EQUATIONS
and thus, (2.15) where R(2)(p, 8, e) is continuous at e = O. Consequently, the transformation (2.14) reduces (2.7) to a higher order perturbation of the autonomous equation
p = eR(l) (p),
(2.16)
where R(l)(p) is the mean value of g(p, 8,0). Suppose ]l(1)(pO) = 0, dR(l)(pO)/dp < 0 for some pO. Then pO is a stable equilibrium point of (2.16) and the equation (2.i5) satisfies the conditions of TheOrem IV.4.3 with x = p and the equations for yand z absent. Consequently, there is an asymptotically stable 2?r-periodic solution of (2.15) and therefore an asymptotically stable 211'-periodic solution of (2.7) given by (2.15). If this is applied to van der Pol's equation where 9 is given by (2.8), then R(1)(p) = 1 -
f2 g(p, 8,0) d8= P- ( 1 -p2) - . 0T
211'0'
2
4
This function R(l)(p) is such that R(1)(2) = 0, dR(1)(2)/dp = -1 < O. Therefore, the van der Pol equation (2.3) has an asymptotically stable periodic solution which for e = 0 is u = 2 cos t. Zero is also a: solution of R(1)(p) = 0 and dR(l)(O)/dp > O. An application of Theorem IV.4.3 implies that zero corresponds to the unstable equilibrium point u = 0 of (2.3).
I
V.3. Averaging
I
One of the most important methods for the determination of periodic and almost periodic· solutions of nonlinear differential equations which contain a small parameter is the so-called method of averaging. The method was motivated in the last section by trying to determine under what conditions one could perform a time varying change of variables which would have the effect of reducing a nonautonomous differential equation to an autonomous one. This idea will be developed to a further extent in this section, but our main interest will lie in the information that can be obtained from only taking the first approximation. Suppose x in on, e ~ 0 is a real parameter,!: R X On X [0, (0) _On is continuous, !(t, ;p, e) is almost periodic int uniformly with respect~o x in compact sets for each ftxed e, has continuous first partial derivatjies With respect to x and !(t,x,e) -+ !(t,x,O), o!(t,x,e)/ox -+ o!(t,x,O)loxas e -+ 0 uniformly for t in (-":,,,00), x in compact sets. Along with the system of equations,
SIMPLE OSCILLATORY PHENOMENA, METHOD OF AVERAGING
191
x = eJ(t, x, e),
(3.1)
consider the" averaged" system,
x=eJo(x),
(3.2) where
1
(3.3)
Jo(x)
= lim -T P-+
00
f J(t, x, 0) dt. p
0
The basic problem in the method of averaging is to determine in what sense the behavior of the solutions of the autonomous system (3.2) approximates the behavior of the solutions of the more complicated nonautonomous system (3.1). There are two natural types of interpretations to give to the connotation "approximates". One is to ask that the approximation be valid on a large finite interval and the other is to ask that it be valid on an infinite interval. The result!'! given here deal only with the infinite time interval. Every solution of the equation i = 0 is a constant. Therefore, system (3.1) is a perturbation of a system which is critical with respect to the class of periodic functions of any period whatsoever and, in particular, with respect to the class of almost periodic systems. On the other hand, it will be shown below that there is an almost periodic transformation of variables which takes x--+y and system (3.1) into a system of the form
y = e[fo(y) +JI(e, t, y)] whereJI(O, t, y) = O. If there is a yo such that Jo(yo) this system can be written as
z = e [ OJa::o) z + {Jo(yO + z) -
Jo(yo) -
= 0 and y = Yo + z, then
OJ~~o) z} +J 1(e, t, Yo + z) ]
.
Therefore, in a neighborhood of Yo the methods of Chapter IV can be applied if the linear system
.
oJo(Yo)
.z=e-oy z is noncritical with respect to the class of functions being considered. In this way, we obtain sufficient conditions for the existence and stability of periodic and almost periodic solutions of (3.1). The reasoning above is the basic idea in the method of averaging. Throughout the remainder of this chapter, we assume the norm in On is the Euclidean norm. The following lemma is fundamental in the theory . . LEMMA 3.1. Suppose g: R X On --+On is continuous, g(t, x) is almost periodic in t uniformly with respect to x in compact sets, has continuous first
192.
ORDINARY DIFFERENTIAL EQUATIONS
derivatives with respect to x and the mean value of g(t, x) with respect to t is zero; that is,
(3.4)
f
1 T
lim T-+ex>
T
= O.
y(t, x) dt
0
For any e > 0, there is a function u(t, x, e), continuous· in (t, x, e) on R X On X (0, (0), almost periodic in t for x in compact sets and e fixed, having a continuous derivative with respect to t and deriv.atives of an arbitrary specified order with respect to x such that, if k(t, x, e)
=
Ou(t, x,e)
at
-
g(t, x),
then all of the functions k, ok/ox, eu, eOu/ox approach 0 as e~O uniformly with respect to t in R and x in compact sets. This lemma is proved in Lemma 5 of the Appendix. It is advantageous at this point to discuss the meaning of Lemma 3.1 when the function g(t, x) satisfies some additional properties. Suppose g(i, x) is a finite trigonometric polynomial with coefficients which are entire functions of x. Then (3.5)
L
g(t, x) =
ak(x)eiA..t, Ak
'7= 0,
l;>;k;>;N
where the Ak are real and each ak(x) is an entire function of x. If
(3.6)
'"
(t, x ) _u
t...
l;>;k;>;N
ak(x) iA.t . e, ~Ak
then u satisfies all of the conditions stated in the lemma. In addition, u is independent of 13 and (3.7)
ou(t, x) - - -g(t, x)=O.
at
In most applications, the manner just indicated is the method for the construction of a function u with the properties stated in the lemma. For a general function g, one may not be able to solve (3.7) in spite of the fact that the mean value of g is zero. In fact, there are almost periodic functions g(t) (see the Appendix) such that M[g]
=0
and
fg is an unbounded function.
The lemma states that, even in this case, one can "approximately" solve (3.7) by an almost periodic function u(t, e). Of course, u(t, e) will become unbounded as e~O, but eu(t, e)~O as e~O. LEMMA 3.2. Suppose / satisfies the properties stated before equation (3.1),/0 is defined in (3.3) and n is any compact set in On. Then there exist an eo> 0 and a function u(t, x, e) continuous for (t, x, e) in R X On X (0, eo],
193
SIMPLE OSCILLATORY PHENOMENA, METHOD OF AVERAGING
almost periodic in t for x in compact sets and e fixed and satisfying the con· clusions of Lemma 3.1, with. g(t, x) =f(t, x, 0) - fo (x) such that the trans· formation of variables (3.8)
x. = y + eu(t, y, e),
(t, y, e)
E
R X 0 X [0, eo]
applied to (3.1) yields the equation
if = efo(y) + eF(t, y, e),
(3.9)
where F(t, y, e) satisfies the same conditions asf(t, y, e) for (t, y, e) X [0, eo], and, in addition, F(t, y, 0) o.
=
E
R X 0
PROOF. If g(t, x) =f(t, x, 0) - fo(x), then the conditions of Lemma 3.1 are satisfied. Let u(t, x, e) be the function given by that lemma. Since eu(t, y, e), eou(t, y, e)/oy, f(t, y, 0) - f(y) - ou(t, y, e)/ot-+O as e-+O uni· formly with respect to t in Rand y in compact sets, we define eu(t, y, e), eou(t, y, e)/oy, f(t, y, 0) - fo(Y) - ou(t, y, e)/8y at e = 0 to be zero. For any e1 > 0, let 0 1 be a compact set in en containing the set {x: x = y eu(t, y. e), (t,y, e) E R X 0 X [0, e1]}. Choose e2 > 0 so that 1 e ou(t, y, e)/oy has a f. bounded inverse for (t, y, e) E R X 0 X [0, e2]. This implies (3.8) can have a~ I' most one solution YEO for each (t, x, e)e" R X.o 1 X [0, £2]. For ~ny Xo E Q~(' the contraction mapping principle implies there is ane3(xo) > 0 such that (3.8) has a unique solution y = yet, x,e) defined and continuous for Iy - xol ~ e3(xo), Ix - xol ~ ea(xo), 0 < e < ea(xo). Since 0 1 is compact, we can choose e4 > 0 independent of :Vo such that the same property holds with ea(xo) replaced by e4. If eo = min(e1, e2, (4), then (3.8) does define a homeomorphism. Therefore, the transformation (3.8) is well defined for 0 ~ e ~ eo, YEO, t E R. If x = y eu(t, y, e), then
+
+
an
+
(1 + :;)iJ = e
efo(y)
+ e[f(t, y, 0) -
+ e[f(t, y + eu, e) -
fo(Y) -
~]
f(t, y, 0)].
From the properties of u in Lemma 3.1 and the continuity off, the indicated property of tb,e equation for y follows immediately. Lemma 3.2 is most important from the point of view of the differential equations since it shows that the transformation of variables (3.8), which is almost the identity transformation for e small, when applied to (3.1) yields a differential equation which is the same as the averaged system (3.2) up through terms of order e. We emphasize again that for functions f(t, x, e) which are trigonometric polynomials with coefficients entire functions of x, the expression for the transformation (3.8) is obtained by taking u to be given by (3.6) where g(t, x) =f(t, x, 0) - fo(x). Let Re A(A) designate the real parts of the eigenvalues of a matrix A.
194
ORDINARY DIFFERENTIAL EQUATIONS
THEOREM 3.1. Suppose j satisfies the conditions enumerated before (3.1). If there is an xo such that jo(xO) = 0, and Re '\'[ojo(xO)/ox] *0, then there are an eo > 0 and a function x*(·, e): R ~en, satisfying (3.1), x*(t, e) is continuous for (t, e) in R X [0, eo], almost periodic in t for each fixed e, m[x*(', e)] c m[f(·, x, e)), x*(·, 0) = x O. The solution x*(·, e) is also unique in a neighborhood of x O. Furthermore, if Re '\'[ojo(XO)/ox] <0, then x*( " e) is uniformly asymptotically stable for < e ~ eo and if one eigenvalue of this matrix has a positive real part, x*( . , e) is unstable.
°
THEOREM 3.2. Suppose j satisfies the conditions enumerated before (3.1). If j (t + T, x, e) = f(t, x, e) for all (t, x, e) in R X en X [0, ex) and there is an xO such that jo(xO) = 0, det[ojo(xO)/ox] 0, then there are an eo> and a function x*(t, e), continuous for (t, e) in R X [0, eo], x*(·, 0) = x O, x*(t + T, e) = x*(t, e), for all t, e and x*(t, e) satisfies (3.1). This solution x*( • , e) is also unique in a neighborhood of x O.
°
*
The proofs of both of these theorems are immediate consequences of previous results. In fact, choose any compact set Q in en containing xO in its interior. Lemma 3.2 implies that we may assume equation (3.1) is in the form (3.9) for (t, x, e) E R X Q X [0, ~oJ. If xO is such that jo(xO) = 0, then y = xO z in (3.9) implies
+
z= deC
=
eAz
+ eF(t, xO + z, e) + e[fo(xO+ z) -
jo(xO) - Az]
eAz+eq(t, z, e),
where A = ojo(xO)/ox. Oh~ can now apply Lemma IVA.3 and Theorem IVA.2 directly to the equation for z to complete the proofs of Theorems 3.1 and 3.2. ln the applications, many problems cannot be phrased in such a way that the equations of motion are equivalent to a system of the form (3.1). However, the underlying ideas in the proof of Theorems 3.1 and 3.2 can be used effectively in more 'complicated situations. One should therefore keep in mind these basic principles and look upon the discussion here as a possible method of attack for the treatment of oscillatory phenomena in weakly nonlinear systems. Another application is now given to a class of equations which seems to occur frequently in the applications. Consider the system (3.10) where e >
°
is I!-
x = ej(t, x) + eh(et, x), real. parameter, J (t, x), h(t, x)
are continuous an<~ have
continuous first derivatives with respect to x, h(t, x) has continuouisecond partial derivatives with respect to x for (t,x) in R X en, j(t,x);ii~ almost periodic in t uniformly with· respect to x in compact sets and there is aT> 0 such that h(t + T,x) = h(t, x) for all t,x. System (3.10) contains both a "fast" time t and a "slow" timeet. The
SIMPLE OSCILLATORY PHENOMENA, METHOD OF AVERAGING
195
averaging procedure is applied to only the fast time to obtain the nonautonomous" averaged" equations
x=
(3.11 )
efo(x)
+ eh(et, x) ~ eG(et, x),
where
f T 1
(3.12)
fo(x) = lim T->O()
T f(t, x) dt.
0
If system (3.11) has a periodic solution xO(et) of period Tie, then the linear variational equation for xO(et) is given by oG(7", xO(7"))
----y
(3.13)
ox
where 7" = et. The next result states conditions under which the equation (3.10) has an almost periodic solution which approaches the periodic solution of the averaged equation (3.11) as e--+O. THEOREM 3.3. Suppose f, h satisfy the conditions enumerated after (3.10). If xO(et) is a periodic solution of (3.11) of period Tie such that no characteristic exponents of the linear variational equation (3.13) have zero real parts, then there are an eo> 0 and a function x*(·, e): R --+On, satisfying (3~1O), x*(t, e) is continuous for (t, e) in R X [0, eo], almost periodic in t for each fixed e, m[x*(·, e)] c m[f(·, x), h(e·, x)]; and x*(t, e) - xO(et)--+O as e --+0 uniformly on R. This solution is also unique in a neighborhood of x O(.). Furthermore, if all characteristic exponents of (3.13) have negative real parts, x*(·, e) is uniformly asymptotically stable and if one exponent has a positive real part, it is unstable. The proof of this result proceeds as follows. Choose any compact set n in On which contains xO(t), 0 ~ t ~ T, in its interior. If g(t, x) =f(t, x) - fo(x) and u(t, x, e) is the function given by Lemma 3.1, then as in the proof of Lemma 3.2, tbere is an eo > 0 such that the transformation of variables x= y eu(t, y, e) is well defined for (t, y, e) E R X n X [0, eo]. This transformation applied to system (3.10) yields the equivalent equation
+
if =
eG(et, y)
+ eH(t, et, y, e),
where H(t, 7", y, 0) = 0 and H(t, 7", y, e) satisfies the same smoothness properties in y as f, h, is almost periodic in t and periodic in 7" of period T. If y(t) = xO(et) z(t), then
+
z=
eA(et)z
+ eZ(t, et, z, e),
196
ORDINARY DIFFERENTIAL EQUATIONS
where A(T) = fJG(T, XO(T))/fJX and Z(t, 8t, Z, 8) satisfies the conditions of Theorem IV.4.2. Suppose P(T)eB'r is a fundamental matrix solution of (3.13) with P(T + T) = P(T) and B a constant matrix. From the hypothesis of the theorem, the eigenvalues of B have nonzero real parts. If z = P(8t)W, then W = 8Bw
+ 8W(t, 8t, w, 8)
where W(t, 8t, w, 8) satisfies the conditions of Theorem IV.4.2. The proof of Theorem 3.3 is completed by a direct application of-this result. Another type of equation that is very common is the system
x= if =
(3.14)
8X(t, x, y, 8), Ay
+ 8Y(t, x, y, 8),
where 8 is a real parameter, x, X are n-vectors, y, Yare m-vectors, A is a constant n X n matrix whose eigenvalues have nonzero real parts, X, Yare continuous on R X en X em X [0, (0), have continuous first derivatives with respect to x, y and are ahnost periodic in t for x, y in compact sets and each fixed 8. The" averaged" equation for (3.14) is defined to be
x=
(3.15)
8X o(X),
where (3.16)
Xo(x)
=
1 lim T-+oo T
f
T
X(t, x, 0, 0) dt.
0
THEOREM 3.4. Suppose X, Y satisfy the conditions enumerated after (3.14). If there is an x Osuch that Xo(x O) =0 and Re A(fJXo(XO)/fJx) #0, then there are an 80 > and vector functions x*(·, 8), y*( . ,8) of dimensions n, m respectively, satisfying (3.14), continuous on R X [0, 80], almost periodic in t for each fixed 8, x*( . , 0) = x O, y*( . , 0) = 0. This solution is also unique in a neighborhood of (xO, 0). Furthermore, if Re A(fJXo(XO)/fJx) < 0, Re A(A) < 0, then this solution is uniformly asymptotically stable for < 8 ~ 80 and if one eigenvalue of either of these matrices has a positive real part, it is unstable.
°
°
The proof follows the same lines as before. Suppose Q is a compact set of X(t, x, 0, 0) - Xo(x) and u(t, x, 8) ia the function given in Lemma 3.1, then as in the proof of Lemma 3.2, there is an 80 > such that the transformation
en containing xOin its interior. If g(t, x) =
°
x-;.-x
+ 8U(t, x, 8), y-;.-y,
is well defined for (t, x, y, 8) E R X Q X em X [0, 80]. This transformation applied to (3.14) yields the equivalent system l·
x=
8X o(x)
..i =
Ay
+ 8Xl(t, x, y, 8) + 8X 2(t, x, y, 8),
+ 8Y1(t, x, y, 8),
SIMPLE OSCILLATORY PHENOMENA, METHOD OF AVERAGING
197
where X1(t, x, y, 0) = 0 and X 2(t, x, 0, 0) = 0 and all functions have the same smoothness and almost periodicity properties as before. If the further trans- ' formation x~x, y~J-;Y is made, then the equations become
x = eXo(x) + eX*(t, x, y, e), if =
Ay
+ Y*(t, x, y, e),
where X*(t, x, y, 0) = 0, Y*(t, x, y, 0) = O. If x~xo + x, y~y, then Theorem IV.4.3 with the variable z absent applies directly to the resulting system to yield Theorem 3.4. Using the same proof, one also obtains THEOREM 3.5. If X, Y satisfy the conditions enumerated after (3.14), are periodic in t or period T and there is an X O such that Xo(x O) = 0 and det[8X o(x O)/8x] 0, then there are an eo> 0 and vector functions x*(·, e), y*( ., e) of dimensions n, m, respectively, satisfying (3.14), continuous on R X [0, eo], periodic in t of period T and x*( . , 0) = x O, y*( . , 0) = o. Furthermore this solution is unique in a neighborhood of (x O, 0) . .An. interesting application of Theorem 3.4 to stability of linear systems is
'*
THEOREM 3.6. Suppose D = diag(B, A) where B is n X n, A is m X m, all eigenvalues of B have simple elementary divisors and zero real parts and all eigenvalues of A have negative real parts. If the (n + m) X (n + m) almost periodic matrix is partitioned as = (jk),j, k = 1, 2, where <1>11 is n X n, <1>22 is m X m and if all eigenvalues of the matrix (3.17)
E=lim T->
00
1 -f. T
T
e-Btn(t)eBtdt,
0
have negative real parts, then there is an eo> 0 such that the system
u = Du + e(t)u,
(3.18)
is uniformly asymptotically stable for 0 < e ~ eo. If one eigenvalue of E has a positive real part, then system (3.18) is unstable for 0 < e ~ eo. The proof of this result proceeds as follows. Let u = (x, y) where x is an n-vector. The'matrix eBt is almost periodic int and therefore the bounded linear transformation x ~ eBtx, y ~ y yields the equivalent system
x = ee-Btn(t)eBtx + ee-Bt12(t)y, 'if =
Ay
+ e21(t)e Btx + e22(t)y.
This is a special case of system (3.14) and the averaged system (3.15) for this situation is x = Ex. From the hypothesis on E, this averaged system satisfies the hypotheses of Theorem 3.4. It is clear from uniqueness of the solution
198 ~*(
ORDINARY DIFFERENTIAL EQUATIONS
. , e), y*( • , e) guaranteed by Theorem 3.4 that
o ~ e ~ eo. This proves the result.
~*(
. , e) = 0, y*( • , e) = 0,
The remainder of this chapter is devoted to applications of the results in .this section.
V.4. The Forced Van Der Pol Equation Consider the equation
(4.1)
il = Z2, i2
where e > 0, Wl
. -Zl
+ e(I - Z~)Z2 + A sin wlt
+ B sin W2 t,
> 0, W2 > 0, A, JJ are real constants and
(4.2) for all integers m, ml, m2 with Iml +Imll + Im21 ~ 4. For e = 0, the general solution of (4.1) is
(4.3)
+ ~2 sin t + Al sin Wlt + Bl sin W2 t, Z2 = -~l sin t + ~2 cos t + Alwl cos Wlt + B l w2 cos W2 t,
Zl = ~l cos t
whereA l "':" A(I - w~)-l, Bl = B(I - w~)-land~l' ~2 are arbitrary constants. To discuss the existence of almost periodic solutions of system (4.1) by using the results of the previous section, consider rela.tions (4.3) as a transformation to new coordinates ~l, ~2. After a few straightforward calculations, the new equations for ~l, ~2" are
(4.4)
XI = i2
- e(I - Z~)Z2 sint,
= e(I -
Z~)Z2 cos t,
where Zt, Z2 are the complicated functions given in (4.3). System (4.4) is a special case of (3.1) and the quasi-periodic coefficients in the right hand sides of (4.4) have basic frequencies 1, Wl, W2. The average of the right hand side of (4.4) with respect to t will have different types of terms depending upon whether the frequencies 1, Wl, W2 satisfy (4.2) or do not satisfy (4.2). If (4.2) is satisfied, then the averaged equations of (4.4) are (4.5)
8il = ul[2(2 - A~ - Bn -
(4 + ·~m,
Si2 = e~2[2(2 - A~ - Bi) - (~i
+ ~m·
Equations (4.5) always have the constant solution ~l = ~2 = 0 and both eigenvalues of the linear variational equation are 2(2 - Ai - Bi). If
199
SIMPLE OSCILLATORY PHENOMENA, METHOD OF AVERAGING
A~ + B~ =F 2, then Theorem 3.1 implies the existence of an almost periodic solution of (4.4) with frequencies contained in the module of 1,Wl,w2, which is zero for e = 0, is uniformly asymptotically stable if A + Bi > 2 and unstable if A + B~ < 2. This implies the original equation (4.1) has an almost periodic solution which is uniformly asymptotically stable (unstable) for A + Bi > (<)2 md this solution fore = 0 is given by
f
i
i
ZI(t) = Al sin wIt
+ BI sin Wz t,
Zz(t) = ZI(t).
+
Notice that the condition A~ A~ > 2 can be achieved if either A or B is sufficiently large for given WI, wz , or if either WI or Wz is sufficiently close to 1 (resonance) for a given A, B. Also, notice that A~ A~ < 2 implies that the averaged equations have a circle of equilibrium points given by x~ ~ = 2(2 - A~ - Bi). There are very interesting oscillatory phenomena associated with this set of equilibrium points, but the discussion is much more complicated and is treated in the chapter on integral manifolds.
+
+
V.5. Duffing's Equation with Small Damping and Small Harmonic Forcing Consider the Duffing equation (5.1)
it
+ e8u + u + eyu3 =
eB cos wt,
or the equivalent system (5.2)
u=v { v= -u -eyu3 -e8v + eB cos wt .
°
+
. where e ~ 0, y, 8 ~ 0, B, w ~ are real parameters. For w 2 = 1 e{3, we wish to determine conditions on the parameters which will ensure that equation (5.1) has a periodic solution of period 27T/W. Since for e = 0, W = 1, the forcing fUI}.ction has a frequency very close to the free frequency of the equation. The free frequency is the frequency of the periodic solutions of the equation (5.1) when e = 0. Such a situation is referred to as harmonic forcing . We have seen previously that the linear equation it + u = cos t has no periodic solutions and in fact all solutions are unbounded. This is due to the resonance effect of the forcing function. As we will see, the nonlinear equation has some fascinating properties and, in particular, more than one isolated periodic solution may exist. Contrast this statement to the results of Section IV.5. To apply the results of Section 3, we make the van der Pol
200
ORDINARY DIFFERENTIAL EQUATIONS
transformation u = Xl sin wt + X2 coswt .. { v = W[Xl cos wt - X2 sin wt]
(5.3)
in (5.2) to obtain an equivalent system Xl
= - [fJu - yu3 - 8'IJ + B cos wt] cos wt
X2
= - ~[fJu -
.
(5.4)
8
w
w
yu3 - 8v + Bcos wt] sin wt
where u, v are given in (5.3). To average the right hand sides of these equations with respect to t, treating Xl, X2 as constant, it is convenient to let (5.5)
X2 = r cos 0/, u
=
Xl
Xl
sin wt
= r sin 0/,
+ X2 cos wt = r cos (wt -
0/),
v = W[Xl cos wt - X2 sin wt] = -wr sin(wt - 0/). This avoids cubing u and complicated trigonometric formulas. The averaged equations associated with (5.4) are now easily seen to be (5.6)
8[
3yr2
]
X2= -2w fJxl-Txl+8wX2 , r2 =x~+x~.
Since X2 = r cos 0/, Xl = r sin 0/, the" equilibrium points of (5.6) are the solu"tions of the equations
( fJ -3yr2) 4 rcos 0/ - 8wrsin 0/ + B = 0, 3yr2) rsin 0/+ 8wrcos 0/=0. ( fJ- 4 These latter equations are equivalent to the equations (5.7)
def 3yo r2 Fo G(r, 0/, w) = w 2 -1 - -4- + -;: cos 0/ = 0,
F(r, 0/, w)~ Fo sin 0/ - 80 wr:== 0,
~
.,""
SIMPLE OSCILLATORY PHENOMENA, METHOD OF AVERAGING
201
where we have put yo = ey, So = e8, Fo = eB, parameters which have a good physical interpretation in equation (5.1). If yo, 80 , Fo are considered as fixed parameters in (5.7); then (5.7) can be considered as two equations for the three unknowns "', w, r: If there exist Wo, ro such that the matrix
"'0,
(5.8)
o(rG) [
or
of or
"'0,
has rank 2 for r = ro, '" = W = Wo, then Theorem 3.2 implies there is an eo> 0 such that equation (5.1) has a periodic solution of period 27T/WO which for e = 0 is given by u = ro cos(wo t - "'0) = ro cos(t - "'0) since wo = I for e = O. In the equations (5.7), one usually considers the approximate amplitude r ofthe solution of (5.1) as a parameter and determines the frequency w(r) and approximate phase "'(r) of the solution of (5.1) as functions of r. The plot in the w, r plane of the curve w(r) is called the frequency response curve. The stability properties of the above periodic solution can sometimes be discussed by making use of Theorem 3.1. Some special cases are now treated in detail.
Case 1. 8 = 0 (No Damping). given by '" = 0 and (5.9)
One solution of (5.7) for 8 = 80 = 0 is 3yor2
Fo
4
r
w2 =1+--+-.
As mentioned earlier, relation (5.9) is called the frequency response curve. The rank ofthe matrix in (5.8) is two if yo = 0, Fo -=1= Ooryo -=1=0, Fo/yo sufficiently small. From Theorem 3.2, there is an eO> 0 such that for 8 = 0,0 ~ e ~ eO, and each value of w, r which lies on the frequency response curve, there is a 27T/Wperiodic solution of (5.1) which for e = 0 is u = r cos wt = r cos t since w = I for e = O. There is also the solution of (5.7) corresponding to '" = 7T, but this corresponds to (5.9) with r replaced by -r. The uniqueness property guaranteed by Theoniin 3.2 and the equation (5.2) fo.r 8 = 0 imply this solution is the negative of the one for'" = O. If Fo ~ 0, the plots of the frequency response curves in a neighborhood of w = I for both the hard spring (yo> 0) and the soft spring (yo < 0) are given in Fig. 5.1. The pictures are indicative of what happens near w = I. The frequency response curves in Fig. 5.1 are usually plotted with Irl rather than r and are shown in Fig. 5.2. Notice how the nonlinearity (yo -=1= 0) bends the response curve for the linear equation. The curve Fo = 0 depicts the
202
ORDINARY DIFFERENTIAL EQUATIONS
----------t---~~--~~====~--~w
('Yo> O)(hard spring) (a)
r
-----------+--------~------------~w
('Yo
< O)(soft spring) (b)
Figure V.5.1
relationship between the frequency wand the amplitude r of the periodic solutions of the "unforced conservative system ii u + you 3 = o. for each given wnear w= 1, there is exactly one periodic solution of period 2
+
SIMPLE OSCILLATORY PHENOMENA, METHOD OF AVERAGING
Fo = 0
1,.1
1,.1
--~----~l----------==--~W
203
Fa = 0
----~--------w-=~l----~~----~W
(1'0> O)(hard spring)
(1'0
< O)(soft spring)
(a)
(1))
Fo = 0
Irl
----~--~--------~------~~----~W
(1'0
= O)(linear equation) (c)
Figure V.5.2
Let us emphasize another striking property of this example which is a direct consequence of the fact that the equation is nonlinear. For 8 = 0, system (5.1) has a free frequency which is equal to 1; that is, all solutions of the linear equation are periodic of period 217. On the other hand, it was shown above that for w 2 -1 of Qrder 8 there are periodic solutions of period 217/W .. In other words, the free frequency is suppressed and has been "locked" with the forcing frequency. This phenomena is sometimes referred to as the locking-in phenomena or entrainment of frequency and, of course, can only occur when the equations are nonlinear.
Case 2. 8> 0 (Damping). If 80 > 0 and Fo = 0, then using the same analysis as in .~he discussion of equation (2.1), one sees that equation (5.1) has no periodic solution except u = O. This is reflected also in equations (5.7) which in this case, have no solution except r = O. If 80 Fo 0, there always exist values of r such that Irl < IFo/8 0 wi and the second of equations (5.7) can be solved for tP as a function of w8 0 r/Fo. Using such a l/J, the first equation yields, up to terms of order 8 2 ,
*
(5.10)
3 r2 J F 2 w2=1+~± ~_82. 4 r2 0
204
ORDINARY DIFFERENTIAL EQUATIONS
This frequency response curve for the hard spring (yo> 0) is shown in Fig. Q.3. The dotted line corresponds to the curve w 2 = 1 3yor2/4. Thus, as for S = 0, for given F o , So with Fo So =1= 0, Theorem 3.2 implies there are three periodic solutions of period 27T/W for some values of wand only one for others. For a given w, which of these solutions are stable and which are unstable? To discuss this, we investigate the linear variational equation associated with these equilibrium points of the averaged equations and apply Theorem 3.l. As is to be expected the analysis will be complicated. Another type of coordinate system which is widely used in applying the method of averaging turns out to be useful for this discussion.
+
r
w=l ('Yo> O)(hard spring)
Figure V.6.3
In (5.2), introduce new variables r, (5.11)
u
t/i by the relations
= r cos(wt - ifs),
v = -rw sin(wt - ifs), to obtain an equivalent set of equations (e{3 (5.12)
r = ~ [...:;. e: sin 2(wt -
= w2 -
1)
ifs) - ef sin(wt - ifs)] ,
~ = w~ [e{32 + e{32 cos 2(wt -
ifs) + ~ f cos(wt - ifs)] , r
+
where f = _yu3 - Sv B cos wt and u, v are given in (5.11). The averaged equations associated with (5.12) are (5.13)
.
1
r = - 2w - [eSwr -
. 1·[ 3
eB sin ifs]
I/J = 2w e{3 - 4eyr2
1
=-
2w
F(r, ifs, w),
1
+ reB cos ifs] = 2w G(r, ifs, w),
SIMPLE OSCILLATORY PHENOMENA, METHOD OF AVERAGING
205
where F, G are defined in (5.7). ThE! equilibrium points of (5.13) are therefore the solutions r, 0/, w of (5.7). If the eigenvalues of the coefficient matrix of the linear variational equation of any such equilibrium point have nonzero real parts, then Theorem 3.1 gives not only the existence but the stability properties of a periodic solution of period 27T/W of (5.1). The solution is uniformly asymptotically stable ifthese eigenvalues have negative real parts and unstable if one has a positive real part. A necessary and sufficient condition for the eigenvalues of this matrix to have negative real parts is that the trace ofthis matrix be negative and the d.eterminant be positive. In terms of F, Gin (5.13), one has an asymptotically stable solution if and only if
Fr+G", 0,
where the subscripts denote partial derivatives and, of course, the functions are evaluated at the equilibrium point. These partial derivatives are easily seen to be
Fr = -Dow, F", = Fo cos 0/,
3 Fo Fo . Gr = - - yor - - cos 0/, G",= - - sm 0/. 2 r2 r
+
Therefore, Fr G", = -2Dow < Oat an equilibrium point and the stability or instability can be decided by investigating the sign of F r G", - Gr F", as long as it is different from zero. It is possible to express this last condition for stability in terms of where the point r, w is situated on the frequency response curve. To see this, consider the equilibrium points of (5.13); that is, the solutions of F(r, 0/, w) = 0 = G(r, 0/, w), as functions of w. Differentiating these equations with respect to CL and letting subscripts denote differentiation, one obtains
-Fco = Frrco + F",o/oo , -Goo = Grroo + G",o/oo , and thus, (5.14) The stability properties of the solution can therefore be translated into the sign of roo(Goo F", - F 00 G",). This expression being positive corresponds to stability and this being negative corresponds to instability. Since F 00 = -Dor, Goo = 2w, it follows that .
GooF", - FooG", = 2wFo cos 0/ + DoFo sin 0/ =
2wr(v 2 - w 2 ) + D~ wr.
206
ORDINARY DIFFERENTIAL EQUATIONS
where v 2 = I + 3yor2/4. From (5.10), w 2 - v 2 = O(s) and since 8~ is 0(S2), the sign of this expression is determined by w 2 - v2 if s is sufficiently small. Using (5.14), we finally have:
A periodic 8olution of Du.tfing'8 equation (5.1) with 8> 0 iB 8table if (dr/ilw) • (v 2 - ( 2) > 0 aruJ, 'U/TUltable if (dr/dw)(v 2 - ( 2) <0, where v2 = 1 + 3'Yor2/4 and 1" iB given a8 a fUnction of w by the frequency re8p0'n8ecurve (5.10). Pictoria,p,y, the sit~ is s~own in Fig. 5.4 'for a hard spring. S~lid black dots represent ~~b~ pomts. The arrows _ represent a typICal manner in which the aril.piitUde of the stable periodic motion would change with increasing wand the backward _ depicts the situation for decreasing w. r
-
~------w-=~l--------------------·W
Figure V.5.4
V.6. The Subharmonic of Order 3 for Duffiing's Equation Consider the equation (6.1)
ii
1
+ scu + 9U+ eyu3 = B cos wt
or the equivalent system (6.2)
U=v,
V=
1
-gU -scv -syu3+ B cos wt,
"
where s ~ 0, c ~ 0, 'Y, B and.w are constants, w -1 = O(s) a's·'s_O. The problem is to determine under what conditions there exists a solution of (6.1) which is periodic of period 617/W and whose zeroth order terms in s are given by
SIMPLE OSCILLATORY PHENOMENA, METHOD OF AVERAGING
207
·u = rsin(i- t+ cp) + Acoswt,
(6.3)
v=
~ r cos (
it + cp) '- wA
sin
wt.
Such a solutio.n if it exists is called a subharmonic· solution of order 3. If any such solution exists, then clearly A must equal -Bj(w 2 -lj9). The free frequency of equation (6.1) is (lj3) and the forcing frequency is w which is approximately 1. Therefore, if a subharmonic solution exists, then the free frequency is again suppressed and is locked with the forcing frequency w in the sense that a solution appears which is periodic of period three times the period of the forcing frequency. The form of (6.1) is important for the existence of a subharmonic solution. In fact, if the damping coefficient in (6.1) is a constant 81 > 0 for all 8 ~ 0, then it was shown in Section IV.5 that no such subharmonic solution can exist. If r, in (6.3) are chosen as new coordinates, the new equations are
cp
(6.4)
r = ~ [~r sin 2(~ t+ cp) -
3(cv + yu3) cos(i
yu3)
t+ cp)
Sin(it+cp)l
=8f3,
where u, v are given in (6.3). The averaged equations are (except for some terms of order (6.5)
l
r= 8r
2w
•
8
ljJ = 6w
8 2)
[-c+ 27yrA COS3CP], 2 [ 2 7 yA
_.f3 + -2- (A - r sin
27yr2] 3cp) + -4-
Using the fact that w 2 -1 = 8(3 and letting '}'O for the equilibrium points of (6.5) are
= 8,}"
Co
= 8C, the equations
(6.6)
where up to terms of order 8 2 , the latter expressions are evaluated with A = -9Bj8. For c = 0 (no damping), the formula for the frequency response
208
ORDINARY DIFFERENTIAL EQUATIONS
curve (6.6b) simplifies to w2= 1
(6.7)
+ 27yoA2 + -27yo (r ± A)2. 4 4
A sketch of the frequency response curves is given in Fig. 6.1. r
L -_ _ _ _..>L...,,-_ _ _ _ _ _ _ _ _ _ _
w2 - 1
27<poA 2 /2 ('Yo> O)(hard spring)
Figure V.6.1
One now uses Theorem 3.2 and Theorem 3.1 exactly as in the previous section to obtain the existence of an exact solution of (6.1) and the stability properties of such a solution for values of w, r, cp satisfying (6.6). The frequency response curves suggest the fact confirmed in Section IV.5 that the equation (6.1) may not have a subharmonic of order 3 if c gets too large, since the above analysis has been shown to be valid only for e small and, thus, w close to 1. Also, notice that once a subharmonic solution is known to exist, two other distinct ones are obtained simply by translating time by 27T/W.
V.7. Damped Excited Pendulum with Oscillating Support A linear damped sinusoidally excited pendulum with a rapidly vertically oscillating support of small amplitude can be represented approximately by the equation (7.1)
u + cu + (1 + e d2~~:t)) sin u -
F cos.wt = 0,
where u is the angular coordinate measured from the bottorq position, h(T) = h(T + 27T), V = e-l, 0 <e ~ 1 and c, F, ware real positive,~~rameters independent of e. To transform (7.1) into a form (3.10) for which Theorem 3.3 is applicable,
SlM?LE OSCILLATORY PHENOMENA, METHOD OF AVERAGING
209
it is convenient to treat all but. the term in c as derivable from the Hamiltonian
H(v u t) = ~ [V2 - 2ev dh sin u _ "2 dt
(7.2)
(dh)2 cos 2 u] dt
132
+ (I-cosu) -uF cos wt, where v is the momentum conjugate to u and the argument of h is vt. The equations of motion are, therefore,
dh 13 2 (dh)2 V=- [ -ev-cosu+sin2u+sinu-Fcoswt ] dt
-c(v -
13
dh sin dt
2
dt
u) '
where the argument of h is vt. If vt = T and differentiation with respect to equations (7.3) become (7.4)
T
is denoted by prime, then
u' = e[v - h' sin u],
v'
= 13 [Vh'
cos u -
~ (h')2 sin 2u -
sin u - cv + ch' sin u]
+ eF cos eWT. where the argument of his T. Equations (7.4) are a special case of the system (3.10), the expressions in the square brackets being periodic in fast time T and the other expression is periodic in the slow time eT. Considering the special case where h(T) = A sin T and A is a constant, the averaged equations (3.H) corresponding to equations (7.4) are (7.5)
u' = eV, v' =
- 13
A2 sin u cos u + sin u + cv ] ["2
+ eF cos eWT.
In terms of the original time t, these equations are equivalent to the equation
(7.6)
u=v,
v = ~ (I + ~2 cos u) sin u'- cv + F cos wt, or the single second order equation (7.7)
ii
+ cu + (I + ~2 cos u) sin u=
F cos wt.
210
ORDINARY DIFFERENTIAL EQUATIONS
If equation (7.7) has a periodic solution of period 27T/W such that the characteristic exponents of its linear variational equation have negative real parts, then Theorem 3.3 implies the original equation (7.1) has an asymptotically stable, almost periodic solution which approaches this solution as e -+"0. To find conditions under which (7.7) has such a periodic solution, one can use the results of Chapter IV or the method of averaging (Theorems 3.1 and 3.2) directly on (7.7). For example, if F is small, one can use the results of Chapter IV very efficiently as follows: For F = 0, the equilibrium solutions of (7.7) are u=O, U=7T and u=cos-i(2/A2) if A2>2. If U=7T+W, then (7.7) becomes
ib + ctb + (~2 cos W - 1) sin W = F cos wt,
(7.8)
and this may be rewritten as
ib+Ctb+(~2 -1)W=Fcoswt+(~2 -1)(W-SinW) A2
+ "2 (1 -
cos w) sin w.
If A 2/2 > 1 and F is small, this equation satisfies the conditions of Theorem IV.2.1 and Theorem IV.3.1. One can therefore assert the existence of an asymptotically stable periodic solution of (7.8) of period 27T/W. This implies in turn that the pendulum described by (7.1) can execute stable motions in a neighborhood of the upright position.
V.S. Exercises EXERCISE 8.1. An infinite conductor through which there tiows a current of magnitude I attracts a conductor A B of length l and mass m along which there flows a current i. In addition, the conductor AB is attracted by a spring C whose force of attraction is proportional to its deflection with proportionality constant k. With x = 0 chosen as the position of AB when the spring is undeflected, the equations of motion of AB are
A_),
x=Y, iJ=.:...~(x __
(8.1)
m
a-x
where A = 2Iil/k. Discuss the behavior of the solutions of this equation along the lines of example 1.3. EXERCISE
8.2.
For c > 0, h > 0, w -1 = O(e) as e -+0, find t,~periodic
solutions of
(S.2)
x + x = e[ -ci + hx3 + X cos 2wt].
Plot the frequency response curves, find the intervals on the w-axis for which
SIMPLE OSCILLATORY PHENOMENA, METHOD OF AVERAGING
211
there are one, two and three periQdic solutions and discuss the stability properties of these solutions. EXERCISE 8.3. The equation of a plane pendulum with vertically oscillating sinusoidal support can be written as
(8.3)
fJ+cB+
If T = wt, III =
13,
(J - Iw 2 sin wt l sinO=O.
e2 k 2 = (J/lw 2 , c/w = 2eoc we have
0" + 2eocO'
+ (e 2k2 -
13
sin
T)
sin 0 = 0,
=d/dT.
Show that the system for 13 small has periodic solutions oLootiQd 21Tfw when 0 is in the neighborhood of either 0, 7r or 27r and cos- 1(-2k2). Determine the stability properties as functions of oc and k. Introduce variables cP and Q by
o=
cP -
0' =
eQ -
13
sin 13
T
cos
sin cP, T
sin cP,
and use averaging. EXERCISE
8.4.
Consider the second order system
x+x=e[f(x) -cx+sinwt],
(8.4)
where c > 0,0 < 13 ~ 1, f(x) is a continuous function of x and w -1 = 0(13). For 13 small, determine w so that the· equation has periodic solutions of period 21T/W. Determine the expression for the approximate frequency response. Study the stability of the solutions. Show that if y = rl(w 2 -1) then the periodic solutions are asymptotically stable if c > 0 and do,
dy [y + G(a) + 0(13)] < 0, where G(a) = (1/1Ta) f:"f(a cos cP) cos cP dcP. Use the transformation x = A(t)·
+
+
+
cos (wt 0(t)), x = -wA(t) sin (wt 0(t)). Note that y G(a) = 0 is the approximate relationship between the amplitude an«i frequency of the periodic motion. of the autonomous problem with c = O.
V.9. Remarks and Suggestions for Further Study The results of Section 1 on conservative systems are extremely simple but at the same time not very informative except for dimension 2. For high dimensional conservative systems, the theory has been and will continue to be a challenge for many years to ~o~e. Recently, Pal~ore [1] and Palamodov [1] have proved H(p,q) = p'p/2 + V(q) real analytic and q not a minimum of V
2i2
ORDINARY DIFFERENTIAL EQUATIONS
implies the equilibrium point (0,0) is unstable. Laloy [1] gave a simple example showing this result is false if one only assumes V is Ceo. For further references on this problem, see Rouche, Habets and Laloy [1]. For other fascinating aspects of conservative systems, see Poincare [2] ,Birkhoff [1], Kolmogorov [1], Arnol'd [2], Moser [13], Siegel!"!.d Moser [1]. The method 01 averaging of Section 3 evolved from the famous paper of Krylov and Bogoliubov [1] and is contained in the book of Bogol~ubov and Mitropolski [1). The basic Lemma 3.1 is due to Bogoliubov [1). Theorem 3.3 as well as the example in Section 7 are due to Sethna [1]. There are many variants of the method of averaging and the reader may consult Mitropolski [1] and Morrison [1] for further discussion as well as additional references. Other illustrations of equations which exhibit interesting oscillatory phenomena may be found in the books of Bogoliubov and Mitropolski [1], Malkin [2], Minorsky rll, Andronov, Vitt and Khaikin [1], Hale [7].
CHAPTER VI Behavior Near a Periodic Orbit
In Section III.6 and Chapter IV, a rather extensive theory of the behavior of solutions of a system (and perturbations of a system) near an equilibrium point has been developed. In this and the following chapter, we consider the same questions except for a more complicated invariant set; namely, a closed curve. Suppose u: R -+ Rn is a periodic function of period w which is continuous together with its first derivative, du(8)/d8 01= 0 for all 8 in (-00, (0) and u: [0, w) -+ Rn is one-to-one. If (1)
r = {x ERn: x = u(8),
o~
8 ~ w},
r is a closed curve and is actually a Jordan curve. Suppose f: Rn -+ Rn is continuous and satisfies enough smoothness properties to ensure that a unique solution of
then
(2)
x =f(x)
passes through any point in Rn. Throughout this chapter, it will be assumed that the curve r is invariant with respect to the solutions of (2); that is, any solution of (2) with initial value on r remains on r for all tin (-00, (0). If r is a periodic orbit of (2), then there is a periodic solution c/>(t) of (2) such that
r = {x: x = c/>(t),
-00
In this case, we may choose the parametric representation of r to be given by the function ?>; that is, we may choose u = c/>. For periodic orbits, such a parametric representation always will be chosen in the sequel. If r is not a periodic orbit, then it must contain equilibrium points and orbits whose ocand w-limit sets are equilibrium points. From the previous remarks, if x*(t), - 00 < t < 00, is any solution of (2) with x*(O) in r, then there are three possibilitieEs: (i) there is aT> 0 such that x*(t) is periodic ofleast period T, (ii) x*(t) is an equilibrium point; (iii) x*(t) has its oc- and w-limit sets as equilibrium points. In the first case, the curve r can
213
214
ORDINARY DIFFERENTIAL EQUATIONS
be represented parametrically by x = x*(O), 0 ~ 0 < 'T and any solution of (2) on r [such a solution must be x*(t «) for some constant 'IX] defines a ftinction 8(t), - 00 < t < 00, such that (J = 1. In either case (ii) or (iii) any solution x*(t) of (1) on r defines a continuously differentiable function O(t) = u-1(x*(t)), - 00 < t < 00, and (J = g(O) where g(O) = [Clu-l(u(O))/ox]f(u(O)) = [du(8)/d8]-lf(u(0)). In particular, in case (ii), there must be a zero of g(O). In all three cases, we can, therefore, assert the existence of a continuous functio~ g(8) such that
+
(3)
du(O)
dO g(O) =f(u(8)),
g(O
+ w) = g(O),
o ~ O~ w.
Furthermore, if r is generated by a periodic solution of (2), then g( 0) can be taken =1 and if r contains an equilibrium point of (2), there must be a zero of g( 0). In this chapter, our main concern is the behavior of the solutions of (2) near a periodic orbit (sufficient conditions for stability and the saddle point property) and the existence of a periodic orbit for the system (4)
i =f(x) + F(x, t),
for the case in which F is a small perturbation independent of t. In the next chapter, the discussion will allow r to be an arbitrary invariant curve and the perturbations F to depend on t. In both cases, the first step will be the introduction of a convenient coordinate system in a neighborhood of r. Such a coordinate system is introduced in this chapter.
VI. I, A Local Coordinate System about an Invariant Closed Curve By a moving orthonormal system along r is meant an orthonormal coordinate system {el(O), ... , en(8)} in Rn for each 8 in [0, wJ which is periodic in 8 of period wand one of the ej( 0) is equal to [du(O)/d8]/ldu( 0)/d81. The first objective is to show there are moving orthonormal systems for r. For this purpose, the following lemma is needed. LEMMA 1.1. If n ~ 3 and v( 0) is a unit vector in Rn which has period w and satisfies a lipschitz condition, then there exists a unit vector, (independent of 0) such that v(O) foraH O.
* ±,
PROOF. This lemma is a consequence of well known facts fr~ real variables. In fact,C the set x = v(O), Iv(O)i = 1, 0 ~ 0 ~ w, is a curv,-~'on the unit sphere Sn-l in Rn. Since v(O) is lipschitzian, this curve is rectifiable and a rectifiable curve on a sphere in Rn, n ~ 3, covers a set of measure zero. There-
BEHAVIOR NEAR A PERIODIC ORBIT
215
fore, there always exists a vector gin Sn-l which is not on this curve or the curve defined by -v(O). Rather than invoke this result, a proof is given here. If S is any set on the unit sphere in Rn of diameter 0, du(())/d() i= 0, 0 ~ () < w, and r is defined in (1), then there is a moving orthonormal system along r which is qjP-l(R, Rn). PROOF. Suppose n ~3. If v(()) = [du(())/d()]/ Idu(())/d()l, then the hypotheses on u imply that v is periodic of period wand lipschitzian. Let el be a constant unit vector (the existence of which is assured by Lemma 1.1) such that el -=i= ± v(()), 0 ~ () ~ w. Adjoin to el any constant vectors e2, ... , en such that {el' ... , en} is an orthonormal basis for Rn. The moving orthonormal system along r is then obtained in the following manner: let S be the (n - 2)dimensional subspace of Rn orthogonal to the plane formed by el and v(()). Rotate the coordinate system about S in the positive sense until el coincides with v(O) (see Fig. l.l). If 6(()), ... , gn(()) are the rotated positions of e2, ... , en, then the moving orthonormal system is given by
(l.l)
{v(()), 6(()), ... , gn(())},
0
~
()
~ w.
If Yj(()), j = I, 2, ... , n, are the direction angles of v(()), ej . v(()) = cos Yj(()), j = 1, 2, ... , n,. then one can show that the vectors gj are given by
j=2,3, ... , n.
(1.2)
The derivation of (1.2) proceeds as follows. Suppose ej
=
ej
+ Aj el + I1-j v,
where ej belongs to S. The final position of gj is then gj =
ej + Aj el + 11-) v,
216
ORDINARY DIFFERENTIAL EQUATIONS
Figure VI.l.l
where A' , p-j are determined from \ ' P-j by a rotation in the ev v plane by an angle Yl with cos Yl = el . v. Therefore
and
Since S is orthogonal to el, v, it follows that el . [ej - Ajel - P-jv] V·
[ej - Ajel - P-jv]
= 0, =
0,
and this implies A' _ J-
-
cos Yl cos Yj sin 2Yl '
Substituting in the expression for gj, one obtains (1.2). For the case n = 2, the moving orthonormal system is easily constructed as . (1.3) where the coordinates of v are Vl, V2 . The explicit formulas (1.1)-(1.3) for the moving orthonormal system
217
BEHAVIOR NEAR A PERIODIC ORBIT
along r clearly imply that the system is rt'p-I(R, Rn) if u is rt'P(R, Rn). This completes the proof of the theorem. Once a moving orthonormal system along a closed curve r is known, it is possible to use this system to obtain a coordinate system for a" tube" around r. In fact, with u(B) as in Theorem 1.1, v(B) = [du(B)/dB]/ Idu(B)/dBI, let (v, 6, ... , gn) be a moving orthonormal system along r. Consider the transformation of variables taking x into (B, p), p = col(p2, ... , pn) given by (1.4)
x
=
u( B)
+ Z( B)p,
o ~ B<w.
The matrix Z is the n X (n -1) dimensional matrix whose columns are the vectors 6, ... , gn . To show that this transformation is well defined in a sufficiently small neighborhood of r, that is, p sufficiently small, let F(x, B, p) ~ u(B)
+ Z(B)p -
x.
The partial derivatives of F with respect to B, pare of du(B) oB =
dZ(B)
dO + dO
p,
of -=Z(B). op
"*
For p = 0, det[oF/oB, of/op] = Idu(B)/dBI . det[v(B), Z(B)] 0 for 0 ~ B ~ w since du(B)/dB"* 0 for all Band v(B), Z(B) is a moving orthonormal system. Therefore, there is a 0 > 0 independent of B such that det[oF/oB, of/op]"* 0 for Ipl ~ 0, 0 ~ B ~ w. Since the closed curve r is compact, a finite number of applications of the implicit function theorem shows that (1.4) is a well defined transformation for 0 ~ Ipl ~ PI, PI> 0, 0 ~ B ~ w. We now make the transformation (1.4) on the differential equation (4) using the fact that u satisfies (3) to obtain new differential equations for B, p. It is assumed that f has continuous first derivatives with respect to x and u satisfies the conditions of Theorem 1.1. If x(t) = u(B(t» + Z(B(t»p(t) satisfies (4), then (1.5)
[
dU( B)
" dZ( B)
]
dO + ----;;;0 p e+ Z(B)p =f(u(B) + Z(B)p) + F(t, u(B) + Z(B)p).
e,
In a pI-neighborhood r p1 of r, the coefficient matrix of p is nonsingular. p asa function of B, p, t. The Therefore, equation (1.5) can be solved for explicit form ofthe equations are obtained as follows. Using (4) and projecting both sides of (1.5) onto v(B), one obtains
e,
(1.6)
e= g(B) + /1(B, p) + h'(B, p)F(t, u(B) + Z(B)p),
218
ORDINARY DIFFERENTIAL EQUATIONS
where h' is the transpose of h, (l. 7)
h(O, p)
h(O, p)
=
-h'(O, p)
=
[Idu(O) dO I+ v'(O) dZ(O) dO p]-1 v(O),
dZ(O)
dO pg(O) + h'(O, p)[f(u(O) + Z(O)p) -
f(u(O))],
By projecting both sides of tl.5) onto Z(O), and using the fact that = Z'(O)g(O) du(O)jdO = 0, where Z' is the transpose of Z, one obtains
Z'(O)f(u(O))
p=
(l.8)
A(O)p
+12(0, p) + Z'(O) [I - d!~O) ph'(O, P)] F(t, u(O) + Z(O)p).
where Z' is the transpose of Z and A(O) = Z'(O)[- dZ(O) g(O) dO
(l.9)
h(O,.p)
=
+ of(u(O)) Z(O)]., ox
, dZ(O) -Z (0) dO ph(O, p)
+ Z'(O) [f(U(O) + Z(O)p) -
f(u(O)) -
l
of~;O)) Z(O)p
From the above expressions for these functions, it is easily seen that h(O, p), 12(0, p) have continuous partial derivatives with respect to p up through order k ~ 1 iff (x) has continuous partial derivatives with respect to x up through order k ~ l. Furthermore,h(O, p) = O(lpi) as p -+0 andh(O, 0)=0, 012(0, O)jop = O. The number of derivatives of these functions with respect to is one less than the minimum of the derivatives off and du(O)jdO. Also, the equations in 0, p contain F in a linear fashion multiplied by matrices which have an arbitrary number of derivatives with respect to p and (p -1) deriva-
o
tives with respect to· 0 if u has p derivatives with respect to O. These results are summarized in TiIEOREM l.2. If u satisfies the conditions of Theorem l.1 with p ~ 2 and f E ~p,:"l(Rn, Rn), then there exist a 8> 0, n-vectors 12(0, p), h(O, p) a. scalar h(O, p), an (n -1) X (n -1) matrix A(O) and an (n -1) X n matrix B(O, p) with all functions being periodic in 0 of period wand having continuous derivatives of order p - 1 with respect to p and p - 2 with respect to ofor 0 ~ Ipl ~ i), -00 < 0 < 00,
(l.lO)
h(O, p)
=
O(lpi)
12(0,0) = 0,
as Ipl-+O,
Oh~~ 0) = 0,
219
BEHAVIOR NEAR A PERIODIC ORBIT
such that the transformation (1.4) applied to equation (4) for Ipl equivalent.system, (1.11)
e= p=
~
8 yields the
+h(8, p) + h'(8, p)F(t, u(8) + Z(8)p), A(8)p +12(8, p) + B(8, p)F(t, u(8) + Z(8)p),
g(8)
where g(8) is given in (3).
VI.2. Stability of a Periodic Orbit
r
In this section, the case is discussed in which the closed curve is generated by a nonconstant w-periodic solution xO of (2) andfhascontimious first derivatives with respect to x. As mentioned before, we can assume that the parametric representation of r is x = xO(8), 0 ~ 8 < w; that is, u can be chosen as xO and u satisfies (3) with g(8) = 1 for 0 ~ 8 ~ w. Since f(x) has continuous first derivatives in x, the function u( 8) has continuous second derivatives in 8. In terms of the local coordinate system (1.4), th'Ol behavior of the solutions of (2) near are given by the solutions of the differential system
r
(2.1)
+h(8, p), A(8)p +12(8, p), Z'(8)[ _ d!~8) + 8f (;;0)) Z(8))}
(j = 1
P= A(8) =
whereh(8, p), 12(8, p)are continuous in 8, p, have continuous first derivatives, with respect to p, and have period w in 8. These functions satisfy (1.10); that is, (2.2)
Ih(8, p)1
= O(lpj),
12(8,0) =0,
as Ipl---+O, 812(8,0)
-"----=0. 8p
Also associated with the periodic solution u(8) of (2) is the linear variational equation (2.3)
dy d8
8f(u(8)) 8x y.
This linear system with periodic coefficients always has a nontrivial w-periodic solution. In fact, du(8)/d8 is a nontrivial w-periodic solution of this equation since (2) implies d 2u/d8 2 = [8f(u(8))/8x] du/d8.· Therefore, at least one characteristic multiplier of (2.3) is equal to unity.
220
ORDINARY DIFFERENTIAL EQUATIONS
LEMMA 2.1. If the characteristic multipliers of the n-dimensional system (2.3) are 1-'-1, ... ,l-'-n-1, 1, then the characteristic multipliers of the (n -I)-dimensional system (2.4) .
dp dO = A(O)p,
where A(O) is given in (2.1), are 1-'-1, •.. , I1-n-1. PROOF. Suppose Z is the n X (n - 1) dimensional matrix defined in (1.4). The vector du(O)fdO and the columns ofZ(O) are orthogonal since v(8), Z(8) is a moving orthonormal system. Thus, for any n-vector y, there is uniquely defined a scalar f3 and (n -I)-vector p such that y = f3 du(8)fdO Z(O)p. If y is a solution of (2.3), then using the fact that d 2ufd0 2 = [8f(u{8»f8x] du/dO and the columns of the matrix (v, Z) are orthogonal, one immediately deduces that p satisfies (2.4). This shows that the normal component of the solution of the linear variational equation coincides with the solution of the linear variational equation of the normal variation.· Now suppose w 2, ••• , w n are (n -1) solutions of (2.3), Y 1 is the n X n-I matrix defined by Y1 = [w 2 , •.. , w n ] and Y(O) = [du(8)/dO, Y 1(O)] is a fundamental matrix solution of (2.3). If K is defined by Y(w) = Y(O)K, then the eigenvalues of K are the characteristic multipliers of (2.3). From the definition of Y, it follows immediately that K must be of the form
+
K=
[~ ~:l
and, therefore, the multipliers 11-1, ... , I1-n-1 of (2.3) defined in the statement of the lemma are the eigenvalues of K~. If Y1 = [du/d8]0( ZR, where 0( = row(0(2, ... , O(n) and R is an (n -1) X (n -1) matrix, then the remarks at the beginning of this proof imply that R is a matrix solution of (2.4) with R(w) = R(0)K1. The matrix R(O) is nonsingular. In fact, if there exists an (n -I)-vector c such that R(O)c = 0, then Y 1(0)c - [du(O)/dO]O(c = O. Since Y(O) is nonsingular, this implies C = 0 and, therefore, R(O) is nonsingular. Since R(O) is nonsingular, R(8) is a fundamental matrix solution of (2.4) and K1 is the monodromy matrix of R. Thus, the eigenvalues of K1 are the multipliers of (2.4) and the proof of the lemma is complete. Let us recall some of the previous concepts of stability. If M is a set in Rn, an 7]-neighborhood U 1j(M) is the set of x in Rn such that dist(x, M) < 7]. An invariant set M of (2) is said to be stable if for any e > 0, there is as> 0 such that, for any xO in U,,(M), the solution x(t, xO) is in Ue(M) for all t ~ O. An invariant set M of (2) is said to be asymptotically stable if it is sta~le and in addition there is b > 0 such that for any xO in U b(M) the solution x(t, xo) approaches M as t -!>- 00. If u(t) is a nonconstant. periodic solution of (2), one says the periodic solution u(t) is orbitally stable, asymptotically orbitally stable if the corresponding invariant. closed curve r generated by u is stable,
+
BEHAVIOR NEAR A PERIODIC ORBIT
221
asymptotically stable, respectively. Such a periodic solution is said to be asymptotically orbitally stable with asymptotic phase if it is asymptotically orbitally stable and there is a b > such that, for any xo with dist(xo, r) < b, there is a 7 = 7(XO) such that Ix(t, xo) - u(t - 7)1-+ as t -+ 00.
°
°
THEOREM 2.1. Iff is in CC1(Rn, Rn), u is a nonconstant w-periodic solution of (2), the characteristic multiplier one of (2.3) is simple and all other characteristic multipliers have modulus less than 1 (characteristic exponents have negative real parts), then the solution u is asymptotically orbitally stable with asymptotic phase. PROOF. The solutions in a neighborhood of the orbit r of u can be describE!d by equation (2.1). The above hypotheses and Lemma 2.1 imply that the characteristic multipliers of (2.4) have modulus less than 1. For a sufficiently small neighborhood of r; that is, p sufficiently small, t may he eliminated in (2.1) so that the second equation in (2.1) has the form dp dB
(2.5)
= A(B)p +h(B, p),
where h(B, p) has continuous first partial derivatives with respect to p, h(B, 0) = 0, 8h(B, 0)/8p = 0. Theorems III.2.4 and III.7.2 imply "there are K > 0, at> 0, 1J > such that for Ipol < 1J the solution p(B, Bo, po), p(Bo, Bo , po) = po, of (2.5) satisfies
°
Ip( B, Bo , po) I ~ Ke-a(O-Oo) Ipol,
(2.6)
B ~ Bo.
e
This proves the asymptotic stability of the solution P = 0 of (2.5). Since > 1/2, we have asymptotic orbital stability of r. Equations (2.5) yield the orbits of (2) near r. The actual solutions of (2) near 'r are obtained from the trltnsformation formulas (1.4) and the vector (B(t), p(B(t), Bo , po)) where p(B, Bo , po) satisfies (2.5) and B(t), B(to) = Bo , is the solution ofthe first equation in (2.1) with p replaced by p(B, Bo , po). To prove there is an asymptotic phase shift associated with each solution of (2), it is sufficient to show that B(t) - t approaches a const-antast -+ 00. If B = t q;, then
+
(2.7)
¢(t)
= h(t + q;, p(t + q;, to + q;o, po)).
e
Since > 1/2, the map taking t to B(t) has an inverse. Making use of this fact and recalling that B(t) = t + l/J(t), we have
(2.8)
= l/J(to) +
f
dt
8 (t)
8(to)
fl «(},p(B),B o ,Po»dB dB . .
ORDINARY DIFFERENTIAL EQUATIONS
222 Since
i/, (6,p)1
~L Ipi, relation (2.6) and
Idt/d() i ~2 imply
I
I/I«()'P«()'()O'PO»~~ ~ 2LKe- a (8 -8°)IPol. Since ()(t) -+ 00 if and only if t -+ 00 and
fao 2LKe- a(8 -8 0 )IP o ld() < 0 0 , we have limt_ ao
r to
II «()(t),p«()(t), () 0, Po»dt
exists. From (2.8), this implies t/J(t) approaches a constant as t completes the proof of Theorem 2.1.
-+
00.
This
EXERCISE 2.1. Give an example of an autonomous system which has an asymptotically orbitally stable periodic solution but there is no asymptotic phase. THEOREM 2.2. If u is a nonconstant w-periodic solution of (2) with (n -1) characteristic multipliers of (2.3) .having modulii different from one, then there exist a neighborhood Wr of the closed orbit r in (1) and two sets Sr, U r such that Sr n Ur = r and any solution of (2) which remains in W r for t ~ 0 (t ~ 0) must lie onSr (U r ). If a solution x of (2) has its initial value in Sr (U r ) then x(t)--+-r as t--+- 00 (t--+- -(0). Furthermore,ifp (q) characteristic multipliers of (2.3) have modulii <1 (>1) then Sr (U r ) is either a p-dimensional (q-dimensional) ball times a circle or a generalized Mobius band. PROOF. In a neighborhood of the periodic orbit, r, the orbits of (2) are given by the transformation (1.4) and the solutions p(8) of the real system of differential equations (2.5). The Floquet representation theorem implies that the principal matrix solution of (2.4) can be expressed as P(8)e BIJ where P(8) is periodic of period w. Furthermore, Exercise III.7.2. asserts that P(8) can always be chosen real if it is only required that P be periodic of period 2w. If P has been chosen in this manner, the real transfo~mation p = P(8)r applied to (2.5) yields the equivalent real system
r = Br +!4(8, r), M! where !4(8; 0) = 0, 0!4(8, O)/or = 0, !4(8, r) is periodic in 8 of perroa 2w, and (2.9)
(f-'"<
the eigenvalues of B have nonzero real parts. The conclusion of the theorem now follows immediately by the application of Theorem IV.3.1 to (2.9) and using the transformation p = P(8)r. .
223
BEHAVIOR NEAR A PERIODIC ORBIT
It is reasonable to oall Sr and Ur the stable and unstable manifolds, respectively, of the periodic orbit The following example illustrates that Sr may be a Mobius band. It looks complicated, but actually isn't if one looks at the example in the manner in which it was invented; namely, the development was from the final result desired ~o the differential equation. Consider the equations
r
(2.10)
(a)
r = A(O)r,
(b) 9=1, where r = (rl' r2), 0 are given in terms of the coordinates x, y, z in R3 by x = (rl 1) cos 47T0, y = (rl 1) sin 47T0, Z = r2, and A(O 1/2) = A(O) with
+
+
+
A (0) = 2 [-cos 4-r:0 -7T sm 47T0
+
+
sin 47T0] . cos 47T0
7T
In R3, there is a periodic solution of period 1/2 and the periodic orbit r is the circle ofradius one in the (x, y)-plane with center at the origin. It will be shown that this periodic orbit has stable and unstable manifolds which are Mobius bands. The principal matrix solution of the equation (2.1Oa) is P(O) exp BO where
The matrix A(O) has period 1/2 whereas the matrix P(O) has period 1 in O. Furthermore, it is easily seen that no Floquet decomposition of the principal matrix solution can have real periodic part and at the same time have period 1/2. The characteristic multipliers of the system are the eigenvalues of P(I/2) exp B which are -e- 1 , - e. The st~ble and unstable manifolds of r are given by Sr = {(x,y,
z) : r = e- 28
Ur = {(x, y, z) : r = e28
(cos 27T0, -sin 27T0)a, 0 ~ 0 <
t,
< a < ao}, (sin 27T0, cos 27T0)b, 0 ~ 0 < t, -b o < b < bo}, -aD
where ao , bo are positive constants. It is clear that these surfaces are Mobius bands. EXERCISE 2.2. Prove that any solution of the stable (unstable) manifoldSr(U r ) of in Theorem 2.2 must approach the orbit with asymptotic phase as t-+ 00 (t-+ -(0).
r
r
224
ORDINARY DIFFERENTIAL EQUATIONS
VI.3. Sufficient Conditions for Orbital Stability in Two Dimensions Consider the real two dimensional system i=X(x,g),
(3.1)
iJ
=
Y(x, g),
where X, Yare continuous together with their first partiaJ derivatives in [f,2. Suppose xO(t), yO(t) is a nonconstant periodic solution of period w of (3.1) whose orbit in the (x, g)-plane is r. The linear variational equations of this solution are aX aX (3.2) i = ox x+ ogg, oY oY iJ=- x+-g, ox By
where the functions are evaluated at xO(t), gO(t). Lemma III.7.3 implies that the product of the characteristic multipliers of (3.2) is
Since I is a characteristic multiplier of (3.2), this exponential must be the other multiplier. Using Theorems 2.1 and 2.2, we can therefore state the following sufficient conditions for stability and instability of r. LEMMA 3.1. An w-periodic solutionxO(t), yO(t) of (3.I)has a multiplier of (3.2) less than I and, thus, is asymptotically orbitally stable with asymptotic phase if
fo
W[OX(xO(t), yO(t»
Ox
+
oY(xO(t), yO(t»]
By
dt
and has a multiplier of (3.2) greater than I and hence unstable if this integral is positive. To illustrate the application of this lemma, consider the scalar equation (3.3)
x +!(x)i + g(x) = 0
or the equivalent system (3.4)
i=y, iJ = -g(x) - !(x)y,
where !, g are continuous together with their first derivatives on R, and! has isolated zeros.
BEHAVIOR NEAR A PERIODIC ORBIT
If E(x, y)
= y2J2 + G(X), G(X) = f:g(S) ds, then
225
the derivative E of E
along the solutions of.(3.4) is given by
E=
(3.5)
_f(x)y2 = 2f(x)[G ~E].
If x(t) is a nonconstant solution of (3.3), then there can be no 1such that i(l) = x(l) = O. Ih fact, if this were the case, then x O= x(l) would be a zero of g and uniqueness would imply x(t) = x Ofor all t. Suppose i(l) = 0 and x(l) i= O. Then the solution x(t) has a local extremum at 1and must be either greater or less than x(l) in a neighborhood of 1. Since the zeros of f(x) are isolated, it follows thatf(x(t)) must be of constant sign near 1and consequently E cannot have an extremum at 1. If r is a closed orbit of (3.4) determined by a periodic solution xO(t) of (3.3), then E assumes a minimum value on r for t = 1 and the above remark implies that f(xO(l)) = 0, y(l) = iO(l) i= O. Th~refore, if G(xO(l)) = h, then E(t) > h for all t. From (3.5), it follows that
E E - h
and integration over
r
+ 2f =
2(G -h)f E- h '
yields
f fdt=f .(G-h)f dt. r r E-h
(3.6)
From Lemma 3.1 and the special form of (3.4), the linear variational equation of the orbit r will have a multiplier <1 if (3.6) is positive and >1 . if (3.6) is negative. In particular, this multiplier will be <1 if G(x) takes the same value 'Y for all zeros of f(x) and [G(x) l1f(x) is positive when f(x) =1= O. As a particular case which will be useful in the sequel, we state LEMMA 3.2. If f, g have continuous first derivatives in R aqd (i) f (x) < 0 for a. < x < f3, f (x) > 0 for x < a., x> f3 and a. < 0 < f3, (ii) xg(x) > 0 for x i= 0, (iii) G(a.)
= G(f3) where G(x) =
f: g(s) ds,
then any closed orbit of (3.4) has a characteristic multiplier in (0, 1) and is thus asymptotically orbitally stable with asymptotic phase. An important special case of Lemma 3.2 is the van der Pol equation x -k(1 -x2 )i+ x=O, k >0.
VI.4. Autonomous Perturbations Consider the system of equations (4.1)
i =f(x) + F(x, e),
226
ORDINARY DIFFERENTIAL EQUATIONS
wherej: Rn~Rn, F: Rn+1~Rn are continuous,j(x),F(x, e) have continuous partial derivatives with respect to x and F(x, 0) = 0 for all x. If system (2) has a'nonconstant periodic solution u(t) of period w whose orbit is r, then the linear variational equation for u is .
(4.2)
oj(u(t»
y=--y.
ox
THEOREM 4.1. If n - 1 of the characteristic multipliers 1-'1, ... , I-'n-1 of (4.2) are =1= 1, then there is an eo > 0 and a neighborhood W of r such that equation (4.1) has a periodic solution u*( . , e) or'period w*(e), 0 ~ lei ~ eo, such that u~J '_,_Q) ~· . .. ,Pn-l is a root of one, then forany constant k >-0, theneighborhood W can be chosen so that u*(· ,e) is the only periodic solution of (4.1) of period ~k in W. If Ill,' .. Pn-l have modulii <1, then u*(· ,E) is asymptotically orbitally stable with asymptotic phase and is unstable if any Ilj has modulus> 1. PROOF. Theorem 1.2 and the transformation (1.4) imply that system (4.1) is equivalent to the system (1.11) with F(t, x) replaced by F(x, e) above. Elimination of t in these equations yields a system
(4.3)
dp dO
=
A(O)p
+ R(O, p, e),
where A(O), R(O, p, e) are w-periodic in 0 and R is in !l'ifi(TJ, M). where !l'ifi(TJ, M) is defined in Section IV.2. Lemma 2.1 implies that the,characteristic multipliers of dp/dO = A(O)p are the numbers 1-'1, ... , I-'n-1. Theorem IV.2.I implies the existence at an eo> 0, po> 0 and an w-periodic solution p*(O, e) of (4.3) continuous in 0, e, 0 ~ lei ~ eo, p*(O, 0) = 0, and p*(O, e) is the only w-periodic solution in the region Ipl < po. Theorem IV.3.I yields the stability properties of p*(O, e) when 1-'1, ... , I-'n-1 have modulii different from one," Usmg this p*(O, e) in the first equation in (1.11) and letting O(t) be the solution of this equation with 0(0) =0, 'One finds there is a unique continuous w*(e), 0 ~ lei ~ eo, w*(O) =w, such that O(w*(e» = w, 0 ~ lei ~ eo. The function O(t), p*(O(t), e), and the transformation (1.4) yield th~ desired periodic solution of (4.1) and the stability properties are as stated in the theorem. Suppose now that no I-',,"j = 1, 2, ... , n - 1, is a root of 1. For' any integer k, the periodic solution u of (2) has period kw and, therefore, all" of the functions in the equations of the above proof can be considered as, ~riodic in 0 of period kw. The hypothesis on the 1-" imply by Theorem IV~2.1 that there is a periodic solution of (4.3) of period kw for 0 ~ lei ~ eo and this solution is unique in the region,lpl < po. Therefore, this solution must be the
BEHAVIOR NEAR A PERIODIC ORBIT
227
p*(8, e) given above. But, any periodic solution x of (4.1) which lies in a suffi-
ciently small neighborhoorl of r must define a closed curve in Rn with a parametric representation of the form x = u(8) + Z(8)p(8, e) where p(8, e) is a periodic solution of (4.3) of period kw for so~e integer k. This completes the proof of the theorem. EXERCISE 4.1. Suppose g(x, y) has cc-ntinuous first derivatives ~ith respect to x., y. Show there is an eo > 0 such that for Ie I < Eo the equation i
X - k(1 -x 2 )i + x
=
eg(x, i), k >0,
has a unique periodic solution in a neighborhood of the unique periodic orbit of the van der Pol equation
x-
k(1 - x 2 )i + x
=
o.
Can you prove that this orbit is asymptotically orbitally stable with asymptotic phase? EXERCISE 4.2. Give an example of an autonomous system i =f(x) which has an w-periodic orbit r whose linear variational equation has 1 as a simple multiplier and yet, in any neighborhood of r, there are other periodic orbits. EXERCISE 4.3. Is it possible for an equation x =f(x) to have an isolated w-periodic orbit r whose linear variational equation has 1 as a simple multiplier and yet, there is a perturbation eg(x) such that in any neighborhood of r there is more than one periodic orbit?
VI.5_ Remarks and Suggestions for Further Study The particular construction of the moving orthonormal system given in Section 1 is based on a presentation given by Urabe [1] for the case in which the' closed curve r is a periodic orbit. The manner in which the coordinate system is used"to obtain a set of differential equations equivalent to (4) in a neighborhood of r is different. Since Urabe discusses only the case of a periodic orbit, it is possible to obtain the equations for the normal variation explicitly in terms of t and not 8 as in (1.11). Lemma 1.1 was given by Diliberto and Hufford [1]. Lemma 3.2 is due to Coppel [1, p. 86]. The stability Theorem 2.1 was known to Poincare for analytic systems. In general, it is very difficult to discuss the behavior of solutions of (2) in a neighborhood of an orbit when ~ore than one characteristic multiplier of (2.3) has modulus equal to one. Hale and Stokes [1] have given the following
228
ORDINARY DIFFERENTIAL EQUATIONS
result. If system (1) has a k-parameter family of periodic solutions, then the linear variational equation has k characteristic multipliers equal to one. If the remaining multipliers have modulus less than one, then every orbit in this family is stable and, in fact, the solutions of (2) near this family approach a member of this family asymptotically with asymptotic phase. Autonomous perturbations of nonlinear systems which possess a k-parameter family of periodic solutions is extremely difficult and is discussed at great length in the book of Urabe [1]. Under the hypothesis that only one multiplier of the linear variational equation has modulus one, the behavior of solutions near a periodic orbit of a nonautonomous perturbation of (2) will be discussed in the next chapter. If the nonautonomous perturbation is "small " for large t; that is, the system is" asymptotically" autonomous, then one can discuss the qualitative relationship between all solutions of the nonautonomous equation and the autonomous one without any hypotheses regarding the solutions of the autonomous one (see Markus [1], OpiaJ [1], Yoshizawa [1]). For nonautonomous perturbations of k-parameter families of periodic solutions, see Hale and Stokes [1], Yoshizawa and Kato [1].
CHAPTER VII Integral Manifolds of Equations with a Small Parameter
Assuming that system (VI.2) has an invariant set which is a smooth Jordan curve r, it was shown in Chapter VI that a transformation of variables could be introduced in such a way that the behavior near r of the solutions of a perturbation of (VI.2), namely (VIA), is reduced to a study of the equations () = g(O) 0(t, 0, p),
+ p =C(O)p + R(t, 0, p),
where p represents the normal deviation from rand g( 0) represents the behavior of the solutions of the unperturbed equation on r. If g(O) == 1, the curve r corresponds to a periodic orbit and if g vanishes at some point, then an equilibrium point lies on r. For g( 0) 1 and the perturbations independent of t, some specific results were given in Chapter VI for the above equation. The analysis for this case is extremely simple since the problem is easily reduced to the study of the behavior near an equilibrium point of a nonautonomous system. On the other hand, if 0, R contain t explicitly, then one cannot reduce the question to such a local problem since t cannot be eliminated from the equations even if g(O) == 1. One must study the behavior near the invariant set itself. If g(O) has a zero for some value of O,then the problem remains nonlocal even if 0, R are independent of t. Other difficulties also arise in this latter case and will be discussed later. The purpose of this chapter is to study such nonlocal problems for even more general systems than the above. More specifically, we will be concerned with equations of the form
=
(1)
() = 0*(t, 0, x, y, e) = i
+ F(t, 0, x, y, e), B(O, elY + G(t, 0, x, y, e),
= A(O, e)x
if =
w(t, 0, e)
+ 0(t, 0, x, y, e),
230
ORDINARY DIFFERENTIAL EQUATIONS
where (e, 8, x, y) is in R X RIt X Rn X Rm, the matrix A(8, e) is of such a nature that the solutions of it = A(8, e)x approach zero exponentially as t -+- 00, the solutions of iJ = B( 8, e)y approach zero exponentially as t -+- - 00, for some class of functions 8(t), and F, G, 0 are sufficiently small in some sense for x, y, e small. These statements will be made more precise later. It will not be assumed that the functions in (1) are periodic in the vector 8 although it will be assumed they are bounded for x, y in compact sets. In .specific applications, a frequently occurring special case of (1) does have all functions periodic in 8. Such problems arise naturally from the study of the local perturbation theory of differential equations near an invariant torus. In many important situations, the flow on the invariant torus is parallel in the sense that all solutions are either periodic or quasiperiodic and then w(t, 8, e) in (1) is a constant vector. If equations (1) arise from the local perturbation theory of a periodic orbit, then 8 is a scalar, w is a constant and the Floquet theory for periodic systems permits one to take the matrices A(8, e), B(8, e) independent of 8. For .the general perturbation theory of invariant torii with parallel flow, the matrices A(8, e), B(8, e) cannot be taken independent of 8 because there is no Floquet theory for general almost periodic systems. The exercises in Section 8 illustrate the many ways in which equations of type (1) arise. Definition 1. A surface S in (z, t)-space is an integral manifold of a system of ordinary differential equations z= Z(z, t) if for any point P in S, the solution z(t) of the equation through P is such that (z(t), t) is in S for a11 t in the domain of definition of the solution z(t).
Our interest in this chapter lies in determining in what sense the qualitative behavior of the solutions of (1) and the system () = w(t,
(2)
8, 0),
it = A(8, O)x,
iJ = B(8, O)y, are the same. This system has an integral manifold S in R X RIt X Rn X Rm whose parametric representation is given by
S
= {(t, 8, x, y): x = 0, y =
O}.
Furthermore, under the stability properties alluded to earlier, the manifold S has a type of saddle point structure associated with it. In fact, there is an ;R X Bit X Rn dimensional manifold of solutions of (2,:" which approach S as t-+-oo and an R X Bit X Rm dimensional manifold Qtaolutions of (2) which approach S as t -+- - 00. Intuitively, one. expects that 0, F, and G small enough for e, x, y small. will imply the existence of some other integral
INTEGRAL MANIFOLDS OF EQUATIONS WITH A SMALL PARAMETER
231
manifold S£ of (1) which is close to S for e small and has the same type of stability properties as the integral manifold S of (2). _ Therefore, in this chapter, we determine conditions on 0, F, G which ensure that (1) has an integral manifold of the form x =1(t, 8, e), y = g(t, 8, e), (~, 8) in R X Rk, which for e = 0 reduce to x = 0, y = O. For this specific case,. such surface will be an integral manifold for (1) if the triple of functions
a
[8(t)
= 8(t, 80, to),
y(t) = g(t, 8(t), e)],
x(t) =1(t, 8(t), e),
is a solution of (1) for every to in R, 80 in Rk. Section 1 consists of a historical and intuitive discussion of the problems involved in determining integral manifolds for (1) as well as possible approaches to the solutions of these problems. The main theorems are stated in Section 2 with the proof being delegated to Sections 3, 4, 5, 6. In Sectipn 7, applications of the results of Section 2 are given for equations which are perturbations of equations possessing an elementary periodic orbit. Also, in Section 7, the method of averaging of Chapter V is extended to averaging with respect to t as well as the variables 8 in (1). The exercises in Section 8 illustrate more fully the implications of the results of this chapter.
m.l.
Methods of Determining Integral Manifolds
This section is devoted to an intuitive discussion. of integral manifolds as well as some methods that have proved successful in the determination of-integral manifolds. We choose for our discussion the problem of perturbing an autonomous system (1.1)
X= X(x),
which has a nonconstant wo-periodic solution u(t) such that n -1 of the characteristic exponents of the linear variational equation (1.2)
.
y=
oX(u(t))
ax
y
have negative real parts. As we have seen in Chapter VI, this implies the orbit described by u is asymptotically stable and this in turn implies that the cylinder (L3)
S = {(t, x): x= u(8), 0 ~ 8 ~ wo,
-00
in (t, x)-space is asymptotically stable. The cylinder S is an integral manifold of (1.1) in R X Rn. If we introduce the coordinate system (1.4)
x = u(8)
+ Z(8)p
232
ORDINARY DIFFERENTIAL EQUATIONS
of Chapter VI, then in a neighborhood of S the solutions of the equation are described by (1.5)
(j
= 1 + 8(0, p),
p=
A(O)p
+ R(O, p),
where 8(0, p) = O(lpi), R(O, p) = O(lp12) as Ipl-O, all functions are woo periodic in 0, and the characteristic exponents of dp/dO = A(O)p have negative real parts. From the Floquet theory, a fundamental system of solutions of this linear equation is of the form P(O)e BB , P(O +wo) = P(O). If we let p(t) = P(O)z(t) and use the fact that (j = 1 + O(lpi), we obtain an equivalent system (1.6)
(j = 1
z=
+ 81(0, z),
Bz +Z(O, z),
where 81(0, z) = O(lzi), Z(O, z) = O(lzI2) as Izl-O and the eigenvalues of B have negative real parts. For the sake of our intuitive exposition, we need the following elementary result proved in Chapter X, Lemma 1.6. There is a positive definite matrix C such that any solution of the linear system z = Bz with initial value on the ellipsoid z'Cz = C > 0, a constant, must enter the interior of this ellipsoid for increasing. time. In Lemma X.1.6, it is shown that the matrix C = Ioc.o eB'teBtdt satisfies the desired properties. Since Z(O, z) = O(lzI2) as
°
Izl-O,
there is a co> sufficiently small such that any solution of (1.6) with initial value on the set
U8(c)
= u(O) + Z(O)P(O)z, < t < oo}, < c ~ Co,
= {(t, x): x -00
°
°
~
0~
Wo,
z'Cz = c,
must enter this set for increasing t (and therefore remain inside this set). The set U 8 (c) projected into the x·space is a tube surrounding the closed curve ce = {x: x = u(O),O ~ 0 ~ wo}. Intwodimensions, it is an annulus around ce. In this case, the geometry is very simple since Us(c) represents a cylinder inside S and a cylinder outside S. Now if the original differential equation (1.1) is perturbed to the form (1.7)
x = X(x) + eX*(t, x)
°
where X*(t, x) is bounded in a neighborhood of S, then for a given c > and e sufficiently small, the solutions will still be entering Us(c) as time i~creases. Therefore, for a fixed c > 0, we would expect some kind of integrll surface inside Us(c} for e small and it should look similar to a cylinder. However, even if such a surface exists we would not expect the solutions on this surface to behave in a manner similar to the behavior of the solutions on the original
INTEGRAL MANIFOLDS OF EQUATIONS WITH A SMALL PARAMETER
233
S since these solutions have no strong stability properties associated with them. This is precisely why we discuss the preservation of the structure and stability properties of the set S rather than the preservation of any such properties for a particular solution on S. Some methods for asserting the existence of integral manifolds for the perturbed equation are now given. If the transformation (1.4) is applied to (1.7), the equivalent equations are (1.8)
0=1 + 0(0,p) + e0*(t,0,p), p = A(O)p + R(O, p) + eR*(t,
0, pl.
Method 1 (Levinson.Diliberto). If we assume that the perturbation term is wI-periodic in t, then the above discussion of the geometric implications of stability allows one to obtain an annulus map via the differential equation (1.7) in the following way. Let U81'(c) be U 8 (c) (\ {t =T}; that is, the cross section of Us(c) at t = T. Actually, all of these cross sections are the same and equal to U so(c). If x(t, xo) is the solution of the perturbed equation (1.7) with initial value Xo at t = 0 and e is sufficiently small, then for any T > 0, the function X(T, .) is a mapping of Uso(c) into the interior of UST(c) = Uso(c); that is, a mapping of Uso(c) into itself. Because of the strong stability properties of the curve q; and the fact that solutions "rotate" around this curve, one would suspect that there is a curve q;£T su~h that X(T, q;£T) = q;£T and q;OT = q;. If this is true and if T is chosen to be WI, then the periodicity in the equation implies X(kWl' q;£()).) = q;£()), for every integer k. By considering the "cylinder" generated by the solutions which start onq;£())" we obtain an invariant manifold of the equation. In this case, the solutions on the cylinder can be described by the solutions of an equation on a torus, the torus being obtained by identifying the cross sections of the cylinder at t = 0 and t = WI. The basis of this idea was proposed and exploited in a beautiful paper of Levinson (1950). Using the idea of the proof of Levinson, Diliberto and his colleagues greatly simplified and improved the work of Levinson as well as discussing integral manifolds of a much more complicated type. In fact, ifthe perturbation·term X* in (l.7) is an arbitrary quasi-periodic,Junction, then it can be written in the form Fet, t, ... , t, x) where F(t1, ... , t p , x) is periodic in tj of period Wj ,j = 1, ... ,p. The functions 0*, R* in (1.8) have this same periodicity structure in t. By artificially introducing variables such that ~j = 1, we obtain from (l.8) a system
'j
. ~j=l,
0=1
P=
j=I,2, ... ,p,
+ 0(0, p) + eP*(O, " p), + R(O, p) +eQ*(O, ',p),
A(O)p
where P*(O, " p), Q*(O. " p), ,~ ('1, ... , 'p), are periodic in
°
and ,. This
284
ORDINARY DIFFERENTIAL EQUATIONS
is a spedal case of (1) with 8 in (1) a (p + I)-vector consisting of the scalar 8 above and the p-vector ,. Notice that the equations no longer contain t explicitly. Diliberto has introduced an ingenious device to discuss integral manifolds for such equations by generaliziri.g the idea of the period map. The inter~ted reader may consult the references for the details of this method aswel1.ll.s the applications. MetkorJ 2 (Partial Differential Equations).
(1.9)
(a)
6=
(b)
i; = A(O)x
Suppose we have a system
w(O)+ 0(0, x),
+ F(8, x),
where .0 in Rk, x in R7I are vectors, and all fun9tions are periodic in 0 of vector period oo. If S = {(O, x): x =J(o'), J(O 00) =J(O), 0 in Rk} is to be an integral manifold of this equation, then x(t) =J(O(t», where O(t) is a solution of
+
6=
(1.10)
w(8)
+ 0(0,J(0»,
must satisfy (1.9b). Performing the differentiation, we satisfy the partial differential equation (1.11)
8J
find
that
J must
.
80 [w(O) + 0(8,j)] -A(O)J= F(O,j), J(O + 00) =J(O).
This method has not been completely developed at this time but a beginning has'been made by Sack~r [1]. The case where w, A are independent of 8 is not too difficult with this approach since one can integrate along characteristics to obtain essentially the same method presented below. If there are values of 0 for which w(O)= 0, then the problem is much more difficult since the solution will not in general be as smooth as the coefficients in the equation. The smoothness properties depend in a delicate manner upon w, A. Another difficulty that arises in this problem is that .the usual iteration procedures involve a loss of derivatives. To circumvent this difficulty, Sacker solves in each iteration the. elliptic equation ,",floJ
8J + 88 [w(8) + 0(8, j)] -
A(8)J = F(8, j),
J(8
+ 00) =J(O),
for,", small and flp the Laplacian operator. The solutions then have as, many derivativescas desired, but there is a delicate analysis involved in q,t,oosing '"' -+0 in such a way as to obtain a solution of the original equation: :Further research needs to be done with this method to see if it is possible to obtain results )Vhen the perturbation terms depend upon t in a general way and also to ~iscuss the stability properties of the manifold.
INTEGRAL MANIFOLDS OF EQUATIONS WITH A SMALL PARAMETE
235
Method 3. (Krylov-Bogoliubov;Mitropolski). Circa 1934, Krylov and Bogoliubov made a significa.nt contribution to this problem in the following way. They defined a mapping of "cylinders" into "cylinders" via the differential equation (1.8) such that the fixed points of this map are integral manifolds of the equation which for e = 0 reduce to S. They also discussed the stability properties of this manifold. These methods do not use any particular properties of the dependence of the perturbation terms upon t. To illustrate the ideas, consider the equation
() = 1 = 9(t,e ,p,e) (1.12)
p= Ap + R(t,e ,p,e)
where.-1 is an n X n constant matrix, ReM < 0, 9(t,e,0,0) = 0, R(t,e,O,O) = 0, 3R(t,e,0,0)/3p = O. We look for integral manifolds of (1.12) of the form p = [(t,e ,e) with I[(t,e ,e) 1 bounded for (t,e) E R2, [(t,e ,e) lying in a small neighborhood of p = 0 for e small. Let S = {f:R X R -+ R n continllous and bounded) with the topology of uniform convergence. Suppose (1.12) has an integral manifold [(t,e,e), with [(.,. ,e) E S. If e (t,to,eo'!) is the solution of the equation (1.13)
() = 1 + 9(t,e.!(t,e,e),e),
e(to,to,eo.!) = eO,
then (e(t,to ,eo.!),f(t,e(t,to,eo.!),e) must satisfy (1.12) for all (to,eo) ER2. Therefore, the variation of constants formula implies
[(t,e(t,to ,eo.!),e) =
e'!- (t-t o)[(to ,eo,e) +
f f!1 to
(t-s)R(s,e(s,to,eo.!).!(s,e(s, to,eo.!),e),e)ds
and
e- A (t-to)[(t,e(t, to, eo,[),e) = [(to,eo,e)
+
fto
f!1(t o-s)R(s,e(s,to,eo .!),
[(s,e(s,to,eo.!),e),e)ds. Since [(t,e,e) is bounded and exp(-A(t - to» formula for the function [(to,eo,e); namely,
-+
0 as t
-+
-00, we have a
t
(1.14)
[(to,eo,e) =
LO.. eA(to-s)R(s,e(s,to,eo.!).!(s,e(s,to,eo.!),e)ds.
For a given [E S, the right hand side of Eq. (1.14) can be considered as a transformation ~:S -+ S and the sought for integral manifolds are flXed points of !F. By careful estimation, one can apply the contraction mapping principle
236
ORDINARY DIFFERENTIAL EQUATIONS
to obtain ftxed points of ~in an appropriate bounded subset of S consisting of functions with a_ sufficien_tly sm~llipschitz constant in 8.
Vll.2. Statement of Results In this section, we give detailed r~sults on the existence and stability of integral manifolds of (1). Let O(G, eo) = {(x; y, e): Ixl < G, Iyl < G, 0< e ~ eo}. The following hypotheses on the functions in (1) will be used: (HI) All functions A, B, Rk X O(G, eo).
W,
0, F, G are pontinuous and bounded in R X
(H 2 ) The functions A, B, W, 0, F, G are lipschitzianin 8 with lipschitz constants r(e1), r(e1), L(e1), 7J(p, e1), y(p, e1), y(p, e1) respectively, in R X Rk X O(p, e1) where r(e1), L(e1), 7J(p, e1), y(p, e1) are continuous and nondecreasing for 0 ~ p ~ G, 0 < e1 ~ eo.
(Hs) The functions 0, F, G are lipschitizian in x, y with lipschitz constants p,(p, e1), S(p, e1), S(p, e1), respectively, in R X Rk X O(p, e1), where p,(p, e1), S(p, el) are continuous and nondecreasing for 0 ~ p ~ G, 0 < e1 ~ eo.
(H4) The functions IF(t, 8,0,0, e)l, IG(t, 8,0,0, e)1 are bounded by N(e) for < e ~ eo where N(e) is continuous and nondecreasing
(t,8) in R X Rk, 0 for 0 <e ~ eo.
(H5) There exist a positive constant K and a continuous positive function <x(e), 0 < e ~ eo, such that, for any continuous function 8(t) defined on (-00, 00), and any real number T, the principal matrix solution <()(t, T), '¥(t, T) of x = A(8(t), e)x, if = B(8(t), e)y, respectively, satisfy (2.1)
I<{)(t, T)I
~ Ke-a(e)(t-T),
I'¥(t, T)I
~ Kea(e)(t-T),
t t
~
~
T,
T.
(H6) Let p(d, D, e), q(d. D, e) be defined by (2.2)
p(d, D, e)
Kr(e)
= y(D, e) + S(D, e)d + - - [S(D, e)D + N(e)], <X
q(d, D, e)
= L(e) + 7J(D, e) + p,(D, e)d,
and suppose that (2.3)
(a) (b)
(c)
> 0, S(D, e)D + N(e) < <x(e)DjK, Kp(d, D, e) < [<x(e) - q(d, D, e)]d, <x(e) - q(d, D, e)
(d) p(d, D, e)p,(D, e) <x(e) - q(d! J), e)
+ S(D
e) ,
< <x(e) , 2K
INTEGRAL MANIFOLDS OF EQUATIONS WITH A SMALL PARAMETER tor 0
<
237
e ~ eo.
We are now in a position to state THEOREM 2.1. If system (1) satisfies hypotheses (H1) - (H6), then there exist functions f (t, e, e) in Rn, g(t, (), e) in Rm which are cOIitinuous in R X Rk X (0, e1], bounded by D, lipschitzian in () with lipschitz constant fl., such that the set Se = {(t, (), x, y): x =f(t, (), e), y
= g(t, e, e), (t,e) in R X Rk}
e
is an integral manifold of (1). If the functions in (1) are periodic ,in with vector period w, then f (t, (), e), g(t, (), e) are also periodic in () with vector period w. If the functions in (1) are T-periodic in t thenf(t, (), e), g(t, (), e) are T-periodic in t. If the functions in (1) are almost periodic in t, then so aref, g. Before proving this theorem, we state some immediate Corollaries which we again state as theorems because of their importance in the applications. THEOREM 2.2. Suppose system (1) satisfies (Hi)-(H5) even for e = 0 and O((e) = 0( a constant. If 7](0, 0) = y(O, 0) = 8(0, 0) = N(O) = 0 and 0( L(O) > 0, then there are e1 > 0 and continuous functions D(e), Ll(e), o < e ~ e1, approaching zero as e -+ 0 such that the conclusions of Theorem 2.1 are valid for this D(e), Ll(e). PROOF. From the hypothesis of Theorem 2.2, 0( - L(O) > O. Therefore, we may take eo such that 0( - L(e) > 0, 0 ~ e ~ eo. Since 7](0,0) = 0, there are positive e1, Ll1, D1 so that (2.3a) is satisfied for 0 < e ~ e1, Ll ~ Ll1, D ~ D1. Since 8(0, 0) = 0 and N(O) = 0, it follows that one can further restrict e1 and choose D(e) so that .p(e) -+0 as e -+0 and (2.3b) is satisfied for 0 < e ~ e1. Since ')1(0, 0) = 0, 8(0,0) = 0 it follows that e1 and a function Ll(e) -+0 as e -+0 can be chosen so that (2.3c) is satisfied for 0 < e ~ e1. Since p(Ll(e), D(e), e), 8(D(e), e) -+ 0 as e -+ 0 if Ll(e), D(e) are chosen as above, it follows that one can further restrict e1 > 0 such that (2.3d) is satisfied· for 0<: e ~ e1. Theorem 2.1 is therefore applicable to complete the proof of Theorem 2.2. If w(t, e) is a constant in (1) then L = 0 and the condition 0( - L > 0 is automatically satisfied from (H5). The hypotheses (H1)-(H4) in Theorem 2.2 merely express smoothness and smallness conditions on the perturbation functions 0, F, G.
e,
COROLLARY 2.1. Suppose e,F,G satisfy hypotheses (Hl) - (H4) even for = O. Suppose w is a constant k-vector, A is an n X n constant matrix whose eigenvalues have negative real parts, B is an m X m constant matrix whose eigenvalues have positive real parts. Then the conclusions of Theorem 2.2 are valid for the system €
238
ORDINARY DIFFERENTIAL EQUATIONS
fJ = w+ 8(t,e,x,y,e)
X=Ax + F(t,e,x,y,e)
y= By + G(t,e, x, y,e) Consider the system
8 = 'W + e0et, 0, x, y, e), i = eA(O)x + eF(t, 0, x, y, e),
(2.4)
y = eB(O)y
where
'W
+ eG(t, 0, x, y, e),
is a constant.
THEOREM 2.3. Suppose A, B, 'W, 0, F, G in system (2.4) satisfy hypotheses (Hl)-(H4) even for e = and y(O, 0) = 8(0, 0) = N(O) = 0. If oc =eOCI, OCI > 0, a constant, in (H 5) and OCI -lim e->o7J(O, e) > 0, then the conclusions of Theorem 2.2 are valid for system (2.4).
°
°
PROOF. It is sufficient to satisfy relatio~s (2.3) with L = and oc(e) replaced by OCI since the form of equation (2.4) implies that e is a common factor of all functions in (2.3). The proof proceeds now exactly as in the proof of Theorem 2.2. Consider the system (2.5)
e8
== 'W + 0(t, 0, x, y, e),
+ F(t, 0, x, y, e), ey = B(O)x + G(t, 0, x, y, e),
ex = A(O)x
where
'W
is a constant.
THEOREM 2.4. Suppose A, B, 'W, 0, F, G in system (2.5) satisfy (HI)-(H4) even for e = 0, y(O, 0) = 8(0, 0) = N(O) = 0. If oc = ocI/e, OCI > 0, a constant in (H5) and OCI -lime->o7}(0. e) > 0, then the conclusions of Theorem 2.2 are valid for system (2.5). PROOF. The proof i" essentially the same as the proof of Theorem 2.3. The proof of Theorem 2.1 will be broken down into simple steps in order to clarify the batlic ideas. Some of these steps are of interest in theIllselves and are stated as lemmas. One could use the same method of proofr!is given below to state results on integral manifolds involving systems Which are combinations of systems (1), (2.4) and (2.5). These results are easily obtained when the need arises and it does not seem worthwhile to state them in detail.
INTEGRAL MANIFOLDS OF EQUATIONS WITH A SMALL PARAMETER
239
Vll.3. A "Nonhomogeneous Linear" System
In the proof ofTheorem 2.1, we will use successive approximations in a manner very similar to that used in Chapter IV for the results on the behavior near an equilibrium point. Let Sn = {f: (-00, (0) X Rk -'?Rn which are continuous and bounded}. Forany I in Sn,define II! II = sup{I!(t, B)1. (t, B) in R X Rk}. The idea for successive approximations in (I)' for an integral manifold is to let x = I(t, B), I in Sn, y = g(t, B), gin Sm, in the first equation in (1) to obtain an equation in B alone say (} = h(t, B,I, g). This equation can be solved for B(t,7", ,,f, g) where B(7") for any 7", , in R X Rk. This function of B is then substituted in the second and third eOl l1ttions in (1) to obtain equations of the form
=,
x= y=
+ j(t, B(t)), B(B(t))y + g(t, B(t)),
A(B(t))x
where j, g depend upon I, g, but, more importantly, the initial data for B(t); namely 7", ,. The problem is then to determine a bounded solution of this system as a function of (7", O. Thus, we are led to a type of "nonhomogeneous linear" system. We refer to the equations in this manner because at each stage in the iteration process the equations are linear in x, y; that is, linear in the coordinates through which the integral manifold is defined. For simplicity in the notation, we will assume that the y-equation in (1) is absent. The general case follows ..itlong the same lines. For any P in Sk, Q in Sn, we first consider the" nonhomogeneous linear" system (3.1)
(a)
(b)
= P(t, B), x = A(B)x + Q(t, (}
B).
For any (7", ') in R X Rk, let B*(t) = B*(t, 7", " P) be the solution of (3.1a) with B*(7") = ,. Since P(t, B) is bounded, such a solution always exists on (-00, oo):We wish to find a function X(·, " P, Q) in Sn such that (B*(t), X(t, B*(t), P, Q)),
tin (- 00, (0),
is a solution (3.1) for t in (-00, (0) and all (7", ') in R X Rk. If we can accomplish this, then the set !/ = {(t, B, x) : x = X(t, B, P, Q), (t, B) in R X R k} is an integral manifold of (3.1). To derive the equation for such a function X, let cI>(t, 7", " P) be the principal matrix solution of the linear system (3.2)
x=
A(B*(t, 7", " P))x
240
ORDINARY DIFFERENTIAL EQUATIONS
The variation of constants formula applied to (3.1b) yields X(t, e*(t,
T, "
P),
1', Q) = (
T, "
P)X(:r, " P, Q)
t
+ f
P)Q(s, e*(s,
T, "
P» ds,
t E (-
00,
(0).
T
If A is independent of e, then this relation is much simpler since
=
" P)X(t, e*(t, T, " P), P, Q)
- f
tE
(-00,.
(0).
T
Since X is assumed to belong to S" and, in particular, is bounded, we can let t approach -00 in this relation and use hypothesis (H5) to obtain (3.3)
X(t, e, P, Q)
=
fo
-co
+ t, e, P)Q(u + t, e*(u + t, t, e, P» du,
where we have replaced T, 'in the end result by t, 8. In the manner in which relation (3.3) was obtained, it follows that (3.3) defines an integral manifold of (3.1) and, furthermore, it is the only such integral manifold which remains in a region for which the x-coordinate is bounded. These facts are summarized in LEMMA 3.1. If (H5) is satisfied, then for any Pin Sk, Qin S", equation (3.1) ~s an integral manifold defined parametrically by X(t, P, Q) in (3.3) and it is the only integral manifold of (3.1) for which the x coordinate is bounded. Our initial goal is to derive some properties of the function X(t, P, Q) defined by (3.3). In particular, we wish to discuss bounds and smoothness properties of X as a function of the bounds and smoothness properties of P, Q. The basic result for the" nonhomogeneous linear" system (3.1) is the following:
e,
e,
LEMMA 3.2. Suppose A(e), P(t, e), Q(t, e) are lipschitzian in e with lipschitz constants r, L(P), M(Q), respectively, and hypotheses (H5) is satisfied with at: > L(P). If X(', " P, Q) is defined by (3.3), then X(', " P, Q) belongs to S", defipes an integral manifold of, (3.1) and satisfies
(3.4) (a)
IX(t, e, Pi Q) -X(t,
8, P, Q)I ~ at: _
K Kr]' L(P) [M(Q) +-;;IIQII Ie -
81,
INTEGRAL MANIFOLDS OF EQUATIONS WITH A SMALL PARAMETER
(b)
IIX(·, ., P, Q) -X(·,
(c)
IIX(·,·, P,Q)
., P,
Q)II
~
K
- IIQ -QII, X(·, oc
241
., P, 0) =0,
K
-X(·,·, P,Q)II ~ oc(oc-L(P))
[M(Q)
X
+ ~r IIQII]IIP-PII,
for all t in R, 8, 8 in Rk, P, Pin Sk, Q,Q in Sn. If P(t, 8), Q(t, 0) a.re periodic in 8 with vector period w, then X(t, 8, P, Q) is periodic in 8 with vector period w. If P(t,8), Q(t, 8) are T-periodic (or almost periodic) in t, then X(t, 8, P, Q) is T-periodic (or almost periodic) in t. If P(t, 8), Q(t, 8) are independent of t, then X(t, 8, P, Q) is independent of t. PROOF. Relation (3.4b) is almost immediate since X(·, ., ., Q) is linear in Q and hypothesis (H 5) implies that
IIX("., ·,Q)II~KIIQII. oc
The relations (3.4a), (3.4c) are more difficult to obtain since 8*(·, ., 8, P) depends in a nonlinear fashion on 8 and P. This nonlinear dependence is also reflected in the function <1>(., ., 8, P) if A depends on 8. Our first objective is to obtain estimates of this dependence on 8, P. Since P(t, 8) is lipschitzian in 8 with lipschitz constant L(P), a simple application of differential inequalities to (3.1a) yields the estimates (3.5)
(a)
W'·(t,
'T, "
P) - 8*(t,
'T,
t
P)I ~
(b)
18*(t,
'T, "
P) - 8*(t,
'T, "
P)i ~
eL(P)IH'I' eL(P)lt-TI
L(P)
~I,
-1 liP -
PII,
for all t, 'T in Rand P, P in Sk. In fact, both of these relations are easily obtained from the relation D+y(u) ~ L(P)y(u)
+ liP -
PII
where D+ is the right hand derivative, y(u) = 18*(u, 'T, " P) - 8*(u, 'T, " p)l to obtain (3.51;» and y(u) = 18*(u, 'T, " P) - 8*(u, 'T, ~, P)I, P = P to obtain (3.5a) We now obtain estimates for the dependence of (t, 'T, 8, P) on 8, P. If we use the fact that this is a principal matrix solution of (3.2) and the difference (u, 'T, 8, P) - (u, 'T, 8, p) is a solution of the matrix equation dx
.
- = A(8*(u, 'T, 8, P))x + [A(8*(u, du
- A(8*(u,
'T, 'T,
8, P))
8, P))](u" 'T, 8, p),
242
ORDINARY DIFFERENTIAL EQUATIONS
then the variation of constants formula yields CP(t, u
+ t, (), P) -
+ t, 8, P) = f
o
CP(t, u
CP(t, v
+ t, (), P)
It
x [A(()*(v + t, u + t, X CP(v
(), P)) - A(()*(v
+ t, U + t, 8, P)]
+ t, U + t, 8, p) dv
R. Therefore, from (Hs) and the lipschitz ian hypothesis on A, we
for all U obtain
E
(3.6)
ICP(t,
U
+ t, (), P) ~ K2retxu
CP(t,
U
fo I
()*(v
+ t, 8, p)1
+ t,u + t, (), P) -
()*(v
+ t, U + t, 8, p)1 du
It
for
U ~
O.
If we let P =
(3.7)
ICP(t,
U
P and use (3.5a) and (3.6), then
+ t, (), P) -
CP(t, u
K2r
+ t, 8, P)i ~ L(P) etxU[e-L(P)u -1]1() -
81
for all u ~ O. If we let () = 8 and use (3.5b) and (3.6), then (3.8)
ICP(t , u + t " ()
+
P) - CP(t u t () '"
K2r
P)I s:- - etxu[e-L(P)U L2( P)
1
+ L(P)u] liP -
PII
for all u ~ O. From (3.3) and (Hs) we have IX(t, (), P, Q) - X(t,
8, P, Q)I
fo o +f ~
KetxUM(Q) I()*(u
+ t, t, (), P) -
()*(u
+ t, t, 8, PI du
-00
ICP(t, u
+ t, (), P) -
CP(t, u
+ t, 8, P)IIIQ II duo
- co
Using (3.5a) and (3.7), we obtain relation (3.4a). Similar estimates using (3.5b) and (3.8) give (3.4c). This completes the proof of the first part of the lemma. Now suppose that P, Q in (3.1) have vector period w in (). The uniqueness theorem implies ()*(" w, P) = ()*(t, T, " P) wand, thus CP(t, T, () + w, P) = CP(t, T, b, r). From formula (3.3), one iPbtains X(t, () + w, P, Q) = X(t, (), P, Q). If P, Q in (3.1) are T-periodic n{: t, then uniqueness of solutions yields ()*(t + T + T, T + T, " P) = ()*(t + 1', T, " P). This fact in turn implies CP(t + T, T, (), P) is T-periodic in T. Since
,'+
+
INTEGRAL MANIFOLDS OF EQUATIONS WITH A SMALL PARAMETER
+
+
243
+
ct>(T, t T,B, P)ct>(t T, T,B, P) = J, the function ct>(T, t T, B, P) is Tperiodic in T. Using (1.3), one has X(t, B, P, Q) is T-periodic in t. If P(t, B), Q(t, B) are independent of t, then the same argument yields X(t, B, P, Q)is periodic in t with an arbitrary period and, therefore, must be independent of t. The case wltere P, Q are almost periodic in t is handled as follows. If 8 is any real number, we define P tJ in Sk, Q" in Sn by P ,,(t, B) = P(t + 8, B), Q,,(t, B) = Q(t + 8, B). Using the same process as in the estimation of the lipschitz function of B*(t, T, ~, P) in P, one easily obtains
IB*(:u + t + 8, t + 8,
B, P) -B*(u
+ t, t, B, P)I ~
-'I L(P) liP" - PII
eL(P)lul
for all u, t, B. Now using the same type of argument that was used to obtain (3.8), one obtains
Ict>(t
+ 8, u + t + 8, B, P) -
ct>(t, u
+ t, B, 'P)I K2r
~ - - eClU[e-L(P)u
- L2(P)
-1
+ L(P)u] lIP" -
PII
for all u ~ O. Using these two relations in (3.3), we arrive at the following inequality
I X(t+ 8, B, P, Q) -X(t, B,
P,
Q)I
K K ~ -; IIQ" -QII + 0((0( _ L(P)) [M(Q)
Kr
+ -;- IIQII] liP" -
PII
for all t,o,O. This inequality and the same type of argument as used in the proof of Theorems 1.1 and 2.1 9f Chapter IV complete the proof of Lemma 3.2. The constant 0( in the statement of Lemma 3.2 is a measure of the rate of approach of the solutions of (3.1) to the integral manifold and the constant L(P) is a measure in some sense of the maximum rate at which solutions on the manifold can converge or diverge from one another. A natural question to ask is whether the condition 0( - L(P) is necessary to obtain lipschitz smoothness of the parametric representation of the manifold. To understand some of the difficulties that might be encountered if 0( L(P) < 0, let us discuss the following example in' the two-dimensional (u, v)-space. If u = r cos B, v = r sin B, the system is cos B (j = k sin B - - . - (r -1)2, rsm B ,
r=
r(1 -r).
+
The circle r = 1 is an integral manifold of this system. If r = 1 p, then the linear variational equation for the manifold r = 1 is P= -p and the constant
244
ORDINARY DIFFERENTIAL EQUATIONS
or; in hypothesis H5) is +1. On the other hand, on the manifold r = 1, the Liptschitz constant L(P) can be taken to be k. Also, on r = 1, there are the stable node (-1,0) and the saddle point (1, 0). The trajectories of this system near r = 1 for k < 1 and >1 are shown in Fig. 3.1. For k < 1 all
"
Figure VII.3.1
orbits enter the node (-1,0) tangent to the circle r = 1 and for k> 1 all orbits enter the node (-1,0) perpendicular to the circler = 1. A small perturbation in the equation for k > 1 could possibly lead to an invariant curve which has a cusp at (-1, 0). The following example shows that a cusp can. arise. Consider llIte system ~.,
8=ksin8, X= -x+f(8),
.'
INTEGRAL MANIFOLDS OF EQUATIONS WITH A SMALL PARAMETER
where f has a continuous first derivative and f(B equation has the solution
B*(t,
7, ' )
+ 27T) =f(B).
245
The first
= 2 arc tan(ek(t-T) tan ~),
and therefore the integral manifold is given from (3.3) by
X(t, B, k)
== X(B, k) = (00 eUf [2 arc tan( eku tan~) ] duo
For any closed interval in the interior of (-7T, 7T),
_o_X_(-=B,_k_) = fO e(1-k)u df [sin 2 ~ + e-2ku cos 2 ~] -1 du oB _ 00 dB 2 2 exists and is continuous for any k. If k > 1, this integral becomes unbounded as B-+ 7T or -7T if dfldB is different from zero in a neighborhood of these points and X(B, k) is not lipschitzian.
vn.4.
The Mapping Principle
To apply Lemma 3.2 to prove Theorem 2.1, we define a mapping whose fixed points coincide with the integral manifold of (1) and prove this mapping is a contraction. It is convenient to formulate this mapping in more general terms and specialize it to system (1) later~ For given constants 6, D, let (4.1)
Sn(6, D) = {fin Sn: Ilf II ~ D, If(t, B) - f(t, 0)/ ~ 61B - 01 for all (t, B, 0) in R X Rk X Rk}.
Let P: R X Rk X sn(6, D) -+ Rk, Q: R X Rk X Sn(6, D) -+ Rn and for any f in Sn{L~., D) suppose that P(·, ., f) is in Sk and Q( ., ., f) is in Sn. For any fin Sn(6, D), we,know from the discussion of system (3.1) that the system
(4.2)
() =
P(t, B, f),
x=
A(B)x + Q(t, B,f),
has an integral manifold defined parametrically by (3.3). We rewrite (3.3) again to emphasize the dependence uponf as
(4.3)
X*(t, B, f) =
f-00o
246
ORDINARY DIFFERENTIAL EQUATIONS
where w.e are using the simplified but hopefully not confusing notation BI (t,
T,
"j) = B*(t,
T,
',f)
=
T, "
P(', " f)), P(', " j)),
T, "
X*(t, B,f) = X(t, B, P(', " j), Q(', " f)).
Ifthere exists anf in Sn(A, D) such that f(t, B) = X*(t, B, j), thenf will .define an integral manifold of (4.2). It is clear that such a procedure will be applicable to system (1) without the y equation simply by defining P(t, B, j) = w(t, B, e)
(4.4)
+ G(t, B,f(t, B), e),
Q(t, B,j) = F(t, B,f(t, B), e).
We now derive some conditions on P, Q which will ensure that the map X*( " " j),f in Sn(A, D) is a contraction.
Assume the following: there are constants {3(D), M(A, D), L(A, D) and a(D), b(D) such that
IIQ(', " j) II ~ {3(D), IIQ(;, " j) - Q(', .,J) II ~ a(D) IIf - j II, IIP(" " j) - P(', .,J) II ~ b(D) Ilf - j II,
(4.5)
IQ(t, B, j) - Q(t, B,f)1 ~ M(A, D)I B - BI, IP(t, B,j) - P(t, B,j)1 ~ L(A, D)IB - 01,
for all (t, B, 0) in R X Rk X Rk andf,.fin Sn(!~., D). If we now use (3.4) with M(Q), L(P) replaced by M(A, D), L(A, D), respectively,
(IIX*" . ,j) II
~
K
-
{3(D),
IX
IX*(t '
B f) - X*(t 0 f)1 ::; ,
IIX*(', " j) -
KM (A D) 1, IB - 01 L(A, D) ,
" - IX -
X*(·,
K [MI(A, D)b(D)
.,J) II ~ -;
M I (A, D)
IX _
L(A, D)
Kr
= M(A, D} + -
+ a(D) ] Ilf - J II,
{3(D).
IX
for all (t, B, 0) in R X Rk X Rk and f,J in sn(A; D) provided that IX L(A, D) > O. From these relations we can state the following;~ LEMMA 4.1. Suppose Q, P satisfy (4.5) and for any fin Sn(A, D), the unique integral manifold X*(·, " j) of (4.2) is defined by (4.3). The mapping
INTEGRAL MANIFOLDS OF EQUATIONS WITH A SMALL PARAMETER
247
T: S1I(tl., D) _S1I(tl., D) defined by Tf = X*(', " f) is a contraction provided that tl., D satisfy the relations
(4.7)
(c)
oe - L(tl., D) > 0, KfJ(D) < oeD, KM1(Ll, D) < [oe - L(tl., D)]tl.,
(d)
K[M1(tl., D)b(D) -; oe _ L(tl., D)
(a) (b)
+ a(D)
]
1
< 2'
If conditions (4.7) are satisfied, then X*(·, " f) has a unique fixed point in S1I(tl., D).
VII.5. Proof of Theorem 2.1
We are now in a position to prove Theorem 2.1 for the case when y is absent. The only thing to do is to find the constants in (4.5) with P, Q defined by (4.4) and substitute these in (4.7). From hypotheses (H 2 )-(H4), one obtains, for 0 < e ~ eo, (5.1)
(a) (b) (c) (d) (e)
fJ(D) = S(D, e)D + N(e), a(D) = S(D, e), b(D) = p.(D, e), M(tl., D) = y(D, e) S(D, e)tl., L(tl., D) = L(e) TJ(D, e) p.(D, e)tl..
+
+
+
If (5.1) is used in (4.7), one obtains relation (2.3) in hypotheses (H6). This completes the proof of Theorem 2.1 when the vector y is absent. The case with y present follows along the same lines if use is made of the fact that the equation
0= P(t, 8), = A(8)x + Q(t, 8), iJ = B(8)y + R(t, 8), i
has a unique integral manifold given by X(t, 8, P, Q), Y(t, 8, P, R) with X given by (3.3) and Y(t, 8, P, R)
where 'Y(t,
'T,
= (') 'Y(i, u + t, 8,
P)R(u+t, 8*(u
+ t, t, 8, P)) du
"P) is the principal matrix solution of the linear system
iJ =
B(8*(t,
'T, "
P))y.
248
ORDINARY DIFFERENTIAL EQUATIONS
Vll.6. Stability of the Perturbed Manifold For the unperturbed part of system (1), namely, (j = w(t, 8, e),
(6.1)
i
= A(8, e)x,
if = B(8, elY, there is a unique .integral manifold with the x and y coordinates bounded and this is given by x = y = o. This manifold has a saddle point structure in the sense that any solution of (6.1) is such that x ---+ 0 exponentially as t ~ 00 and y~O exponentially as t~ -00. One can prove that the saddle point structure is also preserved for the perturbed manifold whose existence is assured by Theorem 2.1. We do not prove this fact in detail,· but only indicate the proof for the special case
0;
B = 1 + 8(t,O,p,e) (6.2)
p=Ap + R(t,O,p,e)
where e(t,O,p,e), R(t,O,p,e) are continuous, bounded and continuously differentiable in 0 ,p in R X R X Q(p,eo), periodic in 0 of period w, e(t,O,O,O) = 0, R(t,8,O,O) = 0, oR(t,8,O,O)/op = 0, and the eigenvalues of A have negative real parts. Corollary 2.1 implies there exists an integral manifold Se = {(t,B,x): x = [(t,O,e), (t,O) E R2}, ~ Ie! ~ eo, [(t,O,e) is periodic in 0 of period w and [(t,O,O) = O. Our first objective is to show there is a neighborhood V C R2 X R n of {(t,O,p):p = O} such that if (to,Oo,po) E V, then the solution (O(t),p(t» of (6.2) through (to,Oo,Po) satisfies
°
!p(t) - [(t,O(t),e)!
~
°
as t ~ 00.
This shows that solutions go back to the integral manifold as t ~ 00. We will also show that this approach is at an exponential rate. To do this, we consider all solutions of the equations (6.2). The variation of constants formula gives
,pet) = eA(t-to)po
+ (eA(t-S)R(s,O(s),p(S),e)dS
(6.3)
B(t) = 1 + e(t,O(t),p(t),e),
OCto) = 0o.
We first show there is a function w(t,O,to,po,e), w-periodic in 0, such that the solution of (6.3) can be represented as
INTEGRAL MANIFOLDS OF EQUATIONS WITH A SMALL PARAMETER
249
pet) = \II(t,to,O(t),po,e).
(6.4)
As in the proof of the existence of the integral manifold Se' we transform this to a problem of fmding a fIxed point of a certain mapping. Let
8 1 = {\II:{(t, to) : t'? to} X R X R n ~ R n , continuous, bounded, together with fIrst derivatives in 0 ,p}. For \II in 81 , let 1\Il1
= sup{1 \II(t,to,O,p) I + 13\I1(t,to,0,p)/30 I + 13\I1(t,to,0,p)/3pl :t'? to, (O,p) ER
X Rn}.
The smoothness assumptions here are only to make the notation simpler. For \II in 8 1 and t ~ T, let
(T\II)(r,t,O,p)
= eA(T-t)p +
r
eA (T-S)R(s, 0*(s) , \II(s,t,O*(s),p),e)
e*(s) = 1 + 8(s, o(s), \II(s,t,O*(s),p),€),
8*(t) =
o.
One can now proceed as before to show that T is a contraction on an appropriate subset of $1 to obtain a fIxed point of T (or one can use the implicit function theorem). If \II = T\II, we have shown that (6.4) is satisfIed. There is a continuous function M(a,eo), a'? 0, lei ~ €o,M(O,O) = 0, constants k>O, a>O such that, for Ipol
13\I1(t,to,8,po,e)/38 I ~M(a,eo)
I3\I1(t, to,O ,po,e)/3p [ ~Ke-cx(t-to )/2, leAtl ~Ke-cxt,
t'? to
t'? 0.
For Po suffIciently small, we know that the solution of (6.2) through (to,Oo,po) satisfIes (6.4) for t '? to. Thus, our integral manifold Se must also have this property; that is, for any t'? to, 0 E R, there is a 00 = 8 0(t,0) such that the solution (8'(s),p'(s» of (6.2) through (to,80,[(to,00,e» satisfIes O'(t) = 8, p'(t) = [(t,O,e). To obtain this assertion, we have simply integrated backwards from" t to to starting from the point (t,8,[(t, 8 ,e». Thus, if Po =[(to,Oo,e), then
[(t,O,e) = \II(t,to,8,p',e) for all t,O and
1[(t,O ,e) - \II(t,to,O ,po,e) I ~Ke-cx(t-to)/2Ipo - Po I. If (O(t),p(t» is any solution of (6.2) with 8(tO) = 0o, p(tO) = po, Po suffIciently small, we have
250
ORDINARY DIFFERENTIAL EQUATIONS
1!(t,8(t),e) - 'Ir(t,to,8(t),po,e) I ~Ke-a(t-to )/21!(to,80(t,8(t),e) - Pol t~to·
This proves the asserted stability of the integral manifold St;.
VII. 7. Applications
In this section, we give some applications of the theory of this chapter. There is such a large variety of applications that it is impossible to do them justice without devoting a treatise to the subject. However, it is hoped that the few mentioned below together with the ones delegated to the exercises in Section 8 may indicate the possible scope of the theory and stimulate further reading in the literature. Suppose the equation
x =j(x)
(7.1)
where j has continuous first derivatives in Rn, has a nonconstant w- periodic solution u whose linear variational equation
.
.(7.2)
oj (u(t))
y=--y
ox
has n - 1 characteristic multipliers not on the unit circle. Suppose also that g(t, x) is continuous for (t, x) in R X Rn, has continuous first derivatives with respect to x and is bounded for t in R and x in any compact set. THEOREM 7.1 Under the above hypotheses, there are a neighborhood U of the periodic orbit 't&' = {x : x = u( 8), 0 ~ 8 ~ w} and an e1 > 0 such that the system
x =j(x) + eg(t, x)
(7.3)
has an integral manifold Se in R X U, 0 ~ le1 ~ e1, So = R X 't&' (a cylinder), Se is asymptotically stable if n":"'l multipliers of (7.2) are inside the unit circle and unstable if one is outside the unit circle. The set Se has a parametric representation given by Se = {(t, x): x = u(8)
+ v(t,
+
8, e), (t, 8) in R X R}, i~'
where v(t, 8,0) = 0, v(t, 8, e) =v(t, 8 w, e) and is almost p~riodic (Tperiodic) in t if g(t, 8) is almost periodic (T-periodic) in t. PROOF. This is a simple consequence of the previous results. In fact, using the coordinate transformation in Chapter VI, one obtains a special
INTEGRAL MANIFOLDS OF EQUATIONS
WITH
A SMALL PARAMETER
251
case of system (1). The above hypotheses permit a direct application of Theorem 2.2 to obtain existence of the manifold for the interval 0 < e ~ el. Replacing e by -e O1').e obtains the result on -el ~ -e < O. Defining the manifold of (1) to be zero at e = 0 completes the proof of existence. Stability is a consequence of the remarks in Section 6. In case g in (7.3) is periodic in t of period T, then the integral manifold S,. given in Theorem 7.1 has a parametric representation which is periodic in t of period T. The cross section of S,. at t = 0 and t = T are therefore the same and the fact that it is an integral manifold implies this cross section is mapped into itself through the solutions of the differential equation: Since the differential equation is T-periodic, following a solution x(t, xo), x(O, xo) = Xo, from t = 0 to t = 2T is the same as following the solution from t = 0 to t = T and then following the solution x(t, x(T, xo)) from t = 0 to t = T. The solutions of (7.3) on S,. in this case can therefore be considered as a differential equation without critical points on the torus obtained by taking the section of the surface S,. from t = 0 to t = T and identifying the ends of this section. The theory of Chapter II then gives the possible behavior of the solutions on S,. . THEOREM 7.2. Supposef(x) satisfies the conditions of Theorem 7.1 and let .g(t, x) be continuous and uniformly bounded together with its first partial derivatives with respect to x in a neighborhood of~. If g(t, x) is almost periodic in t uniformly with respect to x and M[g(·, x)] = lim -1 T-+oo
T
IT g(t, x) dt = 0, 0
then there are a neighborhood U of ~ and an Wo > 0 such that the (7.4)
sys~em
i=f(x)+g(wt, x)
has an integral manifold S{J) in R X U, w ~ WO, S{J) - R X ~ as w - 00, S{J) is asymptotically stable if n -1 multipliers of (7.2) are inside the unit circle and unstable if one is outside the unit circle. The set S{J) has a parametric representation given by
+ v(wt, 8, w-1 ), (t, 8) in R X R}, v(t, 8 + w, ex) and is almost periodic in t.
SQ) = {(t, x): x = u(8)
where v(t, 8,0) = 0, v(t, 8, ex) =
PROOF. From Lemma 5 of the Appendix, for any TJ > 0, there are a function w(t, x, TJ) with as many derivatives in x as desired and a function a(TJ) _0 as TJ _0 such that 18w(t, x, TJ)J8t -g(t, x)1
< a(TJ)
for all t E R and x in a neighborhood of <'C. Also TJw, TJOwJ8x-0 as TJ _0. Therefore, as in the proof of Lemma V.3.2, there is an wo > 0 such that the transformation
252
ORDINARY DIFFERENTIAL EQUATIONS
x=y+~w( wt, y,~) is a homeomorphism in a neighborhood of~. 'If this transformation is applied to (7.4) one obtains a system
iJ =f(y) + G(wt, y, w-1 ), where G(7", y, w-1 ) is continuous in 7", y, w together with its first derivative with respect to y, is uniformly bounded for 7" in R, y in a neighborhood of rc and w ~ wo, and G(7", y, 0) = O. If one uses 'the coordinate transformation in Chapter VI and lets wt = 7", w- 1 = e, then the new system is a special case of system (1) for which Theorem 2.3 applies directly. This will complete the proof of existence of an integral ma~ifold. The stability follows trom Section 6. Another application of the previo~s results concerns a generalization of the method of averaging. Consider the system (7.5)
rfr = e + e'¥(""
p=
p),
eR("" p),
where", is in Rk, p is in RP, e = (1, ... , 1), and '¥("" p), R("" p) are multiply periodic in '" of period w and have continuous first derivatives with respect to p. Let (7.6)
f '¥(", + et, T 1
T
'¥0("', p) = lim T .... '"
. 1 .T R o("', p) = lim R(", T .... '"
p) dt,
0
T
J
+ et, p) dt.
0
From the Appendix, for any 7J > 0, there are functions u("', p, 7J}, v("" p, 7J) with as many derivatives in "', p as desired and a function a(7J) -+0 as 7J -+0 such that
I:; e - '¥("" p) + '¥0("', p) I< a(7J}, I:; e - R("" p) + R o("', p) I< a(7J), and the functions 7Ju, 7Jv, 7J ouJ 0"', 7JovJo"" 7JouJoP, 7Jovjop -+0 as 7J-+0 uniformly for'" in Rk, p in a bounded set. Therefore, as in the proof of Theorem V.3.2, there is an eo > 0 such that the transformation i~ (7.7)
c/J + eu(c/J, r,e), p = c/J + ev(c/J, r, e),
'" =
INTEGRAL MANIFOLDS OF EQUATIONS WI';l'H A SMALL PARAMETER
253
is a homeomorphism for lei < eO, ifJ, cp in Rk and p, r is a bounded set. If the transformation (7.7) is applied to (7.5), then a few simple computations yield
1> = e + e'Yo(cp, r) + e'Yl(cp, r, e),
(7.8)
f = eRo(cp, r)
+ eR1(cp, r, e),
where 'Y 1, R1 have the same smoothness properties as 'Y, R, but now satisfy 'Y 1(cp, r, 0) = 0, R1(cp, r, 0) = 0. In other words, by a transformation (7.7) which is essentially the identity transformation near e = 0, the system (7.5) is t,ransformed into an equation which is a higher order,perturbation of the averaged equations (j = e
(7.9)
+ e'Yo(O, p),
p = eRo(O, pl· One is now in a position to state results concerning the existence of integral manifolds by asserting the averaged equations have certain properties. For example, suppose there is a po such t'hat Ro(O, po) = and furthermore,
°
(7.10)
Ro(O, po + z)
=
C(O)z + H(O, z),
[A~O) ~OJ, 'Yo(O, po + z) = 0(0, z),
C(O)
=
H(O, z)
=
[;::::n,
where all matrices are partitioned so that any matrix operations wIll be compatible. As an immediate consequence of Theorem 2.3 and the form of transformed equations (7.8), one can state . 7.3 .. Suppose A, B in (7.10) satisfy hypothesis (H5) in Section 1 with 0: = e0:1, 0:1 > 0, a constant. If the lipschitz constant L of 0(0. 0) satisfies 0:1 - L > 0, then there.are e1 > 0, continuous functions D(e), ~(e), 0< e ~ e1, approaching zerofds e ~o and a function f(ifJ, e) in Rp which is continuous in Rk X [0, ed, THEOREM
If(ifJ, e) - pol < D(e), If(ifJ, e) - f(ifj, e)1 < ~(e)lifJ - ifjl, such that f(ifJ, e) is multiply periodic in ifJ of vector period wand the set Se = {(ifJ, p) : p = po
+ f(ifJ, e), ifJ in Rk}
is an integral manifold of (7.5). If the y component of z is present in (7.10), then the manifold is unstable and if it is absent, the manifold is stable.
254
ORDINARY DIFFERENTIAL EQUATIONS
A simple corollary of Theorem (7.3) which is useful for the applications is COROLLARY 7.l. Suppose the averaged equations (7.9) are independent of .p; that is, the averaged equations are
(7.11)
+ e'Yo(p) ,
8=
e
p=
eRo(p).
Also, suppose there is a po such that Ro(po)':'- 0 and the real parts of the eigenvalues of the matrix 0 = fJRo(po)/fJp are nonzero. Then the conclusions of Theorem 7.3 remain valid with the manifold being stable if all eigenvalues of 0 have negative real parts and unstable if one eigenvalue has a positive real part. VII.S. Exercises Some of the exercises listed below are rather difficult and a complete discussion of some could lead to interesting new results in the theory of oscillations. EXERCISE S.I. Use Section VI.3 to show that the hypotheses of Theorem 7.1 imposed on (7.1) are satisfied for the van der Pol equation
(S.I)
Xl =X2,
X2 =
-Xl
+k( 1 - Xi)x2 '
k>O.
Use Theorems 7.1 and 7.2 to discuss the existence and stability of integral manifolds of the system (S.2)
Xl =X2, X2 =
-Xl
+ k(1 -
xi)X2
+ e sin wt,
when either e is small or w is large. Use the theory of Chapter IV to discuss the existence and stability properties of a periodic solution of (S.2) of period 27T/W for either Iel small or w large. In Chapter II it was shown that every solution of (S.I) except the zero solution approaches the periodic orbit CC as t ~ 00. Can you use the previous discussion to give the qualitative behavior of. all solutions of (S.2) in some large domain of R2 for· Iel small? Hint: Show there is an open set U in R2 such that the tangent vector to the solution curves of (S.I) on the boundary of U point toward the interior of U. This implies that the tangent v~ptor to the solutions of (S.I) in (t, x)-space is pointing towara· the interi6lof the "cylinder" Rx fJU. Therefore for lei small or w large, the same will be true of the solutions· of (8.2). Now use continuity with respect to the vector field.
INTEGRAL MANIFOLDS OF EQUATIONS WITH A SMALL PARAMETER
255
EXERCISE 8.2. Suppose both of the vector systems x = f (x), 'Ii = F(y) satisfy the condition of Th~orem 7.1. What types of integral manifolds does the system
x
= f (x) + eg(t, x, y),
'Ii =
F(y)
+ eG(t, x, y)
possess for e small if g, G are bounded for t in R and x, y in compact sets? What are the stability properties of each of the manifolds? Generalize this result. Hint: Use the coordinate system of Chapter VI for each periodic orbit. EXERCISE 8.3. For w large, discuss the existence and nature of integral manifolds of the system
Xl =X2,
=
X2
-Xl
+k[1 -
(Xl
+ B sin wt)2]X2'
as a functjon of B. EXERCISE 8.4. For w large, discuss the question of the existence and ,. the dependence of integral manifolds of the system
+ 2 x + x + Kf (x + B sin wt) = 0
53
on th3 constants K and B for odd functions f (x) which are monotone nondecreasing and approach a limit as x --+ 00. This problem is difficult and one could never solve it in this general context. Take special functions f. EXERCISE
8.5.
Discuss the existence of integral manifolds of the
equation Xl =x2,
x2 =
-xl
+ e(1 -
XDX2
+ A sin wt,
for e small and various values of A and w. Let
= p sin (h + A(1 - w 2 )-1 sin wt X2 = P cos {h + Aw(1 - w 2 )-1 cos wt Xl
and 82 = t to obtain a system of differential equations for 81. 82 , P and then apply Corollary 7.1. What are the stability properties of the manifolds? What happens geometrically as A, w vary? EXERCISE 8.6. For e small, discuss the existence and stability properties of integral manifolds of the system
x - e(1 -x2 - ay 2)X + x fj - e(1 -
y2 -
cxx 2 )'Ii
=
0,
+ a 2y =
0,
256
ORDINARY DIFFERENTIAL EQUATIONS
as a function of a, ex, a. Let x = PI cos (h, x = - PI sin (h, y = P2 cos a02, y = -ap2 sin a02 to obtain a system of the same form as system (7.8). Now apply the method Of averaging described above and, in particular, Corollary 7.1 for the case when k + la =1= 0 for all integers k, l for which Ikl +Ill ;;;; 3. What are the periodic solutions? Can you describe geometrically what happens when the stability properties of the periodic orbits change under variation of the constants a and ex? What happens when k la = 0 for some integers k and l with Ikl + Ill;;;; 3? In the equation, change 1 - x2 - ay2 to 1 - x 2 - ay2 bx2y2 and discuss what happens as a function of a, b, ex.
+
+
EXERCISE 8.7. equations
Carry out the same analysis as in Exercise 8.6 for the
x-e(x- ~) +ax+ILY=O, ii -
e(y _~3) - /LX + vy = 0,
for those values of ~he parameters a, IL, v for which the characteristic roots of the equation for e = 0 are simple and purely imaginary, say ±iWl, ±iW2. Under this hypothesis, this system can be transformed by a linear transformation to a system which for e = 0 is given by ii + w~u = 0, ii + w~v = o. Now apply the same type of argument as In Exercise 8.6.
VU.9. Remarks and Suggestions for Further Study Detailed references to the method of Krylov-Bogoliubov may be found in the books of Bogoliubov and Mitropolski [1] or Hale [3]. The original results on integral manifolds using this method considered only the case in which the flow on the unperturbedrllanifold was a parallel flow; that is, w(t, 0, e) in (1) is a conStant. However, the method of proof given in the text is basically the same as the original proof for parallel flow except for the technical details. Other results along this line were obtained by Diliberto [31and Kyner [1]. Kurzweil [1, 2] has given another method for obtaining the existence of integral manifolds and has the problem formulated in such a general framework as to have applications to partial differential equations, difference equations and some types of functional differential equations. Further results may be found in Pliss [1]. It is not assumed in system (i) that the functions are periodic in, ~e vector (}. The results have implications to the theory of center manifolds atid stability theory (Pliss [2], Kelley [1]) and the theory of bifurcation (Chafee [1], Lykova [1]).
INTEGRAL MANIFOLDS OF EQUATIONS WITH A SMALL PARAMETER
257
Hypothesis (H5) in Section 2 is too strong. For example, if w(t,8,e) = 1, one need only assume the exponential estimates are valid for functions 8(t) = t + 80 (see Montandon [1]). For a more general case, see Coppel and Palmer [1] , Henry [1]. The basic result in the proof of the theorems on integral manifolds ~as Lemma 3.2. The proof of this lemma shows that the Lipschitz constant L(P) was used only to obtain estimates on the dependence of the solution O*(t, T,O, P) of = P(t, 0) upon O. There are many waysto obtain estimates of this dependence without using the lipschitz constant of P. For example, if P(t, 0) has continuous partial derivatives with respect to 0, then oO*(t, T, 0, P)/oO is a solution of the equation
a
Consequently, one can use the eigenvalues of the symmetric part of oP/oO to estimate the rate of growth of {. The hypothesis (H 5) can also be verified by using the eigenvalues of the symmetric part of the matrices A(O), B(O). Using the method of partial differential equations mentioned in Section 1, Sacker [1] has exploited these concepts to great advantage to discuss the e!i~tence and smoothness properties of integral manifolds for equations (1) wmch are mae pendent of t and periodic in O. Diliberto [3] has also used these same concepts and the method of the text to find integral manifolds. The first example in Section 3 after Lemma 3.2 is due to McCarthy [1] and the second to Kyner [1). The generalized average defined in relation (16) was first introduced by Diliberto [1]. Some of the exercises in Section 8 can be found in Hale [7].
CHAPTER VIII Periodic Systems with a Small Parameter
In Chapter IV, we discussed the existence of periodic solutions of equa.tions containing a small parameter in noncritical cases; that is systems (1)
where f(t equation (2)
x=Ax+ef(t,x),
+ T, x) =f(t, x) and no solution except x = 0 of the unperturbed x=Ax,
is T-periodic. In Chapter V, the method of averaging was applied to some systems for which (2) has nontrivial T-periodic solutions. The basis of this method is to make a change of variables which transforms the system into One which can be considered as a perturbation of the averaged equations. If the averaged equations are noncritical with respect to the class of T-periodic functions, then the results of Chapter IV can. be applied. If the averaged equations are critical with respect to T -periodic functions, then the process can be repeated. In addition to being a very cumbersome procedure, it is very difficult to use averaging and reflect any qualitative information contained in the differential equation itself into the iterative scheme. For example, if system (1) has a first integral, what is implied for the iterations? For periodic systems, other more efficient procedures are available. The present chapter is devoted to giving a general method for determining periodic solutions of equations including -(I) which may be critical with respect to T-periodic functions-:-This method gives necessary and sufficient conditions for the existence of a T-periodic solution of (1) for e small. These conditions consist of transcendental equations (the bifurcation or determining equations) for the determination of a T-periodic function which is a solution of (2). The bifurcation equations are given in such a way as to permit a qualitative discussion of their dependence upon properties of the right hand side oftd). This is illustrated very well when! in (1) enjoys some even and oddness properties or system (1) possesses a first integral. In general, such systems have families of T-periodic solutions.
258
PERIODIC SYSTEMS WITH A SMALL PARAMETER
259
In addition to the advantage mentioned in the previous paragraph,the method of this chapter can be generalized to arbitrary nonlinear systems. This topic as well as a more general formulation of the method in Banach spaces will be treated in the next chapter. For A = 0, the. basic ideas are very elementary and easy to understand geometrically. For this reason, this case is treated in detail in Section 1. Also, in Section 1, it is shown how to reduce the study of periodic solutions of many equations as well as the determination of characteristic exponents of linear periodic systems to this simple form. Section 2 is devoted to a discussion of the general system (1) as well as results on systems possessing either symmetry properties or first integrals. In Section 3, we reprove a result of Chapter VI using the method of this chapter rather than a coordinate system around a periodic orbit.
VllI.l. A Special System of Equations Suppose I: R X en ~en is a continuous function with ol(t, xl/ox also continuous, 1(t T, x) = 1(t, x), e is a parameter and consider the system of equations
+
i=el(t,x).
(1.1)
Our problem is to determine whether or not system (1.1) has any T -periodic sblutions for e small. For e = 0 all solutions of (1.1) are T-periodic; namely, they are constant functions. The basic question is the following: if there are T-periodic solutions of (1.1) which are continuous in e, which solutions of the degenerate equation do they approach as e ~O? One precedure was indicated for attacking this question in Chapter V. Another method is due to Poincare in which a periodic power series expansion is assumed for the solution as well as the -initial data. The initial data is then used to eliminate the secular terms that naturally arise in the determination of the coefficients in the power series of the solutions. In this section, we indicate another method for solving this problem which seems to hav.e some qualitative advantages over the method of Poincare. Let gl>T = {g: R ~en, gcontinuous, g(t + T) = g(t)}, Ilgl! = supo;:;;t;:;; T Ig(t)l. For any g in f!JT, define Pg to be the mean value of g; that is, (1.2)
Pg = -1
T
IT g(t) dt, 0
If g is T-periodic, then the system
(1.3)
i
= g(t)
IIPgII ~ Ilgll·
260
ORDINARY DIFFERENTIAL EQUATIONS
has a T-periodic solution if and only if Pg = O. Furthermore, if Pg = 0, let :Kg be the unique T-periodic solution of (1.3) which has mean value zero; that
is, (1.4)
:Kg = (1 - P)
r o
g(s) ds,
lI:KglI ~ K IIg II,
K=2T.
Every T-periodic solution of (1.3) can then be written as x=a+:Kg
,where a is a constant n-vector and a = Px. These simple remarks imply the following: Suppose P and :K are defined in (1.2), (1.4): Then (i) x(t) is 'a T-periodic solution of (1.1) only if Pf(·, x(·» , 0; that is, only iff(·, x(·» = (1 - P)f(·, x(·». (ii) System (1.1) has a T-periodic solution x if and only if x satisfies the system of equations LEMMA 1.1.
(a)
(1.5)
(b)
= a +s:K(1 ePf(·, ~rO,
x
P)/(·, xl;
where a is a constant n-vector given by a = Px. PROOF. If x i!iJ a T-periodic solution of (1.1), let g(t) =f(t, x(t» and assertion (i) follows immediately. The fact that x satisfies (1.5) is just as obvious. If x is a solution of (1.5), then f(·, x) = (1 - P)f(·, x) and, therefore, x is a T-periodic solution of (1.1). This proves the lemma.
LEMMA 1.2.
For any
IX
> 0, there is an eo > 0 such that for any a in 0"
lal ~ IX, lei ~eo, there is a unique function x* =
With
x*(a, e) which satisfies
(1.5&). Furthermore, x*(a, e) has a continuous first derivative with respect to a, e and x*(a, 0) = a. If there is an a = a(e) with la(e)1 ~ IX for 0 ~ lei ~ eo and
G(a, e) def Pf(·, x*(a, e» = 0,
(1.6)
then x*(a, e) ,is a T-periodic solution of (1.1). Conversely, if (1.1) ~ a T-periodic solution x(e) which is continuous in e and has Px(e) = a(e), la(e)1 ~ IX, 0 ~ lei ~ eo, then x(e) = x*(a(e), e) where x*(a, e) is the function given above and a(e) satisfies (1.6) for 0 < lei ~ eo. SuppOse IX > 0 is given, fJ is any number >0 and a is an n-vector Define f/'(fJ) = {y in 9p: lIyll ~ fJ, Py = O} and the
with
(1.7)
lal ~ IX.
,y=e:K(1-P)f(·,y+a)
PERIODIC SYSTEMS
WITH A
SMALL PARAMETER
261
+
for y in Y(f3). If Y is a fixed point of ff, then y a satisfies (1.5a). Let M, N be bounds on If(t, x)l, 18f(t, x)/8xl, respectively, for t in R, Ixl ~ f3 IX and choose el so that 4el TM ex,such that Ix(e) - a(e)1 ~ f3 for 0 ~ lei ~ el. For this f3 choose eo ~ el so that ff is a contraction on Y(f3) for 0 ~ lei ~ eo· Then x(e) - a(e) = y*(a(e), e) and x(e) = x*(a(e), e). The conclusion of the lemma follows from Lemma 1.1. Lemma 1.2 asserts the following: one can specify an arbitrary n-vector a and then determine uniquely a T-periodic function x*(a, e) with mean value a in such a way that all of the Fourier coefficients of the function i*(t) - f(t, x*(t)) are zero exceptfor the constant term (this is the same as saying (1.5a) is satisfied). The n-vector a is then used to try to make the constant term equal to zero (that is, satisfy (1.6)). Equations (1.6) are sometimes referred to as the determining equations or bifurcation equations of (1.1). As a consequence of the above proof, the function x*(a, e) can be obtained as the limit of the sequence {x(k)}, x(k) = y(k) a where the y(k) are defined successively by y(k+1) = ffy(k), k = 0, I, 2, ... , y(O) = 0, and ff is defined in (1.7). Using only the first approximation, one arrives at
+
+
THEOREM 1.1. Suppose x*(a, e) is defined as in Lemma 1.2 and let G(a, e) be defined by (1.6). If there is ann-vector ao such that (1.8)
det [
G(ao, 0) = 0,
8G(a o, 8a
0)] =I- 0,
then there are an e1 > 0 and a T-periodic solution x*(e), x*(O) = ao, and x*(e) is continuously differentiable in e.
lei
~ el,
of (1.1),
PRoOF. The hypothesis (1.8) and the implicit function theorem imply there is an el, 0 < el ~ eo, such that equation (1.6) has a continuously differentiable solution a(e), la(e)1 < ex, 0 ~ lei, ~ e1. Lemma 1.2 implies the remaining assertions of the theorem. Notice that (1.9)
G(a, 0) = -I
T
IT f(t, a) dt, 0
and therefore can be calculated without knowing anything about the solutions of (1.1). Compare Theorem 1.1 with Theorem V.3.2. Can you prove Theorem V.3.2 by the above method?
262
ORDINARY DIFFERENTIAL EQUATIONS
In the applications, Theorem 1.1 is not sufficiently general even for equation (1.1). More specifically, it is sometimes necessary to take the terms in the Taylor expansion of G(a, e) in (1.6) of the first, second, or even higher order in e. Except for the numerical computations involved, there is con. ceptually no difficulty in obtaining these terms. To obtain the first order terms of G(a, e) in e, one must compute the first. order terms in e in x*(a, e), etc. This is accomplished by the iteration procedure x(k) = y(k) + a, y(k+l) = ff'y(k), k = 0,1,2, ... , y(O) = a, where the mapping ff' is defined in (l.7). It is easy to generalize the above results to a system of the form (l.lO)
z=
Bz
+ eh(t, z),
where h(t + T, z) = h(t, z), hand 8h/8z are continuous in R X Gm+n, B = diag(On, BI), On is the n X n zero matrix, BI is an m X m constant matrix such that eB1T -1 is nonsingular. This latter condition states that the system if = Bly is noncritical with respect to f!JT. If z = (x, y), h = (f, g) where x, fare n-vectors, then the above system is equivalent to (1.10)'
i = ef(t, x, y),
if = BlY + eg(t, x',
y).
For any h = (f, g) in r!JT define Ph to be the (n vector given by (l.11)
+ m)-dimensional constant
[~fT f(t) dt]
Ph= T o .
o
The nonhomogeneous linear system ( l.12)
Z= Bz+ h(t)
has a T-periodic solution if and only if Ph = O. Furthermore, if Ph = 0, let :f{h be the unique solution z of (1.12) with pz = O. For any h in fJiT, one can therefore define :f{(1 - P)h and there is a constant K > 0 such that (1.13)
11:f{(1 - P)hll
~
Kllhll;
that is :f{(1 - P) is a continuous linear mapping of f!JT into f!JT. Every T-periodic solution of (1.12) can be written as z = a*
+ e:f{h, a*= [~J,
where a is a constant n-vector with a* = Pz.
~
l.3. Suppose P is defmed by (1.11), H is in ~T and :f{(J JiP)H is the unique T-periodic solution z of LEMMA
z=
Bz
+ (1 -
P)H
PERIODIC SYSTEMS WITH A SMALL PARAMETER
263
with pz = O. Then z(t) is a T-periodic solution of (1.10) only if Ph(·, z(·)) = 0; that is, only if h(·, z(·)) = (1 - P)h(·, z(· )). Also, system (1.10) has a T-pcriodic solution z if and only if the system of equations (1.14)
+ e.Yr(1 -
(a)
z = a*
(b)
ePh(·,z(·))=O,
P)h(·, z(· )),
= (a, 0),
a*
= Pz.
is satisfied, where a is a constant n-vector given by a*
LEMMA 1.4. For any IX > 0, there is an eo > 0 such that for any a in Rn with lal ~ IX, lei ~ eo, there is a unique function z* = z*(a, e) which satisfies (1.14a). Furthermore, z*(a, e) has a continuous first derivative with respect to a, e and z*(a, 0) = a*. If there is an ate) with la(e)1 ~ IX for 0 ~ lei ~ eo and (1.15)
G(a, e)
1
= -
T
IT I(t, z*(a, e)(t)) dt = 0, 0
then z*(a, e) is a T-periodic solution of (1.10). Conversely, if (1.10) has a T-periodic solution z(e) which is continuous in e and has Pz(e) = a*(e), a* = (a, 0), la(e)1 ~ IX, 0 ~ lei ~ eo, then z(e) = z*(a(e), e) where z*(a, e) is the function given above and ate) satisfies (1.15) for 0 < lei ~ eo. THEOREM 1.2. Suppose z*(a, e) is defined as in Lemma 1.4 and G(a, e) is defined as in (1.15). If there is an n-vector ao such that
G(ao, 0)
=
det [
0,
8G(a o 8a
,0)] *0,
then there is an el >0 and a T-periodic solution z*(e), lei ~ el, of (1.10), z*(o) = at, a6 = (aQ,O) and z*(e) is continuously differentiable in e. EXERCISE 1.1 Verify all statements made above concerning system (1.10) and prove in detail Lemmas 1.3, 1.4, and Theorem 1.2. EXERCISE 1.2.
Consider the equations
Xl =X2,
X2 =
-Xl
+ e(1 -
xi)x2 + ep cos(wt + IX),
+
where p * 0, e > 0, W, IX are real numbers with w 2 = 1 e{3, (3 * 0. Find conditions on p, w, IX which will ensure that this equation has a periodic solution of period 27T/W for e small. Draw the frequency response curve. Let Xl
X2
and apply Theorem 1.1.
= =
Zl
sin wt
W(ZI
+ Z2 cos wt,
cos wt -
Z2
sin wt),
ORDINARY DIFFERENTIAL EQUATIONS
264
EXERCISE 1.3. Discuss the possibility of the existence of a subharmonic solution of order 2 (that is, a solution whose period is twice the period of the vector field) for the equation (1.16)
iI=X2, X2 = -a2Xl
+ e(3v cos 2t -
xi) - e2(~2
+ p.Xl) -
e3p.X~,
where a = 1. Let
= X2 = Xl
Zl
sin t + Z2 cos t,
Zl
cos t -
Z2
sin t,
and apply Theorem 1.1. EXERCISE 1.4. For a = 2/3, ~ = e~l in (1.16) discuss the existence of a subharmonic solution of order 3 of (1.16). Let Xl
=
Zl
sin at
+ Z2 cos at,
X2 = a(zl cos at -
Z2
sin at),
and determine the Taylor expansion of G(a, e) in (1.6) up through terms of order e and apply an appropriate implicit function theorem. EXERCISE 1.5. For a = 2/n, n a positive integer, show that (1.16) possessing a subharmonic of order n implies that ~ must be 0(e1l - 2 ) as e ~o. You cannot possibly find n terms in the Taylor expansion of G(a, e) so you must discuss the qualitative properties of the expansion. EXERCISE 1.6. equation
Using the method of Exercise 1.5, show that the Duffing
Xl=X2, X2
= -(2n + 1)-2x1 -
eBCx29- etJX1
+ ebx~ + B cos t,
can have a subharmonic solution of order 2n + 1 only if s ~ n. EXERCISE 1.7. Suppose I(x, y) is a continuously differentiable scalar function of the scalars x, y, 1(0,0) = 0 and either I(x, -y) =/(x, y) or I( -X, y) = -/(x, y) for all x, y. Given any ex > 0, show there is an eo > 0 such that for any (xo, YO) with x5 Y5 < ex 2 , there is a periodic solution of the system
+
(1.17)
for
lei;;;; eo.
x=y, if = -x + el(x, y), This implies the equilibrium point (0,0) is a center. Let
PERIODIC SYSTEMS WITH A SMALL PARAMETER
x
265
= p sin e, y = p cos eto obtain the equation dp
de =
eF(p,
e, e)
for the orbits of (1.17). If f( -x, y) = -f(x, y), observe that F(p, -e, e) = -F(p, e, e). Show that the transformation $' in (1.7) in this special case maps even 21T-periodic functions of into even functions of and, therefore, the fixed point must be even. This implies G(a, e) - O. Iff(x, -y) =f(x, y), then let = r/> + Tr/2 and apply the same argument.
e
e
e
EXERCISE
1.8.
Consider the system Xl = X2, •
3
X2= -I-'xl ' where I-' > 0 is a small parameter. A first ~ntegral of this equation is E(xl' x 2) = x~j2 + I-'xt!4 and, thus, all of the orbits are periodic orbits. For I-' = 0, the only periodic orbits are the equilibrium points which lie on the xl-axis. Can you deduce this result by using the above perturbation theory?
JP"
If e = Xl = zl' and x 2 = eZ2, then zl case of system (1.1).
= eZ2 , Z2 = -eZ~ which is a special
The above theory is also useful for determining the characteristic exponents of certain types of linear systems' with periodic coefficients. More specifically, consider the system of equations
w= Cw + ect>(t)W
( 1.18)
where e is a parameter, 0 ~ lei ~ eO, w is an (n + m)-vector, ct>(t) = ct>(t + T), T = 21TjW, IS a continuous1 (n + m) X (n + m) matrix and C = diag(C1 , C2 ) is an (n + m) X (n + m) constant matrix such that all eigenvalues of the n X n matrix eC,T are po and no eigenvalue of eC• T is po. The problem is to determine the characteristic multipliers of (1.18) which are close to po for e =1= O. Suppose Al is an eigenvalue of Cl and then po =·e AIT . From the continuity of the principal matrix solution of (1.18) in e and the Floquet theory, it follows that there is a multiplier p(e) = ell(E)T which is continuous in e and 1-'(0) = AI. Furthermore, to this multiplier there must exist aT-periodic (n m)-vector p(t, e) such that ell(E)tp(t, e) is a solution of (1.18). Conversely, any such solution of (1.18) yields a characteristic exponent of (1.18). Therefore, if one makes the transformation w = elltp in (1.18), then
+
(1.19) 1
jJ
=
(C -I-'I)p
ell could be an integrable function.
+ ect>(t)p,
266
ORDINARY DIFFERENTIAL EQUATIONS
and the problem is to determine p. in such a way that (1.19) has aT-periodic solution. In order to make the previous theory directly applicable, suppose that the eigenvalues of 0 1 have simple elementary divisors. Then the matrix e(C,-A,1lt is T~periodic. If w = (u, v), where u is an n-vector, v is an m-vector, and fJ is any complex number, the transformation u = e(Al+BI1)te(Cl-Al1)tx,
(1.20)
v = e(Al+BlI)ty.
in (1.18) yields (1.21)
i
=-
iJ =
efJx
+ ee-(Cl-All) tct>11(t)e(Cl-A 1)tx + ee-(Cl-A 1)tct>12Y, 1
1
[Oz - (AI + efJ)I]y + ect>zl(t)e(Cl-All)tx + ect>zz(t)y,
where we have partitioned ct> as ct> = (ct>'I). Since e(C,-A,I)t is T-periodic, system (1.21) can be written as
z=
(1.22)
Bz + e'Y(t, fJ)z,
+
where B = diag(On , Oz - All), z = (x, y) and 'Y(t T, fJ) = 'Y(t, fJ) :for all t, fJ. The explicit expression for 'Y is easily obtained from (1.21). If fJ can be determined in such a way that system (1.22) has aT-periodic solutionp(t), then this solution yields from (1.20) a solution of (1.18) of the form w(t)
= e(..t'+Bl1ltp(t),
p(t
+ T) =
p(t).
+
Therefore, Al efJ is a characteristic exponent of (1.18). Conversely, we have seen above that every such characteristic exponent can be obtained in this way. Since system (1.22) is a special. case of system (1.10), we may apply the preceding theory. The function z* given in Lemma 1.4 will now be a function of a, fJ and e and the linearity of (1.22) implies that z*(a, fJ, e) = Z*(fJ, e)a, where Z*(fJ, e) is aT-periodic (n m) X n matrix. If Z* = col (X*, Y*) where X* is n X n then the function G(a, fJ, e) in (1.15) is also linear in a; that is, G(a, fJ, e) = D(fJ,.e)a, where D(fJ, e) can be computed directly from (1.22) as
+
(1.23)
D(fJ, e)
=
1 T T fo ['Yl1 (t, fJ)X*(fJ, e)(t)
+ 'YIZ(t, fJ) Y*(fJ, e)(t)] dt,
where 'Y is partitioned as 'Y = ('Yt1 ). Lemma 1.4 then implies that any f3 for -which det D(fJ, e) = 0 will yield a characteristic exponent Al efJ of (1,18).
+
tha~lthe
EXERCISE 1.9. Suppose D(fJ, e) is given in (1.23). Prove n characteristic multipliers of (1.18) which for e = 0 are equal to po are the n roots of the equation det D(fJ, e) =0.
PERIODIC SYSTEMS WITH A SMALL PARAMETER EXERCISE 1.10.
267
Consider the system W = Ow
(1.24)
+ 8
where
k =1= j.
Show that the characteristic multiplier p(8) of (1.24) such that p(O) = given by p(8) =
where
"'(8) = Aj
ell(e)T,
(~k)'
+ -T8 fT0 CPjj(t) dt + 0(8)
eAiT
is
as 8-*0,
i, k = 1,2, ... , n.
EXERCISE 1.11.
Suppose that the system
x + x = 8f(x, i, y, y), ii + a 2y =
8g(X, i, y, y),
has a periodic solution of period 21T/W(8), w(O) = 1, which for 8 = 0 is given by x = a sin t, y = O. Prove that this solution is asymptotically orbitally stable with asymptotic phase if
fo --:ofoxfo dyog 211
8
211
8
--:-
(a sin t, a cos t, 0,0) dt
< 0,
(a sin t, a cos t, 0, 0) dt
< 0,
and a =1= an integer. The characteristic multipliers. of the linear variational equation relative to this periodic solution are given by 1, 1, e 2nil1 , e- 2nil1 for 8 = O. The hypothesis on a implies that the roots e 2nil1 and e- 271:il1 are simple. Therefore, Exercise 1.10 can be applied to obtain the first order change in these multipliers with 8 provided the constant part of the variational equation is transformed to a diagonal form. Since one multiplier remains identically one for 8 0/- 0 and the product of the multipliers is given by a well known formula, one can evaluate the first order change in the other multiplier. EXERCISE 1.12. Generalize the result of Exercise 1.11 to a system of n-second order equations. EXERCISE 1.13 .. Show that the system
x+ x =
8(I-x2 -y2)i, ii + 2y = 8(1 - x 2 - y2)y,
8>0,
ORDINARY DIFFERENTIAL EQUA.TIONS
268
has two nonconstant periodic solutions both of which are asymptotically orbitally stable with asymptotic phase. EXERCISE 1.14. (1.25)
Consider the Mathieu equation X1=X2, X2
=
-0'2X1 - e(cos 2t)X1,
where 0' = O'(e) and 0'(0) = m, a nonnegative integer. From the general theory of this equation in Chapter III, we know that these values of 0' are precisely the ones which may give rise to instability. In fact, the instability zones are determined from those values of 0' for which equation (1.25) has a periodic solution of period 17 or 217; that is, the multipliers are +1 or -1. Determine approximately these values of O'(e) as a function of e for the case when 0'(0) = 1, 0'(0) = 2. Are the solutions unbounded in a neighborhood of the points (1, 0), (2, 0) in the (0', e)-plane? Let Xl
= %1 sin mt + %2 cos mt,-
X2 = m[%l cos
me -
%2
sin mt],
and m 2 - 0'2 = ef3 and apply the above theory for determiningf3 in such a way that the resulting equations have a periodic solution.
+
EXERCISE 1.15. O(e m) as e-+O.
For 0'(0) = m in Exercise 1.14, show that 0'2(e) = m 2
EXERCISE 1.16. Discuss as in Exercise 1.14 the equation (1.25) with Suppose e ~ 0, let 0'2 = ef3, Xl = %1, X2 = Jez2 . and analyze the resulting equations for periodic solutions. 0'(0)
= O.
EXERCISE 1.17. equation
For what values of w are all of the solutions of the
x + 0'2x =
e(sin wt)y,
ii + p.2y =
e(cos wt)x,
bounded for e small and =F O? Let us use the same ideas as above to discuss the existence of periodic solutions for equations which may contain several independent parameters rather than just a single parameter e as in (1.1), (1.10). Consider the equation'
(1.26)
x=!(t,X,A).
rt" where Ain Rk is a parameter, x is in R n , !(t, x, A) is continuous in t, xiX'together with at least first derivatives in X,A, !(t + T,X,A) = !(t,X,A),
(1.27)
!(t,O,O) = 0,
3!(t,0,0)/3x = 0
PERIODIC SYSTEMS WITH A SMALL PARAMETER
269
Equation (1.26) is a generalization of (1.1) since, for example, for X = (Xl ,X2), it could be of the form
x= XIf(I)(t, x) + X2/(2)(t, x) which makes (1.1) correspond to [(2)(t, x) = O. Equation (1.26) could also be of the form k
x= [(O)(x) + j~l Xj[(j)(t, x) where [(0)(0) = 0, o[(O)(O)/ox = O. If [(O)(x) 'if:. 0, this latter equation corx, X) which can be made small together with its first derivaresponds to an tive by choosing X small and x small. In problems of this type, we will be interested in T-periodic solutions with small norm. Lemmas 1.1, 1.2 have an analogue for system (1.26), (1.27) which we state without proof since the proof Will be the same.
ret,
LEMMA 1.3. Suppose P,Yare defined in (1.2), (1.4). Then x(t) is aT-periodic solution of (1.26) if and only if
(a) x=a+Y(I-P)[(-,x(-),X) (1.28)
(b) P[(-,x(-),X) = 0
where a is a constant n-vector given by a = Px. LEMMA 1.4. If (1.27) is satisfied, then there are Xo > 0, Q(j > 0 such that, for any a in R n , lal ~ Q(j, any X in Rk, IXI ~ Ao, there is a unique T-periodic function x* = x*(a, X) continuous together with its first derivatives, satisfying (1.28a), x*(O,O) = o. If there exist (a,X) such that lal ~ Q(j, IXI ~ Xo and
(1.29)
6 G(a,X) def = T
) (t),X)dt= 0 Jro [(t, x* (a,A T
then x*(a,X) is a T-periodic solution of (1.26). Conversely, if (1.26) has a T-periodic solution x
x= [(t,x,y,X) (1.30)
y=BIY
+ g(t,x,y,X)
where III T - I is nonsingular if one supposes {,g and' their derivatives with respect to x,y vanish at x = 0, y = 0, X = O. We do.state the results explicitly.
270
ORDINARY DIFFERENTIAL EQUATIONS
Having the method formulated so as to apply to (1.26) and (1.30) gives the opportunity to discuss more general problems as well as to discuss old problems more completely. We illustrate this with some examples. The simplest example is the so-called Hopf bifurcation. Suppose A is a scalar, A(A) is a x X 2 matrix continuously differentiable in A with the eigenvalues of A(O) being ±i. Then there is a neighborhood of A = 0 for which the matrix A(A) has two complex conjugate eigenvalues p.(A) ± iv(A), continuous and continuously differentiable in A such that p.(0) = 0, v(0) = i. Let us suppose dp.(O)/tA =1= O. This hypothesis implies that we may redefine the eter A in a neighborhood of zero by replacing A by p.(A), (j(A) = -:[V(p-l(A»)2 and assume A(A) has the form (1.31)
A (j(A») A(A) = ( _(j(A) A '
(j(0)= 1
Consider the second or.der equation
x= A (A) x + f(X,A)
(1.32) where (1.33)
f(O,A)
= 0,
af(O,A)/aX = 0
and A (A) is given in (1.31). The Hopf Bifurcation Theorem is stated explicitly in the follOWing way. THEOREM 1.3. Consider system (1.31), (1.32), (1.33) and suppose fhas continuous second derivatives in x. Then there are Ao > 0, ao> 0, 80 > 0, scalar functions A*(a), w*(a) and an w*(a)-periodic function x*(a) with all functions being continuously differentiable in a for lal < flO such tl}at x*(a) is a solution of the equation
x= A(A*(a»x + f(x, A*(a» and w*(O) = 2A,
A*(O) = 0, x*(a) = (
acost) .
-asmt
+ o(lal).
Furthermore, for IAI < AO, Iw - 21T1 < 80, every w-periodic solution of (1.32) with norm <8 Ii must be of the above type except for a translatioy.' in phase..
:,.
PROOF. Since we are in the plane, every periodic orbit of (1.32) must encircle an equilibrium point. The hypothesis on A(A) implies detA(A) =1=0 for A close to zero. Thus, the solution x = 0 of (1.32) is isolated for IAI sufficiently small
PERIODIC SYSTEMS WITH A SMALL PARAMETER
and any solution of (1.32) with sufficiently small norm must encircle x This justifies the introduction of polar coordinates (1.34)
Xl
= pcosO,
271
= O.
X2 = -psinO
to obtain the equivalent system
() = (3Q..) -!.p [11 sin 0 + h cos 0] (1.35)
p= AP + 11 cos 0 - h sin 0
For IAI, Ipl small, () > 0 and so we can eliminate t to obtain
dp dO
(1.36)
= a(A)p +R(O,p,A)
where R(O + 2rr,p,A) = R(O,p,A), R(O,O,A) = 0, oR (0 ,0, A)/op' = 0, a(O) = 0, a'(O) = 1. Finding periodic solutions of period 2rr of (1.36) is equivalent to finding periodic orbits of the original equation (1.32). If p(O) is a 2rr-periodic solution of (1.36), one obtains the period of the p'eriodic orbit Xl = p(O)cosO, X2 = -p(O)sin 0 by finding that value of w for which O(w) = 2rr where O(t), 0(0) = 0, satisfies the first equation in (1.35) with p = p(O). Lemma 1.4 is applicable directly to (1.36). Let p*(O,.a,A) be the unique solution for a. Asmall of the equation
p = a +%(1 - P)[aQ..)p + R(·,p,A] Then p*(O ,0, A) = 0 since X = 0 is a solution of (1.32). The bifurcation function GH(a, A) in (1.29) is
r
1 21r GH(a,A) = 2rrJ o [aQ..)p*(O,a,A) + R(O,p*(O,a,A),A)] dO. Since p*(O,O,A) = 0, we have GH(O,A) = o. If we define the function GH(a,A) = GH(a,A)la, then GH(O,O) = 0, oGH(O,O)/oA = 1. The Implicit Function Theorem implies there is a unique A*(a), lal
Verify that GH(a,A) in (1.37) is an odd function of a.
From Exercise 1.18, GH(a,A) = agH(a2,A). Suppose AQ..),f(X, A) in (1.32) have derivatives continuous up through order four and
(1.38)
gn(r,A) = a(A) - 'Y(A)r + o(lrl)
*
as Irl-+ 0
where 'Y(O) = 'Yo 0; that is, GH(a,A) = a(A)a - 'Y(A)a 3 + o(laI 3 ) as lal-+ O. Then, gH(r,A) satisfies gH(O,O) = 0, ogH(O,O)/or = -'Yo O. The Implicit
*
272
ORDINARY DIFFERENTIAL EQUATIONS
Function Theorem implies there is a unique r*(X), IXI < Xo, continuously difr*(O) = 0, r*(X) =1= 0 for 0 < IXI < Xo and gH(r*(X), X) = O. If r*(X) > 0 for 0 < IXI < Xo, then we can define a*(X) by the relation a2 = r*(X) to obtain GH(a*(X),X) = 0 for IXI < Xo. Furthermore, b/r*(X) and a = 0 are the only solutions of G(a,X) = 0 for IXI < Xo, lal 0 for 0 < IXI < Xo is equivalent to (sgn'Yo)X > 0,0 < IXI < Xo. Furthermore, +vr(X) and -vr*(X) correspond to the same periodic orbit of (1.32). Consequently, there is a unique nonconstant periodic solution of (1.32) in a neighborhood of x = 0 if the following conditions are satisfied f~rentiable in X,
(1.39)
'Yo> 0,
o <X < >..0
'Yo <0,
-Xo<X
Let us now investigate the stability of this periodic orbit. It is sufficient to investigate the stability of the corresponding 21T-periodic solution p*(a*(X), X) of Equation (1.36). We will use averaging to determine the stability properties. Make a 21T-periodic transformation of variables in (1.36) of the form
(1.40)
p = s + ul (0);'
+ u 2(O)s3
and choose Ul (0), U2(O) to be 21T-periodic and choose a constant 'Yo so that the differential equation for s has the form (1.41)
ds 3 dB = Xs - 'YOs
+ S(O ,s, X) .
where S(O,s,X) = o(lsl(IXI + s2» as IXI-+ 0, Isl-+ O. This is always possible. To be specific in the discussion of stability suppose 'Yo > O. Equation (1.41) can be considered as a perturbation pf the equation (1.42)
y= -'YOy3
which has y = 0 uniformly asymptotically stable. Thus, there is a neighborhood of y = b say Iyl < 5 such that for IXI < XO(5), every solution yeO) of (1.41) with initial value satisfyingly(O) I ~ 5 must satisfy ly(O) I ~ 5 for t ~ O. Suppose ly(O)1 .~5. Since (1.41) is 21T-periodic, y(O + 21T) is also a solution and yeO) < y(21T) implies y(21Tk) < y(21T(k + 1» for k = 0,1,2, .... Thus, , y(21Tk) -+ y,o as k -+ 00 and it is easy to show that the solution of (1.41) with initial value YO is 21T-periodic. A similar argument holds if yeO) > y(21T). Since the equation is 21T-periodic,this implies that, for every solution yet) of (1.41) with ly(O)1 ~ 5, IXI ~ XO, there is a 21T-periodic solution l/>y(O) such that y(~) - l/>y(O) -+ 0, as 0 -+ 00. This means the corresponding periodic orbit of (1.32) in R 2 corresponding to l/>y(O) is attracting the orbit corresponding to yeO). Since 'YO =1= 0; there is only one nontrivial periodic orbit. Since 'Yo > 0, the trivial solution is always unstable for 0 < X < >..0 and we have the nontrivial periodic orbit is asymptotically orbitally stable.
PERIODIC SYSTEMS WITH A SMALL PARAMETER
273
If 'Yo < 0, one argues in a similar way to obtain the nontrivial periodic orbit is unstable. We summarize these remarks as follows. COROLLARY 1.1. ~uppose A(A),f(Z,A) have derivatives contmuous up through order four and suppose GH(a, A) in (1.37) satisfies GH(a, A) = agH(a2, A) where gH(r, A) satisfies
= A- 'Y(A)r + o(lrl) 'Y(O) = 'Yo =1= O.
gH(r,A)
Then there is a neighborhood of A = 0, x following properties:
(i) 'Yo> 0, -Ao < A ~ 0,
o
asr-+O
= 0 in which equation (1.32) has the
= 0 is asymptotically stable and there is no periodic orbit.
x
°
x = is unstable and there is an asymptotically orbitally stable periodic orbit.
= 0 is asymptotically stable and there is an unstable periodic orbit.
x
x = 0 is unstable and there is no periodic orbit. EXERCISE 1.19. Prove there exists asymptotic phase in Corollary 1.1. Hint: Show the characteristic exponent for the solution s of (1.41) is not zero. EXERCISE 1.20. Suppose GH(a,A) in (1.37) satisfies GH(a, A) = agH(a2, A) with g(r,O) = cr m + o(lrlm) as r -+ 0 for some c =1= and integer mZ1. State and prove the analogue of Corollary 1.1.
°
EXERCISE 1.21. Generalize the Hopf bifurcation theorem to a system
x = A(A)X + f(X,y,A) y=
B(A)y + g(X,y,A)
where f,g, and their partial derivatives vanish for x = 0, y = 0, x in R2, yin Rn. Suppose A(A) as before and det [/ - expB(0)21T] =1= O. Give conditions which will ensure there is a unique non constant period orbit for a given A sufficiently small. Discuss the stability properties of this orbit. EXERCISE 1.22. Consider the scalar equation oX = x 2 + f(t,X,A) where f(t + !, x, A) = f(t, x, A) has ~ontinuous derivatives up through order two in
274
ORDINARY DIFFERENTIAL EQUATIONS
X,A;{(t,X,O) = o(lxI2) as Ixi -+ 0, l(t,O,A)
>..0
> 0,
~o
= O(lAI) as A -+ O. Prove there is a
>" 0, and a continuously differentiable function -y(A), IAI
< >..0,
'Y(O) = 0, such that the following properties hold with respect to I-periodic solutions of the above equation with norm
<~o:
(i) 'Y(A) > 0 implies no I-periodic solution (ii) 'Y(A) = 0 implies one I-periodic solution (iii) 'Y(A) < 0 implies two I-periodic solutions. Note that your results are true for A a parameter in a Banach space. Discuss the .stability properties in case (ii). In case (iii), if X1(t,A) < X2(t,A) are the 1periodic solutions, show that X1(t,A) is uniformly asymptotically stable and X2(t,A) is unstable. Hint: Obtain the bifurcation function G(a,A) from (1.29) and observe that G(a,O) = a2 + o(laI 2) as lal-+ O. Thus, G(a, A) has a minimum at some a*(A) for IAI small, a*(O) = O. Let 'Y(A) = G(a*(A),A). For case (ii), note that every bounded solution of the differential equation must approach a I-periodic solution. At a zero a of G(a, A), note that the sign of aG(a, A)/aa is the same as the sign of the characteristic exponent of the corresponding periodic solution if this characteristic exponent is not zero. If two distinct periodic solutions exist and one characteristic exponent is zero, show that an appropriate small perturbation of the vector field will yield at least three periodic solutions near zero, which is a contradiction. EXERCISE 1.23. the equation in R2,
Give an appropriate generalization of Exercise 1.22 to •_
2
(
x - x +1 t,X,y,A) y= -y + g(t,X,y,A) EXERCISE 1.24. Consider the scalar equation
x= -x3 + Atf1(t,X) + A2f2(t,X) where A = (AIoA2) is inR 2,.fj(t + l,x) =.fj(t,x) for all t,x,j= 1,2, 'Y1 = gf2(t,O)dt=F=O . 1
It (t,O) = 0,
'Y2
= fo
[all (t,O)/ax] dt =F= 0
Discuss the existence of I-periodic solutions in a neighborhood of x = 0, A = O. Show there is a cusp r in A-space such that on one side of r there is. one 1periodic solution and the other side there are three. Show the cusp isJ'pproximately given by A~ = (4-yV27'Y~)A~. Hint: Show that the bifurcatioi"tunction G(a,A) is approximately given by -~ + A2'Y2a + Al 'Y1 and deterIIl~e those A1oA2 as functions of a so that G(a, A) = 0, aG(a,A) = O.
PERIODIC SYSTEMS WITH A SMALL PARAMETER
EXERCISE 1.25. 1.24.
275
Discuss the stability of the periodic solution in Exercise
EXERCISE 1.26. Generalize Exercise (1.24) to the equation
x= -x 3 + [(f,x,X) where X is a parameter varying in a Banach space. EXERCISE 1.27. Generalize Exercise (1.24) to the planar system
x=
+ [(f,x,X) y= -y + g(f,x,X)
vm.2.
-X 3
Almost Linear Systems
In this section, the general perturbation scheme given in Section 1 will be .generalized to the system (2.1)
x
= B(t)x + e/(t, x),
where B(t + T) = B(t),f(t + T, x) =/(t, x) and o/(t, x)/ox are continuous for all t in R, x in en . • Let the columns of (t), a matrix of dimension n X p, be a basis for the T-periodic solutions of (2.2)
x=.B(t)x,
and let the rows of 'Y(t), a matrix of dimension p X n, be a basis for the T-periodic solutions of the adjoint equation
if =
(2.3) The p (2.4)
X P
-yB(t).
matrices (' denotes transpose)
" e
=
fo
T
fo
T
'(t)(t) dt,
D
=
'Y(t)'Y'(t) dt
f:
are nonsingular. In fact, if q(t) = (t)a and ea = 0, then q'(t)q(t) dt = 0 which implies q(t) = 0 for all t in [0, T]. But this implies a = O. The same argument holds for the matrix D. As before, let f!JJT be the space of continuous T-periodic n-vector functions with the uniform topology. The fact that e, Din (2.4) are nonsingular allows one to define two projection operators P, Q on f!jJ T in the following way:
276
ORDINARY DIFFERENTIAL EQUATIONS
(2.5)
e-1 f
Pf = <1>( • la,
a=
Qf = 'Y'(. )b,
b = D-l
T
o
<1>'(t)f (t) dt,
fo 'Y(t)f(t) dt. T
It is easy to check that these are projections. Notice that P takes' &T onto the subspace of i?J T spanned by the T-periodic solutions of (2.2) and Q takes i?JT onto the subspace of i?JT spanned by the transpose of the T-periodic solutions of (2.3). The justification for these definitions lies in the following lemma. LEMMA 2.1. Iff is a given element of &T, then a necessary and sufficient condition that the equation
x=
(2.6)
B(t)x + f(t)
has a T-periodic solution is that Qf = O. If Qf = 0, then there is a unique T-periodic solution .:tt"f such that P.:tt"f = O. Furthermore, .:tt"(I - Q) is a continuous linear operator taking i?J T into & T . PROOF. The first part of the lemma is a restatement of Lemma IV.1.l. If Qf = 0, then there is a solution of (2.6) in i?JT. If Xo is the initial value of any solution in i?JT, then Xo must satisfy the equation Exo = b, where
E = X(T) - I, b =
f:
X(T)X-l(s)f(s)ds and X(t) is the principal matrix
solution of x = B(t)x. Let E* be a right inverse of E; that is, a matrix taking the range of E into en such that EE* = 1: Since Qf = 0 implies b is in the range of E, the vector Xo = E*b corresponds to the initial value of a solution x*(j) of (2.6) in &T. If :.tt'J = (1 - P)x*(j) then the operator .:tt" takes &T inte &T and is clearly linear and continuous. Also, P.:tt"f = 0 and .:tt"f is the only solution of (2.6) in &T whose P-projection is zero. This completes the proof of Lemma 2.1. COROLLARY 2.1. Iff is in &T and a is a givenp-dimensional vector, then the unique solution of x = B(t)x (I - Q)f with Px = <1>( • )a is given by <1>a .:tt"(I - Q)f.
+
+
Example 2.1. Suppose B = B(t) is a constant n X n matrix, B = diag(Op, B 1 ) where Op is the p-dimensional zero matrix and eB1T -1 is nonsingular. Then <1>, 'P' may be chosen as <1> = [It
l
'Y = <1>' = [Ip, 0],
where 1 p is the p X P identity matrix. It is easy to see that the matflces e, D in (2.4) and operators P, Q in (2.5) are given by e = D = TIp .
277
PERIODIC SYSTEMS WITH A SMALL PARAMETER
where f p denotes the first p components off. .This is the same operator defined in Section 1 for equation (1.10). If Qf = 0, then the explicit expression for .Yt'f is
(.Yt'f)(t) =
'[
ffp(8)d8
.
1 ,
(e-B,T _ J)-lf T e- B'8fn_p(t + 8) d8 o
where
ffp is the unique primitive of f p of mean value zero, and f n-p
denotes the last n - p components of f. Example 2.2.
Suppose B =B(t) is constant and
[~ ~l
B=
We may choose = col(l, 0) and '¥ = row(O, 1). Then C = D = T and
[~] (~ {T JI(8) d8),
Pf = (')a = Qf = o/'(·)b = where f
[n (~
foT 12(8) d8),
= (h h)· If Qf = 0, one easily computes .Yt'f to be
where" f t " again denotes the primitive of the integrand of mean value zero. LEMMA 2.2. If the operators P, Q and .Yt' are defined as in (2.5) and Lemma 2.1, then system (2.1) has a T-periodic solution x if and only if x satisfies the system 'Of equations (2.7)
(a) (b)
+
x = Px e.Yt'(J - Q)f(·, x), eQf(·, x) =0.
278
ORDINARY DIFFERENTIAL EQUATIONS
PROOF. SUppose X is any element of Q(i: - Bx) = O. In fact,. from (2.l?),
Q(i: - Bx) ='1'"' D-l
='1'"' [)-l
~ T.
We first show that
fo 'I'"(t)[i:(t) - B(t)x(t)] dt fo ['I'"(t)x(t) +'Y(t)x(t)] dt T
T
= '1'"' D-l[,¥(t)x(t)]g' = o. An element x of f?JJT is a solution of (2.1) if and only if
Q(i: - Bx) =eQf ( ., x), (1 - Q)(i: - Bx) =e(1 - Q)f(· , x). Since Q(i: - Bx) = 0, these equations are equivalent to (2.7b) and i: - Bx = (1 -Q)f(·, x). Corollary 2.1 implies this latter equation is equivalent to (2.70.) and this proves the lemma. LEMMA 2.3. For any ex > 0, there is an eo:> 0 such that for any con· stant p.vector a, lal ~ ex, lei ~ eo, there is a unique function x* =x*(a, e) which satisfies (2.8)
x* = a + e~(1 - Q)f ( ., x*).
Furthermore, x*(a, e) has a continuous first derivative with respect to a, e and x*(a, 0) = a. If there is an a = a(e) with la(e)1 ~ ex, 0 ~ lei ~ eo and (2.9)
G(a, e)~ Qf(·, x*(a, e» = 0,
then x*(a, e) is a T'periodic solution of (2.1). Conversely, if (2.1) has a T.periodic solution x(e) which is continuous in e and has Px(e) = a(e), la(e)1 ~ ex, 0 ~ lei ~ eo, then x(e) = x*(a(e), e) where i*(a, e) is the .function defined above and a(e) satisfies (2.9) for 0 < lei ~ eo. PROOF. Suppose ex> 0 is given and a is any p.vector with lal ~ ex. Let fJ be a positive number such that lIall ~ fJ for lal ~ ex. For any y > 0, define 9'(y) = {y in f?JJT: Py = 0, lIyll ~ y} and the operator fF: 9'(y) -+-f?JJT by (2.10)
fFy = e ~(1 -Q)f(·, y + a)
for y in 9'(y). If y*(a, e) is a fixed point of fF in 9'(y), then x*(a, e) = a y*(a, e) is a solution of (2.8). The proof now proceeds exactly as in the proof of Lemma 1.2. ~om th~ po~t of view of appli~tions, i~ is co~venient to obse?e that relatIon (2.5) Impbes that G(a, e) = 0 In (2.9) 18 eqUivalent to '"
+
(2.11)
delfT 'I'"(t)/(t, . x*(a, e)(t» dt =
F(a, e) =
o
O.
.
279
PERIODIC SYSTEMS WITH A SMALL PARAMETER
Lemma 2.3 asserts the following: one can arbitrarily preassign an element a ofthe subspace of fY' p defined by {f in fY' p: Pf = f} and then uniquely determine a solution of (2.7a) with Px replaced by xp. The element xp is then used to attempt to solve the remaining equation (2.7b). Equations (2.9) or (2.11) are called the determining equations or bifurcation equations of (2.1) and can be determined approximately by successive approximations since !F in (2.lO) is a contraction operator. Taking only the first approximation a for x*(a, e) and using the implicit function theorem, one arrives at xp
=
THEOREM 2.1.
Suppose x*(a, e) is defined as in Lemma 2.3 and let
·F(a, e) be defined by (2.11). If there is a p-vector ao such that F(ao, 0) =0,
det[8F(;~, 0)]
:;60,
°
then there are an e1 >0 and a T-periodic solution x*(e), ~ (2.1), x*(o) = ~o and x*(e) is continuously differentiable in e ..
lei ~ e1,
of
EXERCISE 2.1. Iff is in £JJP and Qf = 0, give a constructive procedure for determining the function :Yt'f in £JJP given in Lemma 2.1. EXERCISE 2.2. State and prove the appropriate generalizations of the results of this section for the equation
x = B(t)x + f(t, x, e),
+
where B(t T) = B(t), f (t R xCn xC, and
+ T, x, e) = f(t, x, e) are continuous for (t, x, e) in f(t,
If(t, x, e) - f(t,
0, 0) = 0, y, e)!
~ 1](iel,
a}jx -
yl,
for t in R, e in C, x, y in cn, Ixl < a, Iyl < a and 1](at, iT) is a continuous nondecreasing function for at ~ 0, a ~'o, 1](0,0) = 0. EXERCISE 2.3. (2.12)
Consider the system Xl
= X2,
X2 = e[(a
+ at cos 2t)Xl + bx~ + cx2 ],
where a, at, b, C are constants. Find conditions on these constants which will ensure that 7T~periodic solutions exist. What are their stability properties?
°
EXERCISE 2.4. Find conditions on a so that (2.12) for b = C = has a 7T-periodic solution. Interpret your result in the light of the theory of the Mathieu equation.
ORDINARY DIFFERENTIAL EQUATIONS
280
If the system (2.1) possesses some additional properties, it is sometimes possible to make qualitative statements about the bifurcation equations (2.9) or (2.11) without any successive approximations. The next few pages are devoted to this question. Definition 2.1, A system of differential equations d.: = g(t, x) where x, g are n- vectors is said to have property (E) with respect to S if there exists a symmetric, constant n X n matrix S such that S2 = 1 and Sg( -t, Sx) = -g(t, x)
for all t, x. If x(t) is a solution of a system which has property (E) with respect to S, notice that Sx( -t) is also a solution. In fact, if x is a solution of such a system and w(t) = Sx(t), then tb(t)
= Sd.:(t) = Sg(t, x(t)) = Sg(t, S-lw(t)) = Sg(t, Sw(t)) =
This implies w( -t)
-g( -t, w(t)).
= Sx( -t) is a solution of the original equation.
LEMMA 2.4. Suppose S is an n X n symmetric matrix, S2 = 1, B(t) = B(t T) is an n X n continuous matrix and f(t) = f(t T) is a continuous n-vector such that
+
+
(a)
(2.13)
(b)
= Sf( -t) =
SB(-t)
-B(t)S, -f(t).
Let the columns of the n X p matrix be a basis for the T-periodic solutions of (2.2), :1( be defined as in Lemma 2.1 and Q as in (2.5). Then (2.14)
(a)
S( -t)a = (t)a for all t if S(O)a = (O)a,
(b)
S(Qf) ( -t) = -(Qf)(t),
(c)
S[:1((1 -Q)f]( -t) = [:1((1 -Q)f](t).
PROOF. The fact that B satisfies (2.13) implies that system (2.2) has property (E) with respect to S and thus, if S(O)a = (O)a, then uniqueness of the solutions implies S( -t)a = (t)a for all t. This proves (2.14a). The matrix S( -t) is a basis for the T-periodic solutions of (2.2) and, consequently, there is a p X P nonsingular matrix r such that S( -t) = (t)r for all t. In the same way, there is a p X P nonsingular matrix M such that 'Y( -t)S = M'Y(t) for all t. For any f satisfying (2.13b), we have from (2.5), and S' = S that
fo fo
T
S(Qf) ( -t) = S'Y'( -t)D-l
'Y(a)f(a) da
T
='Y'(t)M'D-l
'Y( -a)f( -a) da
PERIODIC SYSTEMS WITH A SMALL PARAMETER
281
fo 'Y( -cx)Sf(cx) dcx ::: -'Y'(t)M'D-lM f 'Y(cx)f(cx) dcx o == -'Y'(t)M'D-l
p
p
On the other hand, from (2.4) and 8' = S, S2 = J, one shows that M DM' = D. Therefore, M'D-IM = D-l and p
S(Qf)( -t) = -'Y'(t)D-l
f. 'Y(cx)f(cx) dcx o
= -(Qf)(t). This proves (2.14b). If f satisfies (2.13b), then (2.14b) implies that (J - Q)f satisfies S(J - Q)f ( -t) = -(J - Q)f (t).
Therefore, to prove (2.14c), it is sufficient to show that (2.14c) is satisfied with = O. From (2.5), this is true if and only if
(J - Q)f replaced by f and Qf = O. We first show that P8(:Kf)( _.) p
0= =
fo '(t)S(:Kf)( -t) dt
fo <1>'( -t)(:Kf)( -t) dt r' f '(t)(:Kf)(t) dt. o p
r'
p
=
f:
Since P:Kf= Oit follows that '(t)(:Kf)(t)dt= O. Therefore, PS(:Kf)( _.) = O. Since :Kf is the unique solution of (2.6) with P:Kf = 0 and S:Kf ( - . ) is also a solution of this same equation with P-projection zero, it follows that (2.14c) is true. This proves the lemma. THEOREM 2.2. Let the columns of the n X p matri;x: be a basis for the T-periodic solutions of (2.2) and let the rows of the p X n matrix 'Y be a basis for the T-periodic solutions of (2.3). Let x*(a, 8) be the function determined in Lemma 2.3 and suppose system (2.1) has property (E) with respeot toS. Then (2.15)
Sx*(a, 8)( -t) = x*(a; 8)(t)
for all t provided that (J -S)(O)a = O. In addition, if F(a, 8) is defined as in (2.11), ·then (2;16)
(1
+ M)F(a, 8) =
O.
where M is the matrix suoh that 'Y( -t)8 = M'Y(t).
282
ORDINARY DIFFERENTIAL EQUATIONS
PROOF. Suppose 8<1>(0)a = (O)a; let 9'(,,) be defined as in the proof of Lemma 2.3 and 9'*(,,) be the subset of 9'(,,) consisting of those y in 9'(,,) which satisfy 8y( -t) = y(t). From Lemma 2.4, Fy = e%(1 -Q)/(', y a) belongs to 9'*(,,) for lei ~ eo where eo is given in Lemma 2.3. Since this operator has a unique fixed point y*(a, e) in 8(,,),8y*(a, e)( - t) = y*(a,e) (t). Therefore (2.1480) implies x*(a, e) = y*(a, e) a satisfies relation (2.15). Using (2.15), property (E) and the fact that '1'( -t)8 = .M'Y(t), one obtains
+
+
'1'
F(a, e)==
fo
'I'(t)/(t, x*(a, e)(t» dt
fo
'I'(t)/(t, 8x*(a, e)( -t» dt
'1'
=
=-
..
fo
'1'
= -M
'I'(t)8/( -t, x*(a, e)( -t» dt
f '1'( '1'
-..t)f( -t, x*(a, e)( -t» dt
.0
= -MF(a, e). This proves the theorem. In the applications, Theorem 2.2 can be very usefUl and especially the conclusion expressed in (2.16). This relation says that the bifurcation functions given by the components of the vector Fin (2.11) are dependent provided that the p-vector a satisfies (1 - 8)(0)a = 0. This can lead to results for (2.1) involving the existence offamilies of periodic solutions. The following exercises illustrate this point and involve situations in which property (E) expresses some even and oddness· properties of the functions in (2.1). EXERCISE 2.5. Suppose/( -t, x) = -/(t, x) for all t in B, x in Btl. Show that the system i = e/(t, x) has an n-parameter family of periodic solutions for e small. If/(t, x) = A(t)x and A( -t) = -A(t), this implies in particular all characteristic multipliers are 1. EXERCISE 2.6.
Reconsider Exercise 1.7 in the light of Theorem 2.2 .
EXERCISE 2.7.
Show that all solutions of the equation
g + 02y =
eg(y, y,
°
'Ii),
0*,0,
in a neighborhood of y = Y= 'Ii = are periodic for e small if g(y, -]i, 'Ii) = -g(y, y, 'Ii). Write the equation as a third order system i = A(02)x tJ~/(t, x), let x = [exp A(T2)t]z, T2 = 0 2 +efJ and apply the previous results· to the equation for z using fJ as one of the undetermined parameters. EXERCISE 2.8.
Consider the system of second order equations
PERIODIC SYSTEMS WITH A SMALL PARAMETER
283
+ a2u = egl(U, U, v, til, v + p. 2v = eg2(u, U, v, til,
ii
where a =1= mp., p. =1= ma for m
= 0, ±1, ... and
u, v, -til = -gl(U, u, v, til, g2( -u, u, v, -til = g2(U, u, v, til. gl( -u,
Show that this system has two 2-parameter families of periodic solutions. Let Xl = U, X2 = au, X3 = V, V4 = p.ti to obtain a fourth order system i = A(a, p.)x + ef(x) where A(a, p.) = diag(B(a), B(p.)), X = (y, z), y, z two-vectors and let y = [exp B(T)t]Y, z= Z, T = a + efJ and apply Theorem 2.2 to the equation for (Y, Z) using fJ as one of the undetermined parameters. Generalize this result to other types of symmetry as well as a system of n-second order equations. Definition 2.2. IAjirst integrallof a system (2.17)
i=g(t,x)
is a function u: R such that
X
en ~ e which has continuous first partial derivatives ou(t, x) ou(t, x) --g(t,x)+--=O ox ot
for all t, x.
If x(t, to, xo), x(to, to, xo) = Xo is a solution of (2.17) and u is a first integral of (2.17), then u(t, x(t, to, xo)) = u(to, xo) for all t for which x(t, to, xo) is defined. If to = 0, we write x(t, xo), x(O, xo) = xo, for a solution of (2.17). Let V(t, xo) = ox(t, xo)/oxo. In Chapter I, we have seen that V is a principal matrix solution of the equation (2.18)
z=
Hz,
H = H(t, xo) = og(t, x(t, xo))/ox.
LEMMA 2.5. If u is a first integral of (2.17) which has continuous second partial derivatives, then ux(t,xo) ~ ou(t, x(t, xo))/ox is a solution of the adjoint equation (2.19)
w=-wH,
where H is defined in (2.18). PROOF.
= u(O, xo) for all t and all Xo, it follows that ux(O, xo) for all t. Differentiating this relation with
Since u(t, x(t, xo))
ux(t, x(t, xo)) V(t, xo)
=
respect to t, one obtains
ORDINARY DIFFERENTIAL EQUATIONS
284
for all t. Since V is nonsingular, proved.
Ux
must satisfy (2.19) and the lemma is
LEMMA 2.6. If u(t + ,]" x, e) = u(t, x, e) is a first integral of (2.1) which has continuous second partial derivatives with respect to t and x and x*(a, e) is the T-periodic function given in Lemma 2.3, then
Io [UxQf]x=x.(a, E)(t) dt T
(2.20)
=
0
for e =1= 0, where Q is defined in (2.5). PROOF.
Since u is a first integral of (2.1), we have in particular that
Io [ux(Bx + ef) + Ut]x=x.(a, £)(t) dt T
=
O.
Since x*(a, e)(t) and u(t, x, e) are T-periodic in t, we also have
Io -dtd [u(t, x*(a, e)(t), e)] dt Io [uxx* + ue]x=x.(a, E)(t) dt Io [ux(Bx + ef - eQf) + ue]x=x.(a, E)(t) dt. T
0=
T
=
T
=
Subtracting these two expressions gives (2.20) and proves the lemma. Suppose x*(a, e) is the T-periodic function given in Lemma 2.3 and suppose u(t, x, e) = u(t + T, x, e) is a first integral of (2.1) with ux(t, 0, 0) =1= O. From Lemma 2.5, it follows that ux(t, 0, 0) is a nontrivial T-periodic solution of tb = -wB(t). Since x*(O, 0) = 0, ux{t, x*(O, O)(t), 0) = ux{t, 0, 0) =1= 0 is a T-periodic solution ofthe adjoint equation tb = -wB(t). With the rows of the p X n matrix 'I" defining a basis for the T-periodic solutions of this equation, this implies there is a p-dimensional row vector h =1= 0 such that u'x{t, 0, 0) = h'l"(t). Qbviously, there is a continuous function p.(t, a, e) = p.(t + T, a, e) such that p.(t, 0, 0) = 0 and
. ux(t, x*(a, e)(t), e) = ux{t, 0, 0) + p.(t, a,e) =
h'l"(t) + p.(t, a, e).
If F(a, e)is defined as in (2.11), then (2.5) implies that equation (2.20) can be written as
285
PERIODIC SYSTEMS WITH A SMALL PARAMETER
J
T
0=
[h'l"(t)
o
= for
°< lei
~ eo:
°
+ fL(t, a, e)]'l"'(t)D-lF(a, e) dt
[h+ (fL(t,a, e)'l"'(t)D-ldt]F(a, e)
Since h is nonzero and fL(t, 0, 0) == 0, it follows there are
°
(Xl =F 0, el =F such that h + foT fL(t, a, e)'l"'(t)D-l dt = hI is =F for lal ~ (Xl, lei ~ el· Therefore, there is a linear relation among· the .components of F; that is, a linear relation among the bifurcation functions of (2.1). One of the bifurcation functions is therefore redundant if there is a first integral u( t, x, e) = u(t T, x, e) of (2.1) with Ux (t, 0, 0) =F 0. If Ul, ... , Uk are first integrals of (2.1), we say they are linearly independent if the matrix ux(t, 0, 0) = ou(t, 0, OJ/ox, u = (Ul, ... , Uk) has rank k. Using the same argument as above, one obtains
+
THEOREM 2.3. Let u = (Ul, ... , Uk) be k ~ p linearly independent first integrals of (2.1), u(t p., e) = u(t, x, e) be continuous and have continuous second partial derivatives with respect to t, x. Then there are an (Xl> 0, e1> and a k X P matrix H of rank k such that H F(a, e) = for lal ~ (Xl, lei ~ el and F(a, e) defined in (2.11). If k = p, then there exists a p-parameter family of T-periodic solutions x*(a, e), lal ~ (Xl, lei ~ el, where x*(a, e) is given in Lemma 2.3.
+
°
°
The proof of the first part of this theorem is essentially the same as the above argument for k = 1. If k = p, then H is a nonsingular matrix and HF(a, e) = implies F(a, e) == for lal ~ (Xl, lei ~ el; that is, the bifurcation equations are automatically satisfied. An application of Lemma 2.3 completes the proof of the theorem.
°
EXERCISE 2.9. (2.21)
°
(Liapunov's theorem)
Suppose the system
= aX2 + Jr(Xl, X2, y), X2 = -axl + h(Xl, X2, y), Y = Cy + g(Xl, X2, y), Xl
where a > 0, ii, X2 are scalars, y is an m-vector, Jr, h, g are analytic in a neighborhood of Xl = = X2, Y = with the power series beginning with terms of degree at least two. Suppose this system has a first integral W(Xl, X2, y) which has continuous second derivatives with respect to Xl, X2, Y and the eigenvalues AI, ... , Am of the constant matrix C satisfy Ak =F ian for k = 1, 2, ... , m, and every integer n. Prove that this system has a two parameter family of periodic solutions. Introduce polar type coordinates Xl = P cos aO, X2 = - p sin aO in (2.21) and eliminate t in the resulting equa-
°
°
286
ORDINARY DIFFERENTIAL EQUATIONS
tions to obtain differential equations for p, y as functions of 8. Let p -+ ep, y-+eyand apply Theorem 2.3. If all eigenvalues ofe are purely imaginary, do there exist other families of periodic solutions? What additional conditions are sufficient for the existence of other families of periodic solutions? Is it necessary to assume that /1, h , g are analytic? 2.10. Interpret the general theory of this section for the parorder scalar equation
EXERCISE
ticular
nth
dny -d = ej(t, y). tn EXERCISE
2.11.
Consider the system
to = Aow + ect>(t)w,
(2.22)
where ct>(t + T) = ct>(t) is a continuous n X n matrix and Ao is a constant matrix. Give a constructive procedure which is in the spirit of this section for obtaining a principal matrix solution of (2.22) of the form U(t, e) exp[B(e)t] with U(t T, e) = U(t, e), U(t, 0) = I, B(O) = Ao. Consider the n X n matrix equation
+
(2.23)
TV =
AW
+ ect>(t)W
and, for any given n X n constant matrix AI, let W = U exp[Ao - Allt. The differential equation for U is (2.24)
U=Ao U -
UAo
+ UAI + ect>(t)U.
Determine necessary and sufficient conditions that the nonhomogeneous matrix equation (; = Ao U -
F(t
U Ao
+ T) =
+ F(t).
F(t),
have a T-periodic solution. Use this fact to obtain a set of matrix equations r(AI' e) = 0 whose solution is necessary and sufficient for obtaining a T-periodic matrix solution of (2.24). Show there is always a matrix function AI(e) such that r(AI(e), e) = 0 for e small. EXERCISE 2.12. Generalize exercise 2.11 to a class of almost periodic matrix perturbations ct>(t). EXERCISE 2.13. If Ao in (2.22) is a T-periodic matrix func,t;ion, how could the procedure of exercise (2.11) be modified to obtain thEfprincipal matrix solution? .,._____ In Exercise 2.2, the reader' was supposed to prove that one could use the methods of this chapter to obtain T-periodic solutions for equations which con-
PERIODIC SYSTEMS WITH A SMALL PARAMETER
287
tained nonlinear terms when e = 0. Of course, one must then discuss those solutions which remain in a sufficiently small neighborhood of e = 0, x = 0. As in Section 1, a further generalization is needed to discuss equations which contain several independent parameters. In the next pages, we give this generalization and a few illustrations . .Consider the T-periodic system
x= B(t)x + f(t, X,A)
(2.25)
where B(t + T) = B(t) is a continuous n X n matrix, X is in Rk, f(t + T, x, X) = f(t, x, X) has continuous derivatives up through order one in x, X, (2.26)
{(t,O,O) = 0, 3f(t,0,0)/3x =
o.
The analogue of Lemma 2.3 is the following which is stated without proof. lEMMA 2.7. If P,Q are defined in (2.5), %:gxT -+ giT in Lemma 2.1, and (2.26) is satisfied, then ther~ are Xo > 0, ao > 0, such that, for any a in R n , Ia I :::;; ao; any X in Rk, I XI :::;; Xo, there is a unique T-periodic function x* = x*(a,X) continuous together with its first derivative satisfying x*(O,O) = 0,
x* =
~a
+%(1 -Q)f(·,x*,X).
Ifthere exist a, Xsuch that
TJT 0 w(t)f(t,x*(a,X)(t),X)dt =
def.1
G(a,X) =
0
then 'x*(a,X) is a T-periodic solution of (2.25). Conversely, if (2.25) has a T-periodic solution :X(X) which is continuous in X and has Px(X) =~a(X),' la(X) I :::;; ao, IXI :::;; Xo, then :X(X) = x*(a(X),X) where x*(a,X) is the function given above and a(X) satisfies (2.27) for 0 ::; IXI ::; Xo. Let us illustrate the use of this lemma for Duffing's equation. In Section V.5, we have discussed Duffing's equation with a small harmonic forcing by the method of averaging. In order to do this, we. assumed all of the parameters were multiples of a single small parameter e. Also, because of the nature of averaging, we could only discuss a periodic solution if its characteristic multipliers were not on the unit circle. This implies that the information obtained is incomplete in the neighb'orhood of the points in parameter space where the number of periodic solutions changes from one to three. A more complete discussion can be given based on Lemma 1.4. We give an indication of the procedure below for the case of no damping. Consider the equation (2.28) where 'Y
'* 0 is a fixed real number, F,
W -
1 are small real parameters.The
288
ORDINARY DIFFERENTIAL EQUATIONS
objective is to obtain the 27T/w-periodic solutions of (2.28) which have small norm. If
then (2.29) and we determine 27T-periodic solutions of (2.29) of small:norm. Since 'Y the transformation u = IJ.Lol- I /2x for ~ small can be made to obtain
(2.30)
* 0,
x+ x = -(sgn 'Y)x 3 + AIX + A2cOS t;
where A2 = IJ.Lol l / 2J.L2. Therefore, we must analyze the' existence of small 27Tperiodic solutions of (2.30) for the parameter A = (AI, A2) varying in a neighborhood of A = o. We could transform (2.30) to a system of two first order equations, but it is actually more convenient to work directly with (2.30). Let !fl21r = {x:R ~ R, continuous, 27T-periodic} with the usual norm. If
P:!fl21r ~ !fl21r (2.31)
(Ph)(t) =
r
r
!cos t 21r h(s)cossds + !sint 21r h(s)sinsds 7T
Jo
7T
J0
then P is a continuous projection and the equation (2.32)
X +x= h(t)
for h in !fl21r has a 27T-periodic solution if and only if Ph = O. If Ph = 0, then we can obtain a unique 27T-periodic solution%h of (2.32) which is orthogonal to sin t, cos t. The general 27T-periodic solution of(2.32) is (2.32)
x(t) = acos(t - 4» + (rh)(t)
where a,4> are arbitrary real numbers. Therefore, a 27T-periodic solution of (2.30) is given by
x(t) = acos(t - 4» +%(1 - P)[-(sgn 'Y)x 3 + AIX + A2COS' ](t) P[-(sgn 'Y)x 3 + AIX + A2 cos·J
=0
for some a, 4>. The parameters a,4> must be determined through the bif!&cation equations P[ J = O. For the method, it is convenient to put the parameter 4> in the perturbation term by replacing t by t + 4>; that is, consider the equation
(2.33)
x + x = -(sgn 'Y)x 3 + AIX + A2 cos(t + 4»
PERIODIC SYSTEMS WITH A SMALL PARAMETER
289
and a 211"-periodic solution x so that Px = a cos t for some a. SllCh a solution x must satisfy (a) x(t}= acos t +.1"(/ - P)[-(sgn r)x 3 + AIX + A2 cos(· + t/»](t) (2.34)
(b)
P[-(sgn r)x 3 + AIX + A2 co.s(· + t/»]
=0
In the usual way, solve equation (2.34a) for a 211"-periodic function, x*(t/>,a,A) for a,A small, x*(t/>,a,A) = acos t + o( lal) as a~ O. Then the bifurcation equations are (a) A2 cost/> + 1J 21T [AIX*(t) - (sgn r)x*3(t)] costdt = 0 11" 0 (b) A2Sint/> + 1f211 [AIX*(t) - (sgn r)x*3(t)] sintdt = 0 11" 0 EXERCISE 2.14. Prove x*(t/>,a,A) is an even function of t. From Exercise (2.14), Equation (2.35b) is equivalent to A2sint/> = 0 or t/> = 0 if A2 =1= O. If A2 = 0, then the only 211"-periodic solution of (2.33) near x = 0, Al = 0 is x = o. Thus, we may assume t/> = 0 and have the result that there is a 211"-periodic solution of Equation (2.30) near x = 0, A = 0 if and only if this solution is even, x(t) = x*(O, a, A)(t) and (a, A) satisfy the equation d f 1 i'21T (2.36) G(a,A) ~ A2 + [Alx*(O,a,A)(t) - (sgn r)x*3(0,a,A)(t)] cos tdt = 0 11" 0
It remains to analyze Equation (2.36). It is not difficult to show that
G(a,A) = aO(A) + aI(A)a + a2(A)a2 + a3(A)a 3 + o(laI 3) where ao(A) = A2 + O(IAII + 1A21) aI(A)
= Al + O(IAII +'IA21)
a2(A) = O(IAI)
3 a3(0) = -;pgn r The first objective is to obtain the values of A which give rise to multiple solutions of G(a,A) = 0; that is, those values of A for which (2.37)
G(a, A) = 0,
oG(a, A)/oa = 0
The Jacobian of these two functions with respect to A is given by
IJ=
n I
-1.
290
ORDINARY DIFFERENTIAL EQUATIONS
Consequently, there are unique solutions Af(a),A;(a) of (2.37) near a = 0 and they are given approximately by (2.38) Formulas (2.38) are the parametric representation of a cusp given approximately by A~ = (16/81)(sgn 'Y)A~ in the (AbA2)-plane. Since oG(0,0)/OA2 = 1, it follows that there is one solution on one side of this cusp and three on the other. This analysis gives a complete picture of the small 21T-periodic solutions in a neighborhood of x = 0 for A = (AI,A2) varying independently over a full neighborhood of zero. EXERCISE 2.15. (General) Analyze all the exercises in this chapter in the spirit of the above discussion for Duffmg's equation.
Let us consider another example which will illustrate in a more dramatic way the importance of considering independent variations in the parameters. Suppose g(x) has continuous derivatives up through order 2 in thescaiarvariable x and the equation oX +g(x)
(2.39)
=0
has a family of periodic solutions
w(ao) = 1, and
w'(ao) =1= 0 h(a) =
J,o211'p(t)f(t -
a)dt/
r:!" ,-':p2(t)dt
·-0
then h(a) has a unique maximum at aM in [0,21T), a unique minimum at O!m in [0,21T), h"(o:M) =1= 0, h"(O!m) =1= O. .
PERIODIC SYSTEMS WITH A SMALL PARAMETER
291
We can then state THEOREM 2.4. If (Hd, (H2) are satisfied, then there exist enighborhoods U of r, V of A = 0 and curves em ~ V, eM ~ V, defined respectively by Al = Cm(A2), Al = CM(A2) with cm,CM continuously differentiable in A2, em,eM respectively are tangent to the lines Al = h(lXm)A2, Al = h(aM)A2 at Al = A2 = 0, these curves intersect only at zero, the sectors between these curves containing A2 = 0 has no 21T-periodic solutions and the other sectors contain at least two solutions. If the least period of f(t) is k- I , k ~ 1 an integer, .then x(t) a 21T-periodic solution implies x(t + m/k) , m = 0,1, ...• k -i is also a 21T-periodic solution. PROOF. Our first objective l:s--to obtain a convenient coordinate system near r. The vector r(O) = $(O),p(O» is tangent to rand reO) = (ji(0), -/;(0» is orthogonal to r(O). For a fixed ao, an application of the Implicit Function Theorem shows that the mapping
= pea) + ap(a) y = pea) - ap(a)
x
(2.41)
is a homeomorphism of a neighborhood of (ao,O) into a neighborhood of Cp(ao),p(ao». By using the compactness of r and further restricting the size of a, one easily concludes that there is an ao > 0 such that the set
U= {(x,y) given by (2.41) for 0 ~a< 21T, lal
= pet + a) + z(t + a) = pet + a) + z(t + a)
where (z(a),z(a» = area) for some constant a; that is,(z(O),z(O)) is orthogonal to r(a) = $(a),p(a». If x(t + a) = pet + a) + z(t + a), t + a is replaced by t and frit) = f(t - a), then
(2.42)
z+ gCp)z = AIZ + AlP + A2fotCt) + G(t,z)
where G(t + 21T,Z) = G(t,z) and G(t,z) = 0(lzI 2) as Izl-+ O. We now apply our basic method to Equation (2.42). Hypothesis (HI) implies that all 21T-periodic solutions of the linear equation
292
ORDINARY DIFFERENTIAL EQUATIONS
v+g(p)v=O
(2.43)
are constant multiples of p. Let us define f~1rp2 projection P: !P21r -+- !P21r by the relation
= 71 and define the continuous
"21r
(2.44)
Ptp =+
P50
#/71.
For a given 4> E !P21r' the Fredholm alternative implies that the nonhomogeneous equation
v + g(p)v = 4>(t)
(2.45)
has a 21T-periodic solution if and ortly if Ptp = O. If Ptp = 0 and vo(t) = vo(t;4» is any particular solution of (2.45). then each 21T-periodic solution of (2.45) is given by
v(t)= Pp(t) + vo(t,4» for some constant
p. The solution v will have (v(O), v(O» orthogonal to
p[p(0)2
+ p(0)2]
= -p(O)vo(O;4» - p(O)vo(O;4».
This relation uniquely defines P = P(4)) as a function of 4>. If we defme
(2.47)
~4»(t)=
P(4))p(t) + vo(t,4», 4> E (/ - P)!P21r ,
then one can easily verify that.Jr: (/ - P)!P21r -+- !P21r is a continuous linear operator. With this definition oCr, (2.40) will have a solution of the form, x = p + z, if and only if the following equations are satisfied: (2.48a) (2.48b)
z =.Jr(/ - PH-Ali: - AlP + X2fa
+ G(o,z)]
P[-Ali: - AlP + A2fa + G(o,z)] = O.
An application of the Implicit FuJ\.ction Theorem to (2.48a) shows there is a neighborhood U'.£ !P21r of zero and a neighborhood V.£ R 2 of A = 0 such that (2.48a) has a solution Z*(o:,A) for A in V, 0 ~ 0: ~ 21T, this solution is unique in U, has continuous second derivatives with respect to o:,A and z*(o:,O) = 0 for all 0:. Therefore, (2.40) will have a 21T-periodic solution in a sufficiently small neighborhood of r for A in a sufficiently small neighborhood of zero if and only if (0:,X) satisfy Equation (2.48b) with z replaced by Z*(o:,A); that is, if and only if (o:,A) satisfy the bifurcation equation ~, '2
(2.49)
F(O:'A)d:~,fJ01rP[-Ali:*«x'A) -
.
AlP
The function F(o:,A) satisfiesF(o:,O) = 0 and
of,
~.
+ A2fa + G(o,Z*(o:,A»]/71 = o.
PERIODIC SYSTEMS WITH A SMALL PARAMETER
(2.50)
F(a,A) = -AI
293
+ h(a)A2 + h.o.t.
where h(a) is given in (H2) and h.o.t. designates terms which are 0(IAI2) as IAI-+O. For any A2 =1= 0, the Bifurcation Equation (2.50) is equivalent to (2.51)
Al +Q (a'A2,A A I ) =0 h(a)-A2
where Q(a,{3, A) has continuous derivatives up through order two and Q(a, (3, 0) = O. It remains to analyse the bifurcation equation (2.51). Finding all possible solutions of (2.51) in a neighborhood of A = 0 is equivalent to finding all possible solutions of the equation
(2.52)
H(a,{3,A2)~fh(a) - (3 + Q(a,{3,A2) = 0
for all a ER, (3 E R, and A2 in a small neighborhood of zero. For A2 = 0, the only possible solutions (ao,{3o) are those for which h(ao) = (3o. If ao is such that h'(ao) =1= 0, then the Implicit Function Theorem implies there is a o(ao,{3o) >0 and unique solution a*({3,A2) of (2.52) for 1{3 - (3o I, IA21 < o(ao, (3o), a*(fio,O) = ao· It h'(ao)·~ 0, then Hypothesis (H2) implies h"(ao) =1= O. Therefore, there is a o(ao,{3o) > 0 and a function a*({3,A2), a*({3o,O) = ao. such that
aH(a*({3,A2),{3,A2)/aa = 0 for 1{3 - (301,1A21 < o(ao,{3o) and a*({3,A2/ is unique in the region la - aol '''( o(ao,po). Thus, the function M({3,A2)~ H(a*({3,A2),{3,A2) is a maximum or a minimum of H(a,{3,A2) with respect to a for (3,A2 fixed. A few elementary calculations shows that M({3o,O) ~. 0, aM({3o,O)!a{3 = -1. Therefore, the Implicit Function Theorem implies there is a o({3o) and a unique function (3*(A2),{3*(0) = (3o, such that M({3*(A2),A2) = 0 for IA21 < o({3o). There are two simple solutions of (2.52) near ao on one side of the curve {3 = (3*(A2) and no solutions on the other. In terms of the original coordinates A, this implies there are two solutions of (2.51) near ao on one side of the curve AI. = {3*(A2)A2 and none on the other. The curve Al = {3*(A2)A2 is a bifurcation curve and is tangent to the line Al = (3oA2 at A = O. The above analysis can be applied to each of the points CXj in Hypothesis (H2). One can choose a 0 > 0 such that all solutions of (2.51) for 1a - CXji < 0', iA1< 0 are obtained with the argument above. The complement of these small regions 1a - aj 1< 0 in [0, 21T ]is compact and h'(a) =1= 0 in this set. A repeated application of the Impiicit Functioll Theorem now shows that no further bifurcation takes place and all solutions are obtained for 1AI, 1A21 < o. This shows there exist two 21T-periodic solutions in the sectors mentioned in the theorem. If the least period of f(t) is 11k, k ~ 1 an integer, then we can obtain other 21T-periodic solutions by replacing t by
ORDINARY DIFFERENTIAL EQUATIONS
294
t
+ m/k,
k = 0,1,2, ... ,k -
1. This completes the proof of the theorem.
EXERCISE 2.16. Let S be a sector in Theorem 2.4 which contains 21Tperiodic solutions. Suppose r is a continuous curve in S defined parametrically by Aj = Aj«(3), 0 ~ (3 ~ 1, A~«(3) + A~((3) = 0 implies (3 = 0 and let x(t,(3) be a 21T-periodic solution of (2.40) for Aj = Aj«(3) which is continuous in (3 for 0<(3 ~ 1. Prove that x(· ,(3) is continuous at (3 = 0 if and only if limp-+OAI ((3)/A2((3) exists. If this limit is
ho
then x(· ,(3)
~
p(. + ao) where ao is a solution of
h(ao) = ho .
Exercise 2.16 shows the difference between considering parameters independently and considering them varying only along some straight line in A-space. In fact, if one considers A0 = (A ~ ,A~) fixed in R 2 , puts -A = €A 0 and considers Equation (2.40) for € small, then the most interesting part of the qualitative behavior of the solutions is lost. EXERCISE 2.17. Generalize Theorem 2.4 to the case where h(a) has a finite number of extreme values aj on [0, 21T] with h" (aj) =1= O. EXERCISE 2.18. Generalize Theorem 2.4 to the equation
x+ g(x) = Adl (t)x +A2!2(t) VIll.3. Periodic Solutions of Perturbed Autonomous Equations
In this section, part of Theorem VI.4.1 is reproved by using the methods of this chapter rather than a coordinate system near a periodic orbit. Qonsider the equation (3.1)
x=J(x)
+ F(x, e),
whereJ: Rn~Rn, F: Rn+1_R n arecontinuous,f(x), F(x, e) have continuous first partial derivatives with respect to x and F(x, 0) = 0 for all x. If the system (3.2)
x=J(x)
has a nonconstant periodic solution u(t) of period w whose orbit is I{?"then the linear variational equation for u is ;~I (3.3)
iI =
A(t) y,
A(t) = oJ(u(t»,
ox
295
PERIODIC SYSTEMS WITH A SMALL PARAMETER
and this linear periodic system always has at least one characteristic multiplier equal to 1. THEOREM 3.1. If 1 is a simple multiplier of (3.3), then there are an eo =1= 0 and a neighborhood W of C(j such that equation (3.1) has a periodic solution u*(·, e) of period w*(e), O;;;i lei ;;;i eo, such that u*(·, 0) = u(·), w*(O) = w, u*(t, e) and w*(e) are continuous in t, e for tin R, O;;;i lei ;;;i eo. PROOF. Let f!jI co be the space of continuou&W-periodic n- vector functions wIth the supremum norm. For any real number {3, consider the transformation t = (1 (3)-r in (3.1). If x(t) = y(-r), then y satisfies the equation
+
dy d-r = (1
(3.4)
+ (3)[f(y) + F(y, e)].
If (3.4) has a solution in f!jI co, then (3.1) has a solution in u(-r) + z(-r), then z satisfies the equation
f!jI (1+P)co.
If y( T) =
dz d-r
- = A(-r)z + G(-r, z, e, (3),
(3.5)
G(-r, z, e, (3) = (1
+ (3)[f(u + z) + F(u + z, e)] -A(-r)z -f(u).
Since 1 is a simple characteristic multiplier of (3.3), it follows that 'Ii is a basis for the w-periodic solutions of (3.3) and there is an w-periodic row vector rp which is a basis for the w-periodic solutions of the system adjoint to (3.3).
f:
Also, it can be assumed that lifJ(t)12 dt = 1. We now apply the preceding theory to determine w-periodic solutions of (3.5). For any h in f!jIco, let y(h) From Section 2, for any h in dz d-r
=
f!jI co,
f!jI co -+ f!jI co
o
ifJ(t)h(t) dt.
we know that the equation
= A (-r)z + h(-r) - y(h)ifJ'(T)
has a unique solution vlth in vIt:
r
f!jI co
such that
fow
is a continuous linear operator. Let !F:
'Ii'(t)vlth(t) dt = 0 and
f!jI co-+ f!jI (I)
be defined by
!Fz = vltG(·, z, e, (3). In the usual way, one shows there are e1 > 0, (31 > 0, 81 > 0 such that !F is uniformly contracting on f!jI co(O ) = {z in f!jI co: liz II ;;;i 81}. Therefore,!F has a unique fixed point z*(e, (3) which is continuous in e, {3 and continuously differentiable in (3. Also z*(O, 0) = o. This fixed point z*(e, (3) is a solution of
296
ORDINARY DIFFERENTIAL EQUATIONS
the relation
(3.6)
dz dT
= A(T)Z + O(T, Z, e, fJ) -
, B(e, fJ)ifi (T),
B(e, fJ) = y(O(T, z*(e, fJ), e, fJ))·
The function B(e, fJ) is continuous in e, fJ and satisfies B(O, 0) = 0. Also, since z*(e, fJ) is continuously differentiable with respect to fJ, it follows that B(e, fJ) is continuously- differentiable with respect to fJ. Furthermore, one oan
f:
conclude that oB(e, fJ)/ofJ = ifi(T)U(T) dT for e = 0, fJ = 0. This last integral must be different from zero for otherwise the equation dz dT = A(T)Z
+ U(T)
would have a solution in f!} (J) • This is impossible since this equation always has the solution Z(T) = TU(T). Since B(O, 0) = and oB(O, O)/ofJ *0, the implicit function theorem implies the existence of a fJ = fJ(e) such that B(e, fJ(e)) = for lei ~ eo· Thus, the function z*(e, fJ(e)) is an w-periodic solution of (3.1) and the theorem is proved.
°
°
VIllA; Remarks and Suggestions for Further Study Poincare [2], in his famous treatise on celestial mechanics, was the first· to describe a systematic method for the determination of periodic solutions of differential equations containing a small parameter. In 1940, Cesari [3] gave a method for determining characteristio exponents for linear periodic systems which is in the spirit of the presentation given in this chapter although different in detail. The method of Cesari was extended by Hale [1] and Gambill and Hale [1] to apply to periodic solutions of nonlinear differential equations with small parameters. Further modifications of this basic method by Cesari [3] and [5] led to the presentation given in this chapter. Concurrent with this development, Friedrichs [1], Lewis [1] and Bass [1, 2] gave methods which are very closely related to the above. Exercise 1.3 is due to Reuter [1]; Exercises 1.5 and (6 to Hale [6]; Exercise 1.15 to Hale [7]; Exercise 1.17 to Bailey and Cesari [1]. The Hopf bifurcation theorem was first stated by Hopf [1] although the phenomena was known to marty people (see, for example, Poincare: [2], Minorsky[1], [2]). For a recent book on Hopf bifurcation, see Marsden and MacCracJ<;~'n [1]. The result is also known for equations with delays (see Hale [10]). Exercises 1.22 through 1.27 are in the spirit of bifurcation problems discussed from the general point of view. An extensive literature is available (see, for example, Andronov,
PERIODIC SYSTEMS WITH A SMALL PARAMETER
297
Leontovich, Gordon and Maier [1], Sotomayor [1], Newhouse [l]). Except for notational changes, Lemma 2.3 is due to Lewis [1]. A less general form of property (E) of Section 2 is given by Hale [7]. Theorem 2.3 was.stated by Lewis in [1]. Exercises 2.7 and 2.8 are special cases of results by Hale [2]. Exercise 2.9 is due to Lyapunov [1] for analytic systems. Exercise 2.10 is due to Bogoliubov and Sadovnikov [1] while Exercises 2.11 and 2.12 to Golomb [1]. The treatment ofDuffmg's equation in Section 2 follows Hale and Rodrigues [1]. For the case where damping is also considered, see Hale and Rodrigues [2]. The discussion of periodic solutions near a periodic orbit in Section 2 follows Hale and Taboas [1] (see also Loud [2]). For the case where hypothesis (HI) is violated, see Hale and Taboas [2]. For an abstract version of this problem, see Hale [II]. An interesting extension of the idea of periodicity as well as property (E) is contained in the theory of autosynartetic solutions of diiferentialequations of Lewis [2]. For small perturbations of nonlinear second order equations see Loud [1].
CHAPTER IX Alternative Problems for the Solution of Functional Equations
To motivate the discussion in this chapter, let us interpret the procedure of the previous chapter in a more general setting. Let t?l'T be the Banach space of T-periodic n-vector functions with the supremum norm; let A be an n X n matrix whose columns are in t?l'T; let L be the linear operator defined on continuously differentiable functions in t?l'T by (Lx)(t) = i(t) - A(t)x(t); let N: t?l'T-t?l'T be defined by (Nx)(t) = ef(t, x(t)) wherefis a continuously differentiable function of x and T-periodic in t. The problem of finding a solution of the differential equation i - A(t)x = ef(t, x)
in t?l'T is then equivalent to finding an x in t?l'T which is in the domain of L such that Lx = Nx. With P, Q, :f( defined as in relation (VIII.2.5) and Lemma VIII.2.I, the Lemma VII1.2.2 asserts that the equation Lx = Nx is equivalent to the equations (1)
(a)
x = Px + :f(J
(b)
QNx=O.
~ Q)Nx
Equations (1) have a distinct advantage over the original differential equation. In fact, for N "small" and a fixed Xo with PXo = xo, one can determine a solution x*(xo) of (Ia) such that Px*(xo) =.Xo. The.existence of a solution of Lx = Nx is then equivalent to the determination of an Xo such that QN:t*(xo) = O. We have referred to the equations for Xo as the determining or bifurcation equations, but, more appropriately, we are going to say that these equations represent an alternative problem for Lx = N x. It is the purpose of this chapter to determine larger classes of equations which are equivalent to Lx = N x for x in a Banach space and thelspecify alternative problems when N is not necessarily small. This general'approach is taken because the ideas are applicable to problems in fields other than 298
ALTERNATIVE PROBLEMS FOR SOLUTION OF FUNCTIONAL EQUATIONS
299
ordinary differential equations; for example, integral equations, functional differential equations and partial differential equations. However, our applications are confined to ordinary differential equations.
IX.1. Equivalent Equations If X, Z are Banach spaces and B is an operatoJlowhich takes a subset of X into Z, we let ~(B), 8l(B), %(B) denote the domain, range and null . space respectively of B. If E is a projection operator defined on a Banach space Z we denote 8l(E) by ZE andZE will always denote a subspace which is obtained through a projection operator E in this way. The symbol 1 will denote the identity. If L: ~(L) c X -'?- Z is a linear operator, then K is said to be a bounded right inverse of L if K is a bounded linear operator taking gf(L) onto ~(L) and LKz = z for z in 8l(L). Let X, Z be Banach spaces; let N: X -'?-Z be an operator which may be linear or nonlinear; let L: ~(L) eX -'?- Z be a linear operator which may have a nontrivial null space and may have range deficient in Z.
LEMMA 1.1. Suppose %(L) and 8l(L) admit projections, %(L) = X p , gf(L) = ZI-Q and suppose L has a bounded right inverse K with PK = O. The equation
Lx=Nx
(l.l)
is equivalent to the equations (1.2)
PROOF.
(1.3)
+ K(1 -Q)Nx,
(a)
x = Px
(b)
QNx=O.
Equation (1.1) is clearly equivalent to the equations (a)
(1 ---;- Q)(Lx - Nx) = 0,
(b)
Q(Lx-Nx) =0.
Since LX = (1 -Q)Z and Q is a projection, QL = O. Therefore, (1.3b) is equivalent to (1.2b) and (1.3a) is equivalent to Lx = (1 -Q)Nx. Since (1 -Q)Nx belongs to the range of Land K is a bounded right inverse of L, this latter equation is equivalent to x=xo+K(1 -Q)Nx where Xo is in %(L). But, PK = 0 implies Xo = Px and the lemma is proved. The condition P K = 0 in Lemma 1.1 is no restriction. In fact, if M is any bounded right inverse of L, then K ~ (1 - P)M is also a bounded right inverse and P K = O. Even with these few elementary remarks, one is in a position to state some alternative problems for (1.1). More specifically, if N were small
300
ORDINAR Y DIFFERENTIAL EQUATIONS
enough in some sphere so that the contraction principle is applicable to (1.2a) with Px = Xo fixed, then (1.2a) can be solved for an x*(xo) and an alternative problem for (1.1) is QNx*(xo) = O. In other words, one can fix an arbitrary element Xo of %(L), solve (1.2a) for x*(xo) and try to determine Xo so that (1.2b) is satisfied. The alternative problem has the same" dimension" as the null space of L. In many cases, this dimension is finite whereas the original equation Lx ~ N x is infinite dimensional. This result will be stated precisely in the next section, but now we want to obtain other sets of equations which are equivalent to (1.1) and at the same time permit the discussion of cases when N is not small. The idea is quite simple. If one wishes to apply the contraction principle to (1.2a), then K(J - Q)N must be small in somc sense. However, if N is large on the whole space, then one should be able to make the product small by choosing the projection operator Q so that fewer values of x are being considered. This is the idea which now will be made precise. The accompanying Fig. 1 is useful in visualizing the next lemma.
Figure IX.!.! LEMMA 1.2. Suppose P, Q, K are as in Lemma 1.1 and let S be any projection operator on X such that Xs c Bl(K), SP = o. The following conclusions are then valid:
+
. ;, (i) P = P S is a projection operator. (ii) The preimage in ZI-Q under K of X s , X I - P II ~(Li i~duces a projection .J: ZI-Q-+ZI-Q. If XS=KZJ(I-Q), let 1 -Q= (1 -J)(J -Q). Then Q: Z -+ Z is a projection and XI-pll ~(L) = K(ZI-Q)'
ALTERNATIVE PROBLEMS FOR SOLUTION OF FUNCTIONAL EQUATIONS
(iii)
For any x in
~(L),
(1.4)
x
(iv)
301
= K(l -Q)Lx + Px.
QL= LP.
PROOF. (i) Since fJl(S) c rJt(K) and P K = 0 it follows that PS = o. A direct computation now shows that P = P + S is a projection. (ii) K is a one-to-one map of ZI-Q onto XI-P' n ~(L) since KZl = KZ2 implies K(ZI - Z2) = 0 and 0 = LK(ZI - Z2) = Zl - Z2. If Zl, Z2 are the primages under K of X s , X I - P n ~(L), respectively, then Zl n Z2 = {O} and ZI-Q = Zl Et> Z2. If Zn E Z2, Zn ---+z as n ---+ 00, then KZn ---+Kz as n ---+ 00 since K is continuous. Furthermore, there are Xn E XI-P n ~(L) and x E ~(L) such that KZn = x n , Kz = x. Since Xn ---+x and XI-p is closed, we have x E X I - P . Thus x E XI-P n ~(L) and Z2 is closed. In the same way or using the fact that Xs is closed and K is continuous, one sees that Zl is closed. Therefore, a projection J is induced on ZI-Q. If we let Xs = KZJ(I-Q) and 1 - Q = (1 - J)(l - Q) and use the fact that (1 - Q)(l - J) = (1 - J), it is it is clear that 1 - Q is a projection. Since X s = K ZJ(I-Q), it is also obvious that XI-P n ~(L) = KZI-Q. (iii) For any x in ~(L), x = KLx + Px. Since Q is a projection operator,
(1.5)
x
= K(l - Q)Lx + KQLx + Px.
From property (ii), K(l - Q)Lx belongs to XI-S since X I - P c XI-S. Also, a direct computation shows that Q(l - Q) = J(l - Q) and, therefore, KQLx = KQ(l - Q)Lx belongs to Xs. Operating on (1.5) with S and using the fact that !:iP = 0, we obtain Sx = SKQLx = KQLx. Since P = P + S, this proves relation (iii). (iv) Applying L to (1.4) and using the fact that K is a right inverse for L, we have Lx = (1 - Q)Lx + LPx. Therefore, property (iv) is satisfied. This completes the proof of the lemma. LEMMA 1.3. Suppose X, Z are Banach spaces, L: ~(L) ·c X ---+ Z is a linear operator with rJt(L), .;V(L) admitting projections by 1 -Q, P, respectively, L has a bounded right inverse K, PK = 0, and P, Q are the operators defined in Lemma 1.2. For any operator N: X ---+ Z, the equation Lx - Nx = 0 has a solution if and only if (1.6)
(a)
x=Px+K(l:....Q)Nx,
(b)
Q(Lx-Nx)=O.
PROOF. With P, Q as in Lemma 1.2, the equation Lx - Nx = 0 is equivalent to the equations Q(Lx - Nx) = 0, (1 - Q)(Lx - Nx) = O. If (l-Q)(Lx-Nx) =0, then relation (1.4) implies that K(l-Q)Nx=
302
ORDINARY DIFFERENTIAL EQUATIONS
K(l - Q)Lx = (1 - p)x and, thus, (1.&) is satisfied. If Lx - Nx = 0, then (1.6b)- is automatically satisfied. Conversely, if (1.6a) is satisfied, then L(l - p)x = (1 -Q)Nx. Since relation (iv) of Lemma 1.2 is satisfied, this implies (l-Q)(Lx -Nx) =0. If (1.6b) is also satisfied, then Lx -Nx =0. This proves the lemma.
IX.!. A Generalization As we have seen in Lemma 1.3, the geometric Lemma 1.3 permits the establiShment of many sets of equations which are equivalent to equation (1.1). Some assumptions on the linear operator L were imposed in order to obtain these results. However, once the basic relations between the operators P, Qare obtained, one can give an abstract formulation of the basic processes involved. To do this, let X, Z be Banach spaces; let N: !'J(N) c X -+ Z be an operator which may be linear or nonlinear,; let L: !'J(L) c X _ Z be a linear operator and let F = L - N. A solution of Fx = 0 will be required to belong to !'J(L) n !'J(N). The following hypotheses are made:
There are projection operators P: X -X, Q: Z _ Z such thatQL = LP. There is a linear map K: ZI-Q-XI-P such that (i) K(l - Q)Lx = (1 - p)x, x in !'J(L), (ii) LK(l - Q)Nx= (1 - Q)Nx, x in !'J(N). Ha: All fixed points of the operator il = P + K(l -Q)N belong to !'J(L) n !'J(N). Hi: H2:
For the operators L, N satisfying the hypotheses of Lemma 1.3, it was demonstrated that the hypotheses Hi-Ha can be satisfied by a large class of operators P, Q. In fact, hypothesis Hi is just relation (iv) of Lemma 1.2 and the operators P, Q depend upon a rather arbitrary subspace of X. 'Hypothesis H2 (ii) corresponds to the existence of a right inverse of L. Hypothesis H2 (i) is relation (1.4) Lemma 1.2. HypothesesHs was automatically satisfied for the particular P, Q constructed and the bounded right inverse considered. If L, N are. as specified above and K, P, Qexist so that Hi-Ha are satisfied, then it is not true that P, Qcan be obtained by th~ construction of Lemma 1.2. In fact, in that construction it was assumed that L had a bounded right inverse and 9l(L), ,.tV(L) admitted projections. There is no way to deduce these properties from Hi-Ha. Of course, in the applications, Lemma 1.2 is a rather natural way to obtain P, Q.
of ,
LEMMA 2.1. If hypotheses Hi-Hs' are satisfied, then the equation Fx = 0 has a solution x in !'J(L) n !'J(N) if and only if
(2.1)
(a)
x=ilx~P:t+K(l-Q)Nx,
(b) QFx=O.
303
ALTERNATIVE PROBLEMS FOR THE SOLUTION OF FUNCTIONAL EQUATIONS
PROOF. The relation Fx = 0 implies tJFx = 0, (1 - tJ)Fx = O. Therefore, (1 -tJ)Lx = (1 -tJ)Nx and hypothesis H2 (i) implies
K(1 -tJ)Nx = K(1 -tJ)Lx = (1 - p)x. Thus, x = !:l.x. Conversely, suppose (2.1) is satisfied. If x = !:l.x, then (1 - p)x = K(1 - tJ)Nx. Hypothesis Ha implies x in ~(L) (") ~(N) and H2 (ii) implies that
L(1 - p)x = LK(1 -tJ)Nx= (1 -tJ)Nx. But this fact together with HI implies that (1 - tJ)Fx = O. By hypothesis, tJFx = 0 and the lemma is proved.
IX.S. Alternative Problems In addition to the hypotheses HI-Ha imposed on the operators P, K, L, N of the previous section, we also suppose ~(N) = X and
H4:
tJ,
There exist a constant }-' and a continuous nondecreasing function oc(p), 0 ~ p < 00 such that
IK(! -tJ)NXI-K(1 -tJ)NX21
~
OC(p)IXI-X21,
IK(1 -tJ)NXII
~
OC(p)IXII +}-' for lXII, IX21 <po
For any positive constants c, d with c < d, let (3.1)
where
(a)
V(c)
(b)
Y(x, c, d) = {x in X: Px = x, x in V(c), Ixl ~ d}:
=
{x in Xl>:
Ixl
~ c}.
x is a fixed element of V(c).
THEOREM 3.1. Suppose ~(N) = X, V(c), Y(x, c, d), c < d, are defined by (3.1), the hypotheses HI-Ha of the previous section and hypothesis H4 above are satisfied with oc(d) <1, oc(d) d < d - c -}-,. Then there exists a unique continuous function G: V(c)~X, Gx in Y(x, c, d) such that Gx satisfies the equation
x =!:l.x '!Jf Px + K(1 - tJ)Nx.
(3.2)
If there is an (3.3)
x in
V(c) such that
tJFGx=O,
then FGx = O. Conversely, if there is an x such that Fx = 0, Ixl IPxl ~ c, then x = GPx and x = Px is a solution of (3:3).
~
d,
804
ORDINARY DIFFERENTIAL EQUATIONS
PROOF. For any x in V(c) and x in X, let H(x, x) = x + K(J -Q)Nx. For any x in f/(x, c, d),PH(x, x) = x,
IH(x,
x)1
~
c+ a;(d)lxl + II- ~ c+ a;(d) d + II- < d,
and IH(xl' x) - H(X2 , x)1 ~
a;(d)lxl - x21.
Therefore, H(', x): f/(x, c, d) -+ f/(x, c,d) is a contraction and there is a unique fixed point Ox of H( . , x) in f/(x, c, d). The function 0 is obviously continuous and satisfies (3.2) since x = Px. If x satisfies (3.3), then Lemma 2.1 implies that FGx = O. Conversely, if x is a solution of Fx = 0, x = Px, Ixl ~ d: Ixl ~ c, then Lemma 2.1 implies that x = Ax and QFx = O. But the proof of the first part of the lemma showed that A had a unique fixed point with Px = X. Therefore, the solution x of Fx = 0 must satisfy x = Ox, x = Px. Since x must also satisfy QFx = 0, it follows that x satisfies (3.3). This completes the proof of the theorem. In the sense of Theorem 3.1, finding a solution x in V(c) of (3.3) is equivalent to ~ding a solution of equation (l.l~ with Ixl ~ d, IPxl ~ c. Therefore, we refer to equation (3.3) as an alternative problem for equation (1.1).
IX.4. Alternative Problems for Periodic Solutions In this section, we go into some detail on the construction and more detailed meaning of alternative problems in connection with the existence of periodic solutions of ordinary differential equations. Suppose g: (-00, (0) X On-+On is continuous, g(t, u) is locally lipschitzianin u, g(t, u) = g(t 27T, u) for all t e (~oo, (0), u e On. Our problem is the determination of 27T-periodic solutions of the equation
+
(4.1)
u
= g(t, u).
Let us reformulate this problem in terms of our previous notation. Let Y be the Banach space of continuous functions taking [0, 27T] into On with the uniform norm and let B: !,)(B) c: Y -+ Y be the linear operator whose domain !,)(B) = {y e Y such that if e Y} and By(t) = if(t), 0 ~ t ~ 27T, for ye !,)(B). Let M: Y -+ Y be the operator defined by (My)(t) = g(t, y(t», o ~ t ~27T, ye Y; Y -+On be the bounded linear operator defined by ry = y(O) - y(27T), ye Y. Finding a 27T-periodic solution of (4.1)i slfioW equivalent to solving the boundary value problem , ~: ..
r:
(4.2)
By = My,
ry=o,
ye Y.
ALTERNATIVE PROBLEMS FOR THE SOLUTION OF FUNCTIONAL EQUATIONS
305
If X = .;V(r)and L, N are the restrictions of B, M to X, then the boundary value problem (4.2) is equivalent to finding a solution ofthe equation
(4.3)
x E X.
Lx = Nx, Let P: X
~X
be the projection operator defined by 1 2"
Px =
(4.4)
f
-
27T
x(t) dt;
0
that is, Px is the mean value of the 27T-periodic function x in X. With this notation, .;V(L) = X p = {all constant functions in X}, 9t(L) = XI- P = {all 27T-periodic. functions with mean value zero}. Therefore, the operator Q in Section 1 is P. If Z E XI- P , then the equation Lx(t) = x(t) = z(t) has a solution in X and a unique solution with mean value zero which depends continuously upon z. If we designate this solution by Kz, then K: XI-P~XI-P, PK = 0, and K is a bounded right inverse of L. Any x in X has a Fourier series <XI
(4.5)
L
X -
akeikt ,
ak =
f 27T 1
2"
-
e-tktx(t) dt.
0
k~-<XI
For any given integer m > 0, let 8 m : X ~ X be defined by
L
8 mx =
ak e1.kt.
O
One easily checks that 8 m is a projection operator and Lemma 1.2 implies Pm, Qm of that lemma are given by Pm = P 8 m , Qm = Pm. Equation (4.3) is then equivalent to the equations
+
(a)
(4.6)
(b)
x=Pmx+K(I-Pm)Nrc, Pm(Lx - Nx) = 0,
for every integer m:?: 0. From the definitions of Pm, K and the form of x in (4.5),
K(I -:- Pm)x(t) =
L
~k etkt .
Ikl>1JI ~k
Recall that Parseval's relation implies that fo2" Ix(t)l2 dt = Lk~
L
Ikl >m
Iak 1< ( L lakl2 )11o ( L ~)1/2 < (f 2"lx(t)12 dt),11o (L \)110 k
-
Ikl >m k 2
Ikl >m
o
<
Ikl>m k
(27T)1/2 (
II
deC =y(m) X,
L -1 )110 Ixl
Ikl>m k 2
_ lakl2. Since <XI
306
ORDINARY DIFFERENTIAL EQUATIONS
it follows that the Fourier series for K(J - Pm)x(t) is absolutely convergent and IK(I- Pm)xl
or all x E X.'Note that y(m)-+O as m-+ 00 since the series Lkk-2 <00. Since g(t, u) is assumed to be locally lipschitzian in u, there is a constant v> 0 and a continuous nondecreasing function {3(p), 0
INxl - Nx21 <{3(p)I Xl-X21 INxll < {3(p)IXll + v for all IXll < p, IX21 < p. For any constant d > 0 and any m, it follows that
< y(m){3(d)l xl-X21
IK(I- Pm)(NXl -NX2)1
< y(m}{3(d)lxll + y(m)v 0 < c < d are arbitrary constants
IK(J - Pm)NXll
for all IXll < d, IX21 < d. Suppose is chosen so large that y(m){3(d)
< I,
y(m){3(d) d
+ y(m)v < d -
and m
c
The operator I:l. defined by (3.2) is then a contraction mapping of the set !/(x, c, d) in (3.1) into itself. This implies that the equation (4.7)
x=x+K(J -Pm)Nx'
has a solution Gx for every X E {x in X Pm : Ixl for (4.1) is then the equation (4.8)
< c}.
An alternative problem
Pm(L-N)Gx=O.
In words, the existence of this alternative problem implies the following. If (4.1) is to have a 21T-periodic solution u, then all of the Fourier coefficients of the function v(t} = u(t) - g(t, u(t» must be zero. Let u(t) ~ um(t) um(t) u~(t)
= =
+ u~(t),
L
ak eikt ,
L
ake ikt .
Ikl:<=;m Ikl>m
The above remarks imply there is always an integer m such that one can fix x = Um and iletermine Gx = ut(um) in such a way that the (;':,;Fourier series for v(t) contains only the harmonics eikt , Ikl < m (this is the'ineaning of the solution Gx of (4.6a». The alternative problem then involves ,the determination of um(t) in such a way that the remaining Fourier coefficients of v(t) vanish.
ALTERNATIVE PROBLEMS FOR THE SOLUTION OF FUNCTIONAL EQUATIONS 307 There are two serious questions remaining concerning the determination of periodic solutions of (4.1) by this method. First, the size of the finite dimensional problem cannot be determined a priori but involves very careful estimates of the norm of the operator K(1 - Fm). Also, the finite dimensional problem (4.8) involves a function G which is only implicitly known and trying to assert the actual existence of an x is extremely difficult. In fact, one cannot hope to solve (4.7) directly, but must use some approximation scheme. Let us write Gx = x fj where fj is in (1 - Fm)X and is given by fj = K(1 - Fm)NGx. It is impossible to determine fj, but if the existence of Gx is proved by the contraction principle, then one has obtained a priori estimates on fj as
+
(4.9) for some constant 8m . The alternative problem (4.8) will certainly have a solution if one can show that the equation (4.lO)
Fm( L -N)(x+ fj)
=0
has a solution x for every given fj which satisfies the bounds (4.9). To show this latter property, one would naturally look first at the approximate equation obtained by setting fj = 0; namely, (4.11)
Fm(L -N)x=O.
Equation (4.11)'is the mth Galerkin approximation for the solution of Lx = Nx. Methods for the determination of explicit solutions of (4.11) and for showing that (4.lO) has. a solution for every fj satisfying (4.9) would take us too far afield. The interested reader may consult the references for the practical applicability of this method.
IX.5.The Perron-Lettenmeyer Theorem
In this section, we illustrate how the techniques of Sections 1 and 3 can be used to give a proof of a theorem of Perron and Lettenmeyer concerning the number of analytic solutions of a system of linear differential equations of the form (5.1 )
j= 1, 2, ... , n,
where x = COI(Xl' ... ,xn ), N j = row(Njl, ... ,Njn ), the Njk(t) are analytic for It I ;;;; 8,8 >0, and each U1!is a nonnegative integer. THEOREM 5.1. If JL = n - LJ=lUj ~ 0, then there are at least JL linearly independent solutions of (5.1) which are analytic for It I ;;;; 8.
308
ORDINARY DIFFERENTIAL EQUATIONS
PROOF. Let P.l be the set of all scalar functions which are analytic for It I ~ S. If b is in P.l, then b(t) = I~~obmtm and we define Ibl = I~~olbmISm. The function 1'1 is a norm on P.l and P.l is a Banach space. For any nonnegative integer a, let lu : P.l-+P.l be the linear operator defined by
(5.2)
lu b(t) = tUdb(t)jdt.
It is clear that .A'(lu) = {constant functions in P.l}, Bl(lu)
= {c inP.l : c = tUb, bin P.l}.
If the projection operators p, qu on P.l are defined by a-I
qub= Ibmt m,
(5.3)
m~O
then p, 1 - qu are projections onto .A'(lu), 9l(lu), respectively. The right inverse ka of la on Bl(la) such that pka = 0 is given by ~ bm +l1 . ka(l - qa)b(t) =.t.... - - tm+!, m~O m+ 1
(5.4)
and Ika(l-qa)bl ~ Sl-al(l-qa)bl. This proves ka is continuous. Let X be the Banach space which is the cross product of n copies of P.l with Ixlx = maxj IXjl~, x = COI(Xl, ... , xn), Xj in P.l. We write Ixi for Ixlx since no confusion will arise. Let aI, ... , an be the nonnegative integers in (5.1), a = (aI, ... , an) and define operators La, P, Qa, K a , N by Lax = col(la, Xl, ... , lanxn),
(5.5)
Px = COI(PXI' ... , PXn), Qa X = col(qa, Xl, ... , qa nXn), Ka X = col(ka, Xl, ... , k an Xn), Nx = col(NIX, ... , N nX), where the la are defined by (5.2), the p, qU by (5.3), kaJ by (5.4), and NiX in (5.1). Equation (5.1) may now be written as j
(5.6)
j
Lax=Nx,
xinX,
and from Lemwa 1.1 is equivalent to the linear equations (5.7)
X = Px
+ Ka(l -Qu)Nx, xinX.
If the linear operator Ka(l - Qa)N were a contraction operator, then one could fix a constant n-vector Xo and determine a unique solution x* = x*(xo),
ALTERNATIVE PROBLEMS FOR THE SOLUTION OF FUNCTIONAL EQUATIONS
309
Px* = xo, of the equation x = Xo + Ka(l - Qa}Nx. Equations (5.7) would then have a solution if and only if Xo satisfied the equations QaNx*(xo} = O. It is clear that such an x*(xo} is linear in Xo. Thus there would be I::i=lO"j homogeneous linear equations for n unknown scalars (the components of xo) and, therefore n - LJ=lO"j solutions of (5.7). ~his explains the possible appearance of /-t in the statement of the theorem except for the fact that the operator Ka(J - Qa}N may not be contracting. On the other hand, there is a natural way to use Lemma l.2 to circumvent this difficulty. For any given nonnegative integer r, let P = P r+1 be the projection operator on X which takes each component of x in X into the polynomial of degree r consisting of the first (r + I) terms of its power series expansion. The operator Q in Lemma 1.2 is easily seen to be Qr+a, r 0" := col(r + 0"1, ••• , r + O"n}. Thus, Lemma l.3 implies that (5.6) is equivalent to the equations
+
(5.8)
x = Pr+lx
+ Ka(J -Qr+a}Nx,
Qr+a(La - N}x = O.
Furthermore from the definition of K a , it follows immediately that there is a f3 > 0 independent of r such that IKa(l -Qr+a}Nxl ~ f3lxl/(r + I} for all x in X. Consequently, the operator K(J - Qr+a}N is contracting for r sufficiently large. Finally, fixing an n-vector polynomialf (t) of degree r, one can determine a unique solution x* = x*(f) of x = f +Ka(J -Qr+a}Nx which is linear and continuous in f. Therefore, the equations (5.8) have a solution if and only if f satisfies Qr+a(La - N}x*(f) = o. These represent nr LJ=l O"j homogeneous linear equations for the n(r + I} coefficients ofthe polynomialf. Therefore there are always /-t = n- LJ= 10"j solutions. This proves the theorem. If all 0"1 -1 in (5.1), the equation is said to have a regular singular point at t = o. In this case, the above theorem says nothing.
+
XI.6. Remarks and Suggestions for Further Study For the case in which P = P, Q= Q and N in (1.1) is small and has a small lipschitz constant, Theorem 3.1 has appeared either explicitly or implicitly in many papers; in particular, Cesari [5,6], Cronin [1], Bartle [1], Graves [1], Nirenberg [1], Vainberg and Tregonin [1], Antosiewicz [1]. However, Cesari [5, 6] injected a significant new idea when he observed that finite dimensional alternative problems could always be associated with certain types of equations (1.1) even when the nonlinearities N are not small. The construction of P, Q given in Section 1 follows Bancroft, Hale and
310
ORDINARY DIFFERENTIAL EQUATIONS
Sweet [1] and was motivated by Cesari [6]. The abstract formulation in Section 2 was independently discovered by Locker [1] proceeding along the lines of Section 4. Cesari has applied the methods of Section 4 to prove in [5] the existence and bound of a 21T-periodic solution of
x+x3 =sint by using the second Galerkin approximation x = a sin t + b sin 3t and to show in [6] that the boundary value problem
x+x+ cxx3=~t, x(O) = 0,
0 ~ t ~ 1, x(l)
+ i(l) =
0,
has a solution by using the first Galerkin approximation. Borges [1] has applied the same process to obtain the existence and bounds for periodic solutions of nonlinear (periodic or autonomous) second order differential equations by using only the first Galerkin procedure. Knobloch [1, 2] has also used the method" for existence of periodic solutions of second order equations. The papers of Urabe [2] discuss similar procedures for multipoint boundary value problems. Williams [1] has discovered interesting connections between Section 4 and the Leray-Schauder degree when the alternative problem is finite dimensional. When N is small enough to make K(J - Q)N a contraction operator on some subset of the basic space X, Theorem 3.1 is applicable to hyperbolic partial differential equations; see Cesari [7], Hale [9], Rabinowitz [1], Hall [1]. Cesari [8] has also obtained results for elliptic partial differential equations. For applications to functional differential equations, see Perell6 [1,2] and to integral.equations, see Vainberg and Tregonin [1]. The idea for the proof of the Perron-Lettenmeyer Theorem of Section 5 was communicated to the author by Sihuya. Amplifications of this idea appear in Harris, Sibuya and Weinberg [1]. A paper which is not unrelated to the approach of this chapter is the paper of McGarvey [1] on asymptotic solutions of linear equations with periodic coefficients.
CHAPTER X The Direct Method of Liapunov
In the previous chapters, we have repeatedly asserted the stability of certain solutions or sets of solutions of a differential equation. The proofs of these results in most cases were based upon an application of the variation of constants formula, and, as a consequence, the analysis was confined to a small neighborhood of the solution or set under discussion. In his famous memoir., Liapunov gave some very simple geometric theorems (generally referred to as the direct method of Liapunov) for deciding the stability or instability of an equilibrium point of a differential equation. The idea is a generalization of the" concept of energy and its power and usefulness lie in the fact that the decision is made by investig.ating the differential equation itself and not by finding solutions of the differential equation. These basic ideas of Liapunov have been exploited extensively with many books in the references being devoted entirely to this subject. The purpose of the present chapter is to give an introduction to some of the fundamental ideas and problems in this field. There is also one section devoted to a generalization known as the principle of Wazewski.
X.I. Sufficient Conditions for Stability and Instabilityin Autonomous Systems Let 0 c: Rn be an open set in Rn with 0 in O. A scalar function V(x), x in 0, is positive semidefinite on 0 if it is continuous on 0 and V(x) ~ 0, x in O. A scalar function V(x) is positive definite on 0 if it is positive semidefinite, V(O) = 0, V(x) > 0, x =J= O. A scalar function V(x) is negative semidefinite (negative definite) on 0 if - V(x) is positive semidefinite (positive definite) onO. The function V(XI' X2) = xi + x~ is positive definite on R2, V(XI' X2) = xi + x~ - x~ is positive definite on a sufficiently small strip about the Xl-axis, V(XI, X2) = x~ + xi/(l + xt) is positive definite on R2. In each of these cases, there is a co> 0 such that {(Xl, X2): V(XI, X2) = c} is a closed curve for every nonnegative constant c ~ co. Near (Xl, X2) = (0, 0), each of these functions have the qualitative properties depicted in Fig. 1.1. Any positive definite 311
312
ORDINARY DIFFERENTIAL EQUATIONS
function in Rn has some of these same qualitative properties, but a pl/ecise characterization of the level curves of V has not yet been solved. Howbver, the follawing lemma holds. LEMMA 1.1. If V is positive definite on n, then there is a neighborhood U of x = 0 and a constant Co > 0 such that any continuous curve from the
origin to
au must intersect the set {x:
PROOF.
V(x)
= c} provided that
0
~
c ~ co.
Let U be a bounded open neighborhood of 0 in U with (j c
n.
If l = minz in au V(x),
then l > O. Since V(O) = 0, V(x) ~ t for x in au, and V is continuous, it follows that the function V(x) must take on all values between 0 and l along any continuous curve going from 0 to au. LEMMA 1.2. If V(x) = V p(x) + W(x) where W(x) = o(lxl p ) as Ixl-+O is continuous and V p is a positive definite homogeneous polynomial of degree p, then V(x) is positive definite in a neighborhood of x = O. PROOF.
If minlzl=l V p(x) = k, then k
= IxlpV(x/ Ixl) ~ klxl p there is a S(e) > 0 such
V p(x)
*
for x O. For any e > 0, Ixl < S(e). If e = k/2, then V(x)
if 0
> o. Therefore that IW(x) I < elxl p for
~ V p(x) -I W(x)l ~ klxl p - (~) Ixl p = (~)lxIP
< Ixl < S(k/2). This proves the lemma.
LEMMA 1.3. Suppose V p(x) is a homogeneous polynomial of degree p. If p is odd, then V p cannot be sign definite.
**
PROOF. If x = (Xl, ... , xn), Xl 0, let X = XlU. Then V p(x) = xfV p(u). For a given value of U for which V p(u) 0, say positive, the function V p(x) has the sign of xf. If p is odd, this implies V p(x) must change sign in every neighborhood of X = O. General criteria for determining the positive definiteness of an arbitrary homogeneous polynomial of degree p are not known, but for p = 2, we have LEMMA 104.
(Sylvester).
The quadratic form x'Ax = L~. ;=1 ai;xix;,
A' = A, is positive-definite if and only if det(~j,i,j
= 1, 2, ... ,8) > 0,
8=
1, 2, ... , n.
A proof of this lemma may be found in Bellman [2].
THE DmECT METHOD OF LIAPUNOV
(1.1)
313
Consider the differential equation i=f(x),
where f: Rn ~ Rn is continuous and satisfies enough smoothness conditions to ensure that a solution of (1.1) exists through any point, is unique arid depends continuously upon the initial data. For any scalar function V defined and continuous together with oVfox on an open set n of Rn, we define V = V(l.l) as (1.2)
V(x)
=
oV(x) f(x). ox
If x(t) is a solution of (1.1), then dV(x(t»fdt = V(x(t»; that is, V is the derivative of V along the solutions of (1.1). Notice that V can be computed directly from f (x) and, therefore, involves no integrations. The theory below also remains valid for scalar functions V defined and only continuous on an open set n of Rn provided that V is given as V(~)
-1
= lim
i" [V(x(k, ~»
h ... O+ 'h
- V(~)],
where x(k, ~) is the solution of (1.1) with initial value ~ at k = O. If Vis locally lipschitz continuous, then this latter definition can be shown to be equivalent to V(~)
-1
= lim
i" [V(~
h ... O+'h
+ kf(~» -
V(g)].
In the applications, it is sometimes necessary to consider functions V(x) which do not have continuous first partial derivatives at all points x. On the other hand, the functions V(x) are usually piecewise smooth with the sets of discontinuities in the derivatives of V occurring on surfaces of lower dimension than the basic space Rn. Since the proofs of the theorems below do not use the differentiability of V, they are stated without this hypothesis. However. in the majority of the applications, V willbe given by (1.2). THEOREM 1.1. If there is a continuous positive definite function V(x) on V;£ O,.then x = 0 is a solution of (1.1) and it is stable. If, in addition, V is negative definite on n, then the solution x = 0 is asymptotically stable.
n with
PROOF. Let B(r) be the ball in Rn of radius r with center at the origin. For 0 < e < r, 0 < k = minlzl=-e V(x). There is an r > 0 such that B(r) Suppose that 0 < 3 ;£ e is such that V(x) < k for Ixl ;£ 3. Such a 3 always exist since V(O) = 0 and V is continuous. If xo is in B(8), then the solution x(t) of (1.1) with x(O) = xo is in B(e) for t ~ 0 since V(x(t»;£ 0 implies V(x(t» ;£ V(xo), t ~ O. This prov'3s x = 0 is a solution and it is stable.
en.
314
ORDINARY DIFFERENTIAL EQUATIONS
v
---+1---+-++---%1
~'--------%2,
Figure X.I.I
Since x = 0 is stable, there is a bo > 0, H > 0, such that the solution x(t) exists and satisfies Ix(t)1 < H for t ;;;;; 0 provided that Ixol < bo. Also, for any e > 0, there is a 8(e) > 0 such that Ix(t)1 < e for t;;;;; 0, Ixol < 8(e). To prove asymptotic stability, it is sufficient to show there is a T >0 such that Ix(t)1 < 8(e) for t ;;;;; T, Ixol < bo . Suppose there is an Xo with Ixol < bo and Ix(t)1 ;;;;; 8(e) for t ;;;;; O. If y > 0 is such that V(x(t» ~ - y for 8(e) ~ x < H: t ;;;;; O,then V(x(t» '~ V(~o) - yt.
If f3 > 0, k satisfy 0 < f3 ~ Vex) ~ k for 8(e) ~rxl< H, choose T = (k - f3 )/y. For t;;;;; T, V(x(t»
.
I
Let r >0 be such that B(r) c: n. For any 0 < 8 ~ r, there is an Xo :# 0 iIi U () B(s) and thus V(xo) > O. Since V ;;;;; 0 in U () n, it follows that the solution x(t) with x (0) = Xo satisfies V(x(t» ~ V(xo) > 0 for x(t) in U () n. PROOF.
THE DIRECT METHOD OF LIAPUNOV If oc = min{V(x): x in U V(x(t))
(1
0 with V(x)
= V(xo) +
315
~ V(xo)}, then oc >0 and
fo V(x(t)) dt ~ V(xo) + oct t
for all t for which x(t) remains in U (1 O. If 0 0 is any bounded neighborhood of x = 0 which is contained in 0, this implies x(t) reaches 000 since V(x) is bounded on U (1 0 0 . This completes the proof of the theorem. Theorems 1.1 and 1.2 give an indication of a procedure for determining stability or instability of the equilibrium point of an autonomous equation without explicitly solving the equation. On the other hand, there is no general way for constructing the functions V and the ingenuity of the investigator must be used to its fullest extent. For linear systems (1.1), the following lemma is useful. In this lemma, Re A(A) designates the real parts of the eigenvalues of a matrix A. LEMMA 1.5.
Suppose A is a real n'X n matrix. The matrix equation A'B+ BA =-C
(1.3)
has a positive definite solution B for every positive definite matrix C if and only if Re A(A) <0. PROOF.
Consider the linear differential equation
i = Ax,
(1.4)
and the scalar function V(x) (1.5)
= x' Bx where B is a symmetric matrix.
V (1.4) (x)
= x'[A' B
+ BA]x.
If there exists a positive definite B such that B satisfies (1.3) with C positive definite, then Theorem 1.1 implies that all solutions of (1.4) approach zero as t -+ 00. Of course, this implies Re A(A) < O. Conversely, if Re A(A) < 0 and C is a positive definite matrix, define B
(1.6)
=
{Xl eA'tCeAt dt. o
The matrix B is well defined since there are positive constants K, oc such that leAtl ~ Ke- at , {'~ O. Furthermore, it is clear that B is positive definite. Also, A' B
+ BA = f"" -d (eA'tCeAt)dt o dt
=-C. This proves the lemma. From the proof of Lemma 1.5, it is clear that we have proved the following converse theorem of asymptotic stability for the linear system (1.4).
316
ORDINARY DIFFERENTIAL EQUATIONS
LEMMA 1.6. If the system (1.4) is asymptotically stable, then there is a positive definite-quadratic form whose derivative along the solutions of (1.4) is negative definite. Let us now apply Lemma 1.5 to the equation (1.7)
i=Ax+f(x)
where f has continuous first derivatives in Rn with f(O) = 0, of (O)/ox = o. If Re A(A) < 0 then Lemma 1.5 implies there is a positive definite matrix B such that A'B + BA = -1. Let V(x) = x'Bx. Then V = V(l.7) is given by V= -x'x+2x'Bf(x).
Lemma 1.2 implies that -.:.... V is positive definite in a neighborhood of x = 0 and Theorem 1.1 implies the solution x = 0 of (1.7) is asymptotically stable. This is the same result as obtained in Chapter III using the variation of constants formula. If Re A(A) =1= 0 and an eigenvalue of A has a positive real part, then we can assume without loss of generality that A = diag(A_, A+) where Re A(A-) < 0, Re A(A+) >0. Let Bl be the positive definite solutionqf A'-Bl + BJA- = ~1 and B2 be the positive definite solution of ( -A )B2 + B2( -A = -1 which are guaranteed by Lemma 1.5. If x = (u, v) where u, v have the same dimensions as B 1 , B 2 , respectively, let V(x) = -u' BIU + v' B2 v. Then V(x) = V(l.7)(X) = x'x 0(lxI2) as Ixl-O. Lemma 1.2 implies V(x) is positive definite in a neighborhood of x = O. On the other hand, the region U where V is positive obviously satisfies the conditions of Theorem 1.2. Thus, the solution x = e of (1.7) is unstable if there is an eigenvalue of A with a positive real part. This result was also obtained in Chapter III using the variation of constants fomlUla. The above results are easily generalized. To simplify the presentation, we say a scalar function V is a Liapunov function on an open set G in Rn if V is continuous on 0, the closure of G, and V(x) = [oV(x)/ox]f(x) ~ 0 for x in G. Let
+
+)
+
8
= {x in 0:
V(x)
= O},
and let M be the largest invariant set of (1.1) in 8. THEOREM 1.3. If V is a L,iapunov function on G and y+(xo) is a bounded orbit of (1.1) which lies in G, then the w-limit set of y+ belongs to M; that is, x(t, xo)_M as t;+ 00. PROOF. Since y+(xo) is bounded, V(x(t, xo)) is bounded ~ow for t ~ 0 and V(x(t, xo)) ~ 0 implies V(x(t, xo)) is nonincreasing. therefore, V(x(t, xo)) _a constant cas t_ 00 and continuity of ViIpplies V(y) = c for
THE DIRECT METHOD OF LIAPUNOV
317
any y in w(y+). Since w(y+) is invariant, V(x(t, y)) = c for all t and y in w(y+). Therefore, w(y+) belongs to S. This proves the theorem. COROLLARY l.l. If V is a Liapunov function on G = {x in Rn: V(x) < p} and G is bounded, then every solution of (l.I) with initial value in G approaches M as t ~ 00. COROLLARY l.2. If V(x) -7" 00 as Ixl ~ 00 and V ~ 0 on Rn, then every solution of (1.1) is bounded and approaches the largest invariant set M of (1.1) in the set where V = O. In particular, if M = {O}, then the solution x = 0 is globally asymptotically stable.
PROOF. For any constant p, V is a Liapunov function on the bounded set G = {x: V(x) < p}. Furthermore, for any Xo in Rn there is a p such that Xo belongs to G. Corollary l.1 therefore implies the result. Notice that V < 0 in G\{O} implies M = {O} and one obtains from Theorem l.3 the asymptotic stability theorem in Theorem l.l. THEOREM 1.4. Suppose x = 0 is an equilibrium point of (l.I) contained in the closure of an open set U and let n be a neighbcyhood of x = O. Assume that (i) V is a Liapunov function on G = Un n, (ii) M n G is either the empty set or zero, (iii) V(x) < TJ on G when x =I- 0, (iv) V(O) = TJ and V(x) = TJ when x is on that part of the boundary of G in n. Then x = 0 is uristable. More precisely, if no is a bounded neighborhood of zero with contained in n, then any solution with initial value in (U n no)\{O} leaves no in finite time.
no
PROOF. If Xo is in (U n no)\{O}, then V(xo) < TJ. Furthermore, V ~ 0 implies V(x(t, xo)) ~ V(xo) < TJ for all t ~ O. If x(t, xo) does not leave no, then w(y+(xo)) is not empty and in As in the previous proof, w(y+(xo)) c: M. Since w(y+(xo)) is nonempty, this implies w(y+(xo)) = {O} and belongs to the set where V(x) = TJ. This contradicts the fact that V(x(t, xo)) ~ V(xo) < TJ and proves the theorem.
no.
Example 1.1. (l.8)
Consider the van der Pol equation x
+ e(x2 -1)x + x = 0,
and its equivalent Lienard form (l.9)
x=y_e(~3 Y=
-x.
-x),
818
ORDINARY DIFFERENTIAL EQUATIONS
In Chapter II, it was shown that this equation has a unique asymptotically stable limit cycle for every 8 > O. The exact location of this cycle in the (x, 1/)-plane is extremely difficult to obtain but the above theory allows one to determine a region near (0, 0) in which the limit cycle cannot lie.' S~ch a region can be found by determining the stability region of the solution x = 0 of (1.8) with t replaced by -to This has the same effect as taking 8 < O. Therefore, suppose 8 < 0 and let V(x, 1/) be the total energy of (1.9); that is, V(x, 1/) = (x2 + 1/2)/2. Then
V(x, 1/) = _8X2(~2
-1)
and V(x, 1/) ~ 0 if x 2 < 3. Consider the region G = {(i, 1/): V(Xj 1/) < 3/2}. It is clear that G is bounded and V is a Liapunov function on G. Furthermore, S = {(x, 1/): V = O} = {(O, 1/), < 3}. Also, from (1.9), M == {(O, O)} Qnd Corollary 1.1 implies every solution starting in the circle x 2 + 1/2 < 3 Ej.pproaches zero as t_ 00. Firially, the limit cycle of (1.8) for 8 >0 must be ootside,this circle. '
y2
Example 1.2.
Consider the equation
x +I(x)x + k(x) = 0, where xk(x) > 0, x ::1= 0, I (x) > 0, x ::1= 0 and H(x) = (1.10)'
f:
k(s) ds _00 as Ixl- 00.
Equation (1.lO) is equivalent to the system (1.11)
x=1/,
'Ii =
-k(x) -
I (x)1/.
+
Let V(x, 1/) be the total energy of the system; that is, V(x, 1/) = 1/*/2 H(x). Then V(x, 1/) = -I (X)1/2 ~ O. For any p, the function V is a Liapun~v function on the bounded set G = {(x, 1/): V(x, 1/)
Use Theorems 1.1 and 1.2 to prove Lemmas V.1.2 and Consid~r
the second order system
x = 1/ - xl (x, 1/),
'Ii =
-x - 1/1 (x, 1/).
Discuss the stability properties of this system when I has a fixedr' EXERCISE 1.3.
' ~,
Consider the equation
x + ax + 2bx + 3x2 = 0,
a>O,b>O.
THE DIRECT METHOD OF LIAPUNOV
319
Determine the maximal region of asymptotic stability of the zero solution which can be obtained by using the total energy of the system as a Liapunov function. EXERCISE 1.4. F(O)
=
Consider the system
0, a> 0, c > 0, aF(y)jy
>c
x=
for y
y,
if =
z - ay, Z = -cx - F(y),
* 0 and f:
[F(g) - cgja] dg -+
00
as IYI-+ 00. If F(y) = ky where k >,cja, verify that the characteristic roots of the linear system have negative real parts. Show that the origin is asymptotically stable even when F is nonlinear. Hint: Choose Vas a quadratic form plus the term
f:
F(s)ds.
EXERCISE 1.5. Suppose there is a positive definite matrix Q such that J'(x)Q QJ(x) is negative definite for all x 0, where J(x) is the Jacobian matrix ofJ(x). Prove that the solution x = 0 of x = J(x), J(O) = 0, is globally asymptotically stable. Hint: Prove and make use of the fact that J(x) =
*
+
fol J(sx)x ds. EXERCISE 1.6. Suppose you have the function V = y 2 e- X defined in the whole (x, y)-plane and that relative to some differential equation 17= _y2V. Can one conclude anything about the solutions of the original differential equation? If not , what is the trouble? EXERCISE 1.7. Suppose h(x, y) is a positive definite function such that -+ 00 as x 2 + y2 -+ 00. Discuss the behavior in the phase plane of the solutions of the equations
h(x,y)
+y -
x=
ex
if =
ey - x--:- yh(x, y),
xh(x, y),
for all values of e in (- 00, (0).
+
Consider the n-dimensional system x = J(x) g(t) where for all x and Ig(t)j ~ M for all t. Find a sphere of sufficiently large radius so that all trajectories enter this sphere. Show this equation ,pas a T-periodic solution if g is T-periodic. If, in addition, (x - y)'[f(x) - J(y)] < 0 for all x y show there is a unique T-periodic solution. Hint: Use Brouwer's fixed point theorem. EXERCISE 1.8.
x'f(x) ~ -klxI 2, k
> 0,
*
EXERCISE 1.9. Suppose J, g are as in Exercise 1.8 except g(t) is almost periodic. Does the equation·x = J(x) g(t) have an almost periodic solution?
+
EXERCISE 1.10. Prove the zero solution of (1.7) is unstable if there is an eigenvalue of A with a positive real part even though some eigenvalues may have zero real parts.
320
ORDINARY DIFFERENTIAL EQUATIONS
X.I. Vircuits Containing Esaki Diodes In this section, we give. an example which illustrates many of the previous ideas. Consider the circuit shown in Fig. 2.1. The square box in this diagram represents an Esaki diode with the characteristic function f(v) representing the current flow as a function of the voltage drop v. Kirchoff's L
i = f(v)
R Figure X.2.1
. laws imply that the relation between the current i and voltage v are given by deC] • E R'$-V= L di dt= ($,V),
(2.1)
where E, R,O, L are positive constants and vf(v)
~
0 for all v.
LEMMA 2.1. If there is an A >0 such that xf(x) >E2/R for Ixl >A, then every solution of (2.1) is bounded. In fact, every solution is ultimately in a region bounded by a circle. PROOF. If W(i, v) = (Li2+ Ov 2)/2, then the derivative of Walong solutions of (2.1) is
Let Wo=[L(E/R)2+0A2]/2. If W(i,v»Wo, then either lil>E/R or Ivl >A. If Iii >E/R, then W < 0 and if Iii ~ E/R, Ivl >A, then
For i
W< -
[ Ri2. - Ei
= 0,
> A,
Ivl
+ -E2] = R
we have also
- [ Ri2 - E (i E - -) ] :::;; -Ri2 :::;; O~;.'
R
W < O.
Therefore,
-
W< 0
-j.",
in the region
THE DffiECT METHOD OF LIAPUNOV
321
W(i, v) > Wo. Since the region W < P is bounded for any p and W(i, v) _ 00 as Iii, Ivl- 00, it follows that every solution of (2.1) is bounded. This proves the lemma. The problem at hand is to find conditions on J and the parameters in (2.1) which will ensure that every solution of (2.1) approaches an equilibrium point as t _ 00. Let f'(v) = dJ(v)fdv.
LEMMA 2.2. If the conditions of Lemma 2.1 are satisfied andf'(v) > 0 for all v, then every solution of (2.1) approaches the unique equilibrium point of (2.1). PROOF. First of all, it is clear there is only one equilibrium point of (2.1) if f'(v) > 0 for all v. If 1 1 Q(i v) =_12+- V 2 (2.2) , 2L 20 then the derivative of Q along the solutions of (2.1) is (2.3)
Q= -(RL- 212 + f'0-2V2)
~
o.
Since Lemma 2.1 implies all solutions of (2.1) are bounded and Q= 0 only at the equilibrium point, it follows from Theorem 1.3 that the assertion of Lemma 2.2 is true. • The most interesting cases in the applications are when f' changes sign and, in fact, can take on values < -R-l so that equation (2.1) has three equilibrium points. However, iff' > -R-l (only one equilibrium point), then a limit cycle may appear unless there are other restrictions onJ. In fact, one can prove THEOREM 2.1. If -f' < R-l, maxv( -f'10) > Rf L, then there is a value of E such that equation (2.1) has at least one periodic orbit. PROOF. Choose E so that the equilibrium point (io, vo) is such that -f'(vo)/O > RIL. From Lemma 2.1, there is a circle n with center at (0,0) such that the trajectories of (2.1) cross n from the outside to the inside. If i = io + u, v = Vo + w, x = (u, w), then
_ [-RL-l A0-1
(2.4)
-L-l ] -J~O-1
'
where ... represents higher order terms in x andJ~ = f'(vo). The hypotheses of the theorem imply that the eigenvalues of A have positive real parts. Replacing t by -t in (2.4) has the same effect as replacing A by· -A, a matrix whose eigenvalues have negative real parts. Lemma 1.6 implies there exist a positive definite matrixB such that the derivative ofW(x) = x' Bxalong the solutions of (2.4) satisfies W(x) = -x'x o(lxI2) as Ixl-O. Returning to the original
+
322
ORDINARY DIFFERENTIAL EQUATIONS
time scale and using Lemma 1.2, one sees that the trajectories of (2.4) are crossing the ellipses x' Bx = c for c > 0 sufficiently small from the inside to the outside. The annulus bounded by one of these ellipses and the circle n contains a positive semiorbit of (2;1). The Poincare-Bendixson theorem implies the result. To obtain more information for the case when f' changes sign, observe first that system (2.1) can be written as L di = (JP dt Oi •
(2.5)
dv (JP -0-=-. dt Ov
where
.
(2.6)
P($, v)
Ri2 = E$. -""2 -
.
w
+ IV f(8) ds o
[2
= -~+ U(v) 2R
and (2.7)
U(v) = THEOREM 2.2.
(E -v)2 2R
If there is an A
~
+ IV0 f(8)ds. 0 such that vf (v)
~
0, vf (v)
> E2t R for
Ivl >A and l'(v)
R
-+->0 o ·L
(2.8)
for all v, then each solution of (2.1) approaches an equilibrium point of (2.1) as t-+ 00. PROOF. Consider the function 8 = Q is defined in (2.6) and
+ >"P where Q is defined in (2.2), P
(2.9) Some straighforward but tedious calculations show that tions of (2.1) is given by
P along the solu-
(2.10)
Thus, relation (2.3) and (2.10) imply that S =
S=
-[(R - >"L)L-2[2
Q+ >..P satisfies
+ {f' + >"0)0-2V2] ~ 0
THE DffiECT METHOD OF LIAPUNOV
323
f(v)
f(v)
--------~~------------~v
--------~~--------------~v
Figure X.2.2
by our choice of A. Furthermore S = 0 if and only if 1 = 0, V = 0; that is, only at the equilibrium points of (2.1). Since all solutions of (2.1) are bounded from Lemma 2.1, the conclusion of the theorem follows from Theorem 3.1. It is of interest to determine which equilibrium points of (2.1) are stable. Let 8 be defined as in the proof of Theorem 2.2. If A satisfies (2.9), then one can show that the extreme points of 8 are the equilibrium points of (2.1). Since S ~ 0, it follows from Theorems 1.1 and 1.2 that the stable equilibrium points of (2.1) are the minima of 8 and the unstable equilibrium points are the other extreme points of 8. LEMMA 2.3. If Asatisfies (2.9), A=1= 0, then the extreme points of 8(i, v) coincide with the extreme points of U(v) in the sense that the extreme points of 8 coincide with the solutions of 1 = 0, 8U18v = o. Every strict local minimum of U(v) represents a local minimum of 8 and hence a stable equilibrium point of (2.1). PROOF.
A few elementary calculations yields
88 8i
88
=
-(RL-l - A)1
+ 0-1 V,
-- = -R-l(RL-l - A)1 -
8v
f
,
0-1 V
A8U +. OV
If 8UI8v = 0 then R-l(E - v) =f(v). If 1 = 0, then i = R-l(E - v). Thus, 1 = 0, 8UI8v = 0 implies V = 0 and this in turn implies an extreme point of 8. If 88j8i = 0, 88jav = 0, then 1 = V = 0 and this in turn implies 8Ujav = O. This proves the first part of the lemma. . The proof of the second part of the lemma is left as an exercise and
324
ORDINARY DIFFERENTIAL EQUATIONS
involves showing that f)2U /&2> 0 implies the quadratic form f)2S f)i2
evaluat~d
f)2S
g~ + 2 f)i &
f)2S
gog!
+ &2 g~
at an equilibrium point is positive definite.
EXERCISE 2.1. Interpret the above results for I(v) having the shape shown in Figs. 2.2a,b.
X.3. Sufficiel),t Conditions for Stability in Nonautonomous Systems In this section, we state nonautonomous systems (3.1)
som~
extensions of the results in Section 1 to
i=l(t,x)
where I: [1', (0) X R"-+R", l' a constant, is smooth enough to ensure that solutions exists through every point (to, XII) in [1', (0) X R", are unique and depend continuously upon the initial data. Let Q be an open set in R" containing zero. A function V: [1', (0) X Q-+ R is said to be positive definite if V is continuous, V(t, 0) = 0, and there is a positive definite function W: Q -+R such that V(t, x) ~ W(x) fgr all (t, x) in [1', (0) X Q. V(t, x) is said to possess an infinitely'small upper bound if there is a positive definite function W(x) such that V(t, x) ~ W(x), (t, x) in [1', CQ) X Q. If V: [1', (0) X Q -+ R is continuous, we define the derivative V(t, g) of ,V along the solutions of (3.1) as V(t,
g) =
-1 lim - [V(t k-+O+
h
+ h, x(t + h, t, g» -
V(t, g)],
where x(t, a, g) is the solution of (3.1) passing through (a, g) E [1', (0) X R". If V(t, g) has continuous partial derivatives with respect to t, g, then V(t,
g) = f)V~ g) + f)V~i g) 1ft, g).
THEOREM 3.1. If V: [1', (0) X Q-+R is positive definite and V(t, x) ~ 0, then the solution x = 0 of (3.1) is stable. If, in addition, V has an infinitely small upper bound then the solution x = 0 is uniformly stable. If, furthermore, - V is positive definite; then the solution x = 0 of (3.1) is uniformly,asymptotically stable. ,
I-
PROOF. Since V(t, x) is positive definite, there exists a positive definite function W(x) such that V(t, x) ~ W(x) for all (t, x) E [1', (0) X Q .. Let B(r)
THE DIRECT METHOD OF LIAPUNOV
325
be the ball in Rn of radius r with center at the origin. There is an r > 0 such that B(r) c Q. For 0 < c; < r, 0 < k = minlxl~EW(x). For any to E [T, (0), let 0< S < c; be a number such that V(to, x) < k for Ixl ~ S. Such a S always exists since V(to, 0) = 0 and V is continuous. If Xo is in B(S), then the solution x(t) of (3.1) with x(to) = Xo is in B(c;) for t ~ to since V(t, x) ~ 0 implies V(t, x(t)) ~ V(to, xo), t ~ to. This proves stability of x = o. If, in addition, V has an infinitely small upper bound, then there is a positive definite function W1(x) such that V(t, x) ~ Wl(X) for all (t, x) E [T, (0) X Q. With c;, k as above, chooseS > owith W1(x) < k for Ixl ~ S. Then, for any to E [T, (0), Xo in B(S), the solution x(t) of (3.1), x(to) = xo, is in B(c;) for t ~ to since V(t, x(t)) ~ V(to, xo) ~ W1(xo) for all t. This proves uniform stability of x = o. If, furthermore, - V(t, x) is positive definite, then there is a positive definite function W 2(x) such that - V(t, x) ~ W 2 (x) for all (t, x) E [T, (0) X Q. The proof now proceeds in a manner similar to the proof of Theorem 1.1. Let R+ = [0, (0), V(t, x): R+ X Rn -?- R be continuous, G be any set in Rn and G be the closure of G; We say V is a Lia-punov function of (3.1) on G if (i) given x in G there is a neighborhood N of x such that V(t, x) is bounded from below for all t ~ 0 and all x in N n G. (ii) V(t, x) ~ - W(x) ~ 0 for (t, x) in R+ X G and W is continuous on G. If V is a Lyapunov function for (3.1) on G, we define E
=
{x in
G:
W(x)
= o}.
THEOREM 3.2. Let V be a Liapunov function for (3.1) on G and let x(t) be a solution of (3.1) which is bounded and °remains in G for t ~ to ~ o. (a) If for each pin G, there is a neighborhood N of p such that If(t, x)1 is bounded for all t ~ 0 and all x in N n G, then x(t) -?-E as t-?- 00. (b) If W has continuous first derivatives on G and W = (oWfox)f(t, x) is bounded from above (or from below) along the solution x(t), then x(t)-?-E as t -?- 00. PROOF. Let p be a finite positive limit point of x(t) and {tn} a sequence of real numbers, t n -?- 00 as n -?- 00 such that x(tn ) -?- p as n -?- 00. Conditions (i) and (ii) in the definition of a Liapunov function imply that V(tn, x(t n )) is nonincreasing and bounded below. Therefore, there is a constant c such that V(tn, x(t n )) -?- cas n -?- 00 and since V(t, x(t)) is nonincreasing, V(t, x(t)) -?-Il as o
t -?- 00. Also, V(t, x(t))
~
V(to, x(to))
-::-1: W(x(s)) ds and hence
( ' W(x(s)) ds to
<
00.
326
ORDINARY DIFFERENTIAL EQUATIONS
Part (a). Assume p is not in E. Let S > 0 be such that W(p) > 2S > o. There is an e > O·sl1ch that W(x) > S for x iIi S2e(P) = {x: Ix - pi < 2e}. Also e can be chosen so that S2e(P) eN, the neighborhood given in (a). If x(t)
remains in 8 2.(p) for all t ~ tl ~ to, then ft~ W(x(s» ds = 00 which is a contradiction. Since p is in the limit set of x(t), the only other possibility is that x(t) leaves and returns to .:s'2e(P) an infinite number of. times. Since 11(t, x)1 is bounded in S2e(P), th.is implies each time that x(t) returns to S2e(p), it must remain in S2e(P) at least a positive time T. Again, this implies fOO W(x(s» ds = 00 and a contradiction: Therefore W(p) = 0 and E contains
+
+
Jto
all limit points. Part. (b). Since
ft~ W(x(s) ds <
Example 3.1.
Consider the equation
and W(x(t» is bounded from above (or from below), it follows that W(x(t» ~o as t ~ 00. Since W is continuous W(p) = 0 and this proves .(b). (3.2) where p(t)
~
S > O. If V(x,
00
x=y, iJ = -x·- p(t)y, y) = (x2 + y2)/2, then V = _p(t)y2 ~ -Sy2,
and V is a Liapunov function on R2 with W(x, y) = -Sy2. Also, W ~ -2S(xy+ p(t)y2) ~ -2Sxy. Every solution of (3.2) is clearly bounded and, therefore, condition (b) of Theorem 3.2 is satisned. The set E is the x'-axis and Theorem 3.2 implies that each solution x(t), y(t) of (3.2) is such that y(t) ~O as t ~ 00. On the other hand, if p(t) = 2 et , then there is a solution of (3.2) given as x(t) = 1 e- t , y(t) = -e-t . Since the equation is linear, every point on the x-axis is a limit point of some solution. This shows that the above result is the best possible without further restrictions on p. Notice that the condition in (a) of Theorem 3.2 is not satisfied in Example 3.1 unless p(t) is bounded. A simple way to verify that a solution x(t) of (3.1) remains in G for t ~ to is given in the following lemma whose proof is left as an exercise.
+
+
LEMMA 3.1. Assume" that V(t, x) is a continuous scalar function on R+ X Rn and there are continuous scalar functions a(x), b(x) on Rn such that a(x) ~ V(t, x) ~b(x) for all (t, x) in R+ X Rn. For any real number p, let Ap = {x in Rn: a(x) < p} and let Go be a component of Ap. If G is the component of the set "Bp = {x in Rn: b(x) < p} contained in Go and V ~ O~bn Go, then any solution of (3.1) with x(to) in G remains in Go for all t ~ ~,. Notice that a(x) ~ 00 as Ixl ~ 00 in Lemma 3.1 implies boundedness of the solutions of (3.1).
THE DmECT METHOD OF LUPUNOV
327
X.4. The Converse Theorems for Asymptotic Stability In this section, we show that a type of converse of ';l'heorem 3.1 for uniform asymptotic stability is true. If V(t, x) is a continuous scalar function on R+ X Rn and x(t, to, xo) is the solution of (3.1) with x(to, to, xo) = xo, we define the derivative V(3.1)(t, g) along the solutions of (3;1) as in Section 3; namely, V(3.1)(t,
-1
g) = lim h~ [V(t + h, x(t + h, t, g)) -
V(t,~)] .
h-+O+
Our first converse theorem deals with the linear system since the essential ideas emerge without any technical difficulties. THEOREM 4.1. system
If A(t) is a continuous matrix on [0, (0) and the linear
(4.1)
i
= A(t)x
is uniformly asymptotically stable, then there are positive constants K, ex and a continuous scalar function V on R+ X Rn such that (4.2)
(a)
Ixl ~ V(t, x) ~ K lxi,
(b)
V(4.1)(t, x) ~ -ex V(t, x),
(c)
jV(t, x) - V(t, y)1 ~ Klx -YI,
for all tin R+, x, yin Rn. PROOF. From Theorem III.2.1, there are positive constants K, ex such that the solution x(t, to, xo) of (4.1) with x(to, to, xo) = Xo satisfies Ix(t, to '. xo)1 ~ Ke-cx(t-to) Ixol
(4.3) for all t
~
to ~
°
and all Xo in Rn. Define for t
(4.4)
V(t, xo)
~
0, Xo in Rn
= sup Ix(t + T, t, xo)jecxr. T~O
It is immediate from (4.3) that V(t, xo) is defined for all t, Xo and satisfies (4.2a). To verify (4.2c) observe that
IV(t, xo) -
V(t, yo)1 ~ sup Ix(t
+
T,
t, xo) - x(t
+
T,
T~O
= sup Ix(t + T, t, Xo -
yo)le cxr
T~O
=
V(t, Xo -yo) ~ Klxo -yol·
t, yo)je cxr
328
ORDINARY DIFFERENTIAL EQUATIONS
The proof of (4.2b) proceeds as follows: V(4.1)(t, xo)
= lim ~ [sup Ix(t + h .... O+
h
= lim ~ [sup Ix(t + h .... o+h
~ lim ~
h .... O+ h
~
=
+ h, t + h, xo)ieaT - sup Ix(t +
T,
t, xo)lea(T-h) - sup Ix(t +
T,
t, xo)lea(T-h) - suplx(t +
,,;0>;0
T;O>;O
-a; V(t,
xo).
t, xo)!e aT ]
T,
t, Xo)ieC:T]
T;O>;O
[sup Ix(t +
h
T,
T;O>;O
T;O>;h
-lim [1- sup h .... O+
T
T;O>;O
T,
t, xo)le aT ]
T;O>;O
Ix(t +
T,
t, xo)leaT(e- CXli -1)
It remains only to show that V(t, ;Co) is continuous. Notice that
+
IV(t + 8, Xo + Yo) - V(t, xo)! ~ IV(t + 8; Xo yo) - V(t +8, X(t+8, t, xo+yo))1 + IV(t +8, x(t+ 8, t, Xo + yo)) - V(t, Xo + yo)! + IV(t, Xo + Yo) - V(t, xo)l· The· fact that the first and third terms can be made small if 8, yare small follows from (4.2c). The continuity of the solution of (4.1) in t and an argumentsimilar to that used in proving that V(4.1\ existed shows the second term can be made small if s is small. This proves right continuity. Left continuity is an exercise for the reader. Theorem 4.1 is a converse theorem for exponential asymptotic stability. In fact, if there is a function V satisfying (4.2), and x(t) = x(t, to, xo) is a solution of (4.1), then V(t, x(t)) ~ e-a(t-to) V(to, xo) ~ Ke-a(t-to) Ixol.
The fact that V(t, x) ~ Ixl implies that x(t) satisfies (4.3). The basic elements of the proof of the above theorem are the estimate (4.3) and the lifting of the norm of x in the definition of V. If the lifting factor eaT had not been applied one would have only obtained V ~ O. Our next objective is to extend Theorem 4.1 to the nonlinear equation (3.1). We need LEMMA 4.1. Supposef(t, 0) = O. The solution x = 0 of(3.1) is ,niformly stable if and only if there exists a function p(r) with the following'pt-operties: (a) p(r) is defined, continuous and monotonically increasing in an interval 0 ~ r ~ rl,
(b)
p(O) =0,
THE DIRECT METHOD OF LlAPUNOV
(c)
329
For any x in Rn, Ixl < rl and any to ~ 0, the solution x(t, to, xo), = xo, of (3.1) satisfies
x(to, to, xo)
Ix(t, to, xo)1 ~ p(lxo/)
for
t ~ to.
PROOF. The sufficiency is obvious. To prove necessity, suppose £ > 0 is given and let ~(£) be the least upper bound of all numbers 8(£) occurring in the definition of uniform stability. Then, for any x in Rn, Ixl ~ b(£), and any to ~ 0, the solution x(t, to, xo) of (3.1) satisfies Ix(t, to, xo) ~ £ for t ~ to. For every 81 >~, there is an Xo in Rn, Ixol < 81 such that Ix(t, to, xo)1 exceeds £ for some value of t. Consequently, $(£) is nondecreasing, positive for £ > 0, and tends to zero as £ tends to zero. However, $(£) may be discon,tinuous. Now choose 8(£) continuous, monotonically increasing and such that 8(£) < $(£) for £ > o. Let p be the inverse function of b. For any Xo in Rn, Ixol ~ 8(£), there exists an £1 such that Ixol = 8(£1) and, thus, Ix(t, to, xo)1 ~ £1 = p(lxo/). This proves the lemma. LEMMA 4.2. The solution x = 0 of (3.1) is uniformly asymptotically stable if and only if there exist functions p(r), O(r) with p(r) satisfying the conditions of Lemma 4.1 and O(r) defined, continuous and monotonically decreasing in 0 ~ r < 00, O(r)~Oasr~ 00 such that, for any Xo in Rn, Ixol ~rl:, and every to ~ 0, the solution x(t, to, xo) of (3.1) satisfies Ix(t, to, xo)1 ~ p(lxo/)O(t - to),
The proof of this lemma is left as an exercise. THEOREM 4.2. Iff(t, 0)= 0, f(t, x) is locally lipschitzian in x uniformly with respect to t and the solution x = 0 of (3.1) is uniformly asymptotically stable, then there exist an rl > 0, K = K(rl) > 0, a positive definite function b(r), a positive function c(r) on 0 ~ r ~ rl and a scalar function V(t, x) defined and continuous for t ~ 0, x in Rn, Ixl ~ rl, such that (4.5)
(a)
Ixl ~ V(t, x) ~ b(lxl),
(b)
V(3.1)(t, x) ~ -c(lx/) V(t, x) ~ -Ixlc(/x/),
(c)
IV(t, x) -
V(t, y)1 ~ K Ix - yl,
for all t ~ 0, x, y in Rn, Ixl ~ rl, Iyl ~ rl. PROOF. From Lemma 4.2, there exist functions p(r), O(T), defined, continuous and positive on 0 ~ r ~ rl, 0 ~ T < 00, such that p(r) is strictly increasing in r, p(O) = 0, O(T) is strictly decreasing in T, O(T) ~O as T ~ 00 and forany x in lin, Ixl ~ rl, the solution x(t, to, xo) of (3.1) satisfies Ix(t, to, xo)1 ~ p(lxo/)O(t -to),
t ~ to ~ O.
330
ORDINARY DIFFERENTIAL EQUATIONS
One can also assume that 8(T) has a continuous negative derivative, and 8(0) = 1. From the .properties of 8(T), there exists a function ac(T) defined, continuously differentiable and positive on T > 0, ac(O) = 0, ac(7) strictly increasing, ac(T) _ 00 as T- 00 such that T~O.
Suppose q(T) is a bounded continuously differentiable function on 0 ~ T < CX), such that q(O) = 0, q(T) > 0, q'(T) < ac'(T) for T > 0, q' monotone decreasing, and define Vet, xo)
= sup Ix(t + T, t, xo)/ etl(T). T;;:;O
Tne function Vet, xo) is defined for t
~
0, Xo in Rn, Ixol .
Ixol ~ Vet, Xo) ~ p(lxo/) sup e-a(T)+tl(T)
~ rl
and satisfies
df
= p(/xo/) ~ b(lxo/).
T;;:;O
Furthermore, since there exists a continuous positive nondecreasing function per), 0 < r ~ rl such that p(/x/)e-a(T)+tl(T) ~ Ixl
for T ~ P(lx/),
it follows that Vet, xo)
=
sup
Ix(t
+ T, t,xo)letl(T).
O;>;T;>;P(lzol)
Consequently, for any h > 0, there is a Tt, 0 that is continuous in hand
T:
Vet + h, x(t + h, t, xo))
~ Tl ~ P(lx(t
+ h, t, xo)/) such
= Ix(t + h + Tt, t, xo)letl(T:).
If k + .,.: =
"'h, then Vet + h, x(t + h, t, xo)) = ~
Ix(t + Th, t, xo)letl(T_)e-tl(T_)etl(T_-h) Vet, xo)e-tl(T_)etl(T_-h).
Therefore,
= - Vet, XO)q'(TO) ~ - Vet, xo)q'(P(/xo/))
~ -c(lxoD Vet, xo) ,
. ~ -Ixol c( Ixol) ' :
which proves (4.5b). Since fin (3.1) is locally Lipschitzianin x unifqrily in t, for any ro > 0, there is a consta.nt L = L(ro) such that Ix(t, to, xo) - x(t, to, 1/0)1 ~ eL(Ho) Ixo - 1/01,
331
THE DffiECT METHOD OF LIAPUNOV
for all t ~ to and xo, Yo for which Ix(t, to, xo)i ~ ro, Ix(t, to, Yo)1 ~ ro. Choose ro = p(rl). Sin?e P(r) is nondecreasing In r, it follows that
IV(t, xo) -
V(t, Yo)1 ~
sup
Ix(t
+
T,
t, xo) - x(t
+
T,
t, yo)le(l(T)
O~T~P(r)
~[
sup eLTe(l(T)] Ixo - yol O~T~P(r)
~ K(r) Ixo - yol, for Ixol, Iyol ~ r. For r = rl, K = K(rl), this proves V satisfies the inequalities of the theorem.. To prove V(t, xo) is continuous in t, Xo we observe that
IV(t + h, Xo + Yo) - V(t, xo)1 ~ IV(t + h, Xo + Yo) - V(t + h, x(t + h, t, Xo + Yo))1 + IV(t + h, x(t + h, t, Xo + Yo)) - V(t, Xo + Yo)1 + IV(t, Xo + yo) - V(t, xo)l· The fact that the first and third terms can be made small if h, Yo are small follows from the lipschitzian property of V and the continuity of the solution of (3.1). The second term can be seen to be small if h is small by an argument similar to that used in showing that V existed. This shows right continuity. Left continuity is an exercise for the reader. In Section 2, it was shown that a function V satisfying (4.5) implies that the solution X= 0 of (3.1) is uniformly asymptotically' stable. Therefore, Theorem 4.2 is the converse theorem for uniform asymptotic stability.
X.5. Implications of Asymptotic Stability In most of the applications that arise in the real world, the differential equations are not known exactly. Therefore, specific properties of the solutions of these equations which are physically interesting should be insensitive to perturbations in the vector field with, of course, the types of perturbations being dictated by the problem at hand. Therefore, one reasonable criterion on which to judge the usefulness of a definition of stability is to test the sensitiveness of this property to perturbations of the vector field. In this section, we use the converse theorem of Section 4 to discuss the sensitiveness of uniform asympto~ic stability to arbitrary perturbations as long as they are "small." We need LEMMA 5.1. Suppose f satisfies the conditions of Theorem 4.2, V is the function given in that theorem, g(t, x) is any continuous function on R+ X R1I
332
ORDINARY DIFFERENTIAL EQUATIONS
intoR1I and consider the equation i =f(t, x) + g(t, x).
(5.1) Then (5.2)
V(5.1)(t, x) ~
-c(lxl) V(t, x) + Klg(t, x)1
for all t ~ 0, Ixl ;;;; rl. PROOF. Let x*(t, to, xo), x*(to, to, xo) = xo, be a solution of (5.1) and x(t, to, xo), x(to, to, xo) == xo, a solution of (3.1). Using the definition of V(5.1) and relations (4.5b), (4.5c), one obtains
-1 V(5.1)(t, xo) = lim i" [V(t
+ n, x*(t + n, t, xo)) -
h ... O+ ,.
-1
= lim i" [V(t + n,x(t + n, t, xo)) -
V(t, xo) .
~~n
+ V(t + n, x*(t + n, t, xo)) ~ V(S.l)(t, xo)
+K
V(t
V(t, xo)]
+n, x(t + n, t, xo))]
-1 . Hm -I x*(t + n, t, xo) - x(t + n, t, xo)l
h ... O+
n
+ K Ig(t, xo}1 ;;;; -c(lxoI)V(t, xo) + Klg(t, xo)l· ;;;; V(U)(t, xo)
This proves the lemma. Many results now follow directly from Lemma 5.1. In fact, if (3.1) is linear, then equation (5.1) is (5.3)
i = A(t)x
+ g(t, x),
and Theorem 4.1 states that c(lxl) in (4.5) can be chosen as the inequality (5.2) becomes (5.4)
V(5.S)(t, x) ~ -oc V(t, x)
+K
IX> O. Therefore,·
Ig(t, ~)I.
If w(t, r) is a continuous function on R+ X R+, nondecreasing in r, such that (5.5)
Ig(t,
x)1 ;;;; w(t, lxI),
then one sees that (4.2a) and (5.5) imply
(5.6)
V(5.S)
~ -oc V
+ Kw(t, V).
Using our basic result on differential inequalities in Section 1.6 and relation (4.2), one can state
THE DIRECT METHOD OF LIAPUNOV THEOREM 5.1. the equation
Suppose w(t, u) in (5.5) is such thatfor any to
(5.7)
'It
=
-ocu
333 ~ O,.uo ~
0,
+ Kw(t, u)
has a solution passing through to, Uo which is unique. If u(t, to, uo), u(to, to, uo) = uo, is a solution of (5.7) which exists for t ~ to and Uo ~ Klxol, Xo in Rn, Ixol ~ rl, then the solution x(t, to, xo), x(to, to, xo) = xo, of (5.3) satisfies Ix(t, to,
xo)1
~ u(t, to, uo),
Another application of Lemma 5.1 is THEOREM 5.2. If f satisfies the conditions of Theorem 4.2 and the solution x = of (3.1) is uniformly asymptotically stable, then there is an rl > such that for any e, < e < rl and any r2 , e < r2 ~ rl, there is as> and aT> such that for any to ~ and any Xo in Rn, e ~ Ixol ~ r2, the solution x(t, to, xo), x(to, to, xo) = Xo of (5.1) satisfies Ix(t, to, xo)! < e for t ~ to T provided that Ig(t, x)! < S for all t ~ 0, Ixl ~ rl.
° ° °
°
°
°
+
PROOF. The hypotheses of the theorem imply there is a function V(t, x) satisfying the conditions (4.5) of Theorem 4.2. Choose rl as in Theorem 4.2. Without loss of generality, we may assume the functions b(8), C(8) in (4.5) are nondecreasing. For any e, r2 as in the theorem, we know that e ~ Ixl ~ r2 implies c(e) > and e < V(t, x) < b(r2)' Furthermore, Lemma 5.1 implies
°
V(5.1)(t, x) ~ ~
-c(e) V(t, x) + Klg(t, x)1 -c(e)e + Klg(t, x)1
for all t ~ 0, e ~ Ixl ~ r2. Consequently, if x = x(t) represents a solution of (5.1) with x(to) = xo,' e ~ Ixol ~ r2 and K8 < ec(e)f2 with Ig(t, x)1 < 8 for all t ~ 0, Ixl ~ rl, then V(5.1)(t, x(t)) ~ -ec(e)/2 as long as e ~ Ix(t)! ~ r2. This clearly implies that Ix(to t, to, xo)1 < e for t ~ T > 2bh)/ec(e). This completes the proof of the theorem. One could continue to deduce other more specific results but Theorems 5.1 and 5.2 should indicate the usefulness of the converse theorems of Liapunov.
+
X.6. Wazewski's Principle The purpose of this section is to give an introduction to a procedure known as Wazewski's principle for determining the asymptotic behavior of the solutions of differential equations. This principle is an important extension of the method
334
ORDINARY DIFFERENTIAL EQUATIONS
of Liapunov functions. To motivate the procedure and to bring out some of the essential ideas, we begin with a very elementary example. Let us consider the scalar equation
x= x + [(t,x)
(6.1)
where [(t,x) is continuous, smooth enough to ensure the uniqueness of solutions and there is a continuous, nondecreasing function K(a), a ~ 0, constants kE [0,1), ao > 0, such that aK(a)::; ka 2
(6.2)
I[(t, x) I ;::; K(a)
for a ~ao for
t~O, Ixl ::;a.
The objective is to show that hypotheses (6.2) imply that Eq. (6.1) has a solution which exists and is bounded for t> 0. If V(x) = x 2 /2, then the derivative Vof V along the solutions of Eq. (6.1) satisfy V(x)
(6.3)
= x 2 + x[(t,x) ~(l- k)lxl 2
if Ixl ~ ao. Consequently, on the set Ixl = ao, the vector field behaves as depicted in the accompanying figure.
If x(t,xo)
Sr.
de~ignates
the solution of Eq. (6.1) satisfying x(O,xo) = xo, let
= {xo ER: IXol ::; ao and there is a T with X(T,Xo) = aol
S"!... = {xo E R :Ixol ::; ao and there is a T with X(T,Xo) = -aO} S*
=S+ US"!....
335
THE DIRECT METHOD OF LIAPUNOV
The set S* represents the initial values of all those solutions with initial value Xo in {Ixl :::; aO} which leave the region {Ixl :::; ao} at some time, the ones which leave at the top of the rectangle and S~ those that leave at the bottom of the rectangle. Clearly ao E S+, -ao E S~. It is also intuitively clear that S*=I= [-ao,ao] by the way some paths want to go to the top and some to the bottom. Let us make this precise. We prove first that are open in [-ao,ao]. If xo' E S*, let r be chosen so that Ix(r, xo) I = ao. Inequality (6.3) implies there is an e > 0 such that Ix(t,xo)1 > ao for r < t :::; r + e. Choose a neighborhood U of x(r + e,xo) such that y E U implies Iyl > ao. By continuous dependence on initial data, there is a neighborhood V of xo, V ~ [-ao,ao] of Xo such that, for every z in V, there is a fez) such that' the solution x(t,z) satisfies x(T(z),z) in U. In particular, Ix(f(z),z) I > aO' Consequently, there is a r(z) such that Ix(r(z),z) I = ao. This proves are open in [-ao,ao]. Obviously, S S~ = ep. Thus, [-ao, ao] \S * is not empty. This is equivalent to saying that there is an Xo E [-ao,ao] such that the solutionx(t,xo) exists and satisfies I x(t, xo) 1< ao for t ~ 0, as was to be shown. Let us now generalize these ideas to n-dimensions. Suppose n is an open set in R x R n , f: n .... R n is continuous and the solution cp(t,to,xo),cp(to,to, xo) = Xo, of the n-dimensional system
S+
S:
S:
+n
(6.4)
x = f(t, x)
depends continuously upon (t,to,xo) in its domain of definition. For notation, let P designate a representative point (t, x) of n, (cx(P),(3(P» designate the maximal interval of existence of the solution cp(t,P) through PiP(t,P) = (t,cp(t,P» be the point on the trajectory through P at time t, and for any interval I C (cx(P),(3(P», let CP(I,P) = {tlI(t,P) = (t,ep(t,P», tEl} be that part of the trajectory through P corresponding to t in I. Let wen be a fixed open set, W be the closure of w in n, and let ow denote the boundary of win n.
Definition 6.1. A point Po E ow is a point of egress from w with respect to Eq. (6.4) and the set n if there is a 8 > 0 such thatcp([to - 8,to),Po) C w. An egress point Po is a point of strict egress from w if there is a 8 > 0 'such that rf>«to,to + 8] ,Po)' c n\w. The set of points of egress is denoted by S and the points of strict egress by S *. Definition 6.2. If A C B are any two sets of a topological space and K:B -+ A is' continuous, K(P) = P for all P in A, then K is said to be a retraction from B to A and A is a retract of B.
With these definitions, we are in a position to prove the following result known as the principle of Wazewski.
336
ORDINARY DIFFERENTIAL EQUATIONS
THEOREM 6..1. If S = S* and there is a set Z C w U S such that Z n S is a retract of S but not a retract of Z, then there exists at least one point Po in Z\S such that"¥([to,I3(Po)),Po ) C w. PROOF. For any point Po in w for which there is aTE [to,I3(Po)] such that ~(T,PO) is not in W, there is a first time tpo for which"iP(tpo ,Po) is in S, ~(t,PO) is in w for t in [to, tpo ). The point «tp0 ,Po) is called the consequent of po. and denoted by Q:Po). The set of points in w for which a consequent exists is designated by G, the left shadow of S. Suppose now S = S* and define the map K:G U S ~ S, K(P) = C(P) for PEG, K(P) = P for PES. We prove K is continuous. If P E. w and C(P) = (tp,l/>(tp,p»), then S = S* implies there is a 5> such that~«tp - 5,tp),P) C w, ~«tp,tp + 5),P) C n\w. Since cf>(s,P) is continuous in (s,P), for any € > 0, there is an 1/ > such that Il/>(s . Q) - cf>(s,P) I < € for sE (tp - 5,tp + 5) if IQ - PI < 1/. Ihis clearly implies that C(Q) ~ C(P) if Q ~P. If Pis in S = S*, then one repeats the same type of argument to obtain that K is continuous. Since K is continuous, K is a retract of GUS into S. If the conclusion of the theorem is not true, then Z\S C G, the left shadow of S. Thus, Z C GUS. Since Z n S is a retract of S, there is a mapping H:S ~ Z n S such that H(P) = P if P is in zns. The mapHK:G us~zns is continuous, (HK)(P) = P if P is in Z n S. Thus GUS is a retract of Z n S. Since Z C GUS, the map HK:Z ~ Z n S is a retraction of Z onto Z n S. This contradiction proves the theorem. As an example, consider the second order system
°
°
x=f(t,x,y)
(6.5)
y=g(t,x,y)
for (t,x,y) En = {t ~ O,x,y in R}. Let w = {(t,x,y):t~O,lxl 0, b > 0 are fixed constants. Suppose -
(6.6)
xf(t,x,y»O
on Ixl =a, Iyl
yg(t,x, y) <0
on Ixl
For the set w, we have S = S* = {(t,x;y):t ~ 0, Ixl = a, Iyl < b}. Let Z = {(t,x,y):t = 0, y = 0, Ixl ~ a}. Then Z ns is clearly a retract of S. However, Z n S is not a retract of Z. Thus, Theorem 6.1 implies there is a solution of Eq. (6.5) which remains in w for t ~ 0. EXERCISE 6.~.· Generalize the previous example by replacing the conditions (6.6) onf,gby the following:
w = {(t,x,y):t~ 0, Ixl < u(t), Iyl < v(i), u, v continuous together with their first derivatives u' , v'}
THE DIRECT METHOD OF LIAPUNOV
337
x/(t, x,y) > U(t)U'(t)
on Ixl = u(t), Iyl
yg(t,x,y) < V(t)V'(t)
on Ixl < u(t), Iyl
= vet).
Give some examples of u, v,[,g which satisfy these conditions and interpret your results in these special cases. EXERCISE 6.2. Let HrR X R n -+ R, j =.1,2, ... ,m be continuous together with their first derivatives and let H/to, xo) be the derivative of Hj(t,x(t,x(t,to,xo» at (to,,xo) along the solutions of the Eq. (6.4). Define
w= {(t,x)inRn+~:Hj(t,x)
j= 1,2, ... ,m}
= {(t,x)inR n + 1:Hk(t,X),= O,Hj(t,x)-:;' 0,
The rk are called the faces of w. Suppose for each k (t,x) E rk, either
j
= 1,2, ... , m}.
= 1,2, ... , in, and each
o~
(b) (t,x) is not a point of egress. Let Lk = {(t,x) in rk satisfying (a)},Mk S=S* = Uk=ILk\Uk=IMk.
= {(t,x) in rk
satisfying (b)}. Pi:ove
ExERCISE 6.3. One can use the results in Exercise 6.2 to prove results on the asymptotic equivalence of systems. Consider the two systems (6.7)
Xj = fj(t)Xj + gj(t, x),
(6.8)
Yj=fj(t)Yj,
j
= 1,2, ... , n,
j= 1,2, ... ,no
Suppose g = (g1, ... ,gn),f= ([1, ... ,In) are continuous for t~ T,(t,x)En = {(t,x):T -:;. t < oo,x in Rn} and a unique solution of Eq. (6.7) exists through each point in n. If there exists a continuous function F: [T,oo) -+ [0,00) and constant K such that
K. ~
r
fj(s)ds,
f- F(t)exp (f;fj(S)ds) dt < 00, Ig(t,x)1
~
j= 1,2, ... ,n,
Ix IF(t) On n
then, for every solution y of Eq. (6.8), there is a solution x of Eq. (6.7) such that
ORDINARY DIFFERENTIAL EQUATIONS
338
limt .... [x(t) - y(t)J GO
= O.
EXERCISE 6.4. Consider the second order equation (6.9)
x+ !(t,x,x) = 0
where !(t, x, y) is continuous for t ~ 0, x, yin R n and a uniqueness result holds for the initial value problem for Eq. (6.9),
x· !(t,x,y)+ y. y> 0
for all t~ O,X inRn,y* 0 inRn.
Then there exists a family of solutions of Eq. (6.9) depending on at least n parameters such that, for any solution x in this family, there is a to > 0 such that x(t) • x(t) is nonincreasing for t ~ to. Hint: Let b> 0, Wb = {(t,x,y) in R2n+ 1 :R(t,x,y) O} with (to,O,O) not in Zb' The principle of Wazewski yields a solution such that x(t)x(t) < b for t> to. To show x(t)x(t) ~ 0, take a sequence bm -+ O.
X.7. Remarks and Suggestions for Further Study The main topic in the previous pages has been the application of the direct method of Liapunov to the determination of the stability of solutions of specific differential equations as well as the application of this method to the deduction of qualitative implications of the concept of stability. This chapter should serve only as an introduction and more extensive discussions may be found in Hahn [1], Krasovskii [1], LaSalle and Lefschetz [1], Halanay [1], Yoshizawa [2], Zubov [1]. The definition of Liapunov function used in Sections 1 and 3 may be found in LaSalle [2]. Theorem 1.1, Lemma 1.5 and Theorem 3.1 are due to Liapunov [1], Theorem 1.2 is essentially due to Cetaev [1], Exercise 1.5 to Hartman [2]. Section 2 is based on the paper of Moser [1]. The idea for the proofs of the converse theorems in Section 4 is due to Massera [1]. The basic idea of the direct method of Liapunov is applicable to systems described by functional differential equations as well as partial differential equations. For a discussion of functional differential equations as well as further references, see Krasovskii [1], Halanay [1], Hale [8], Cruz and Hale [1]. For partial differential equations, see Zubov [1], Infante and Slemrod [1]. The fonnulation of the topological principle in Section 6 was given by Wazewski [IJ a!though the idea had been used on examples before (see Ii,irtman [1 J for more details). Exercise 6.2 is due to Onuchic [1], Exercise- 6.3 to Onuchic\ [2J. Further extensions of the ideas of Wazewski and a more general theory of isolating blocks may be· found in Conley [1 J.
Appendix.
Almost Periodic Functions
In this appendix, we have assembled the information on almost periodic functions which is relevant for the discussion in the. text. Only the most elementary results are proved. More details may be found in the following books. [A.l] [A.2] [A.3] [A.4] [A.5] [A.6]
H. Bohr, Almost Periodic Functions, Chelsea, New York, 1947. A. S. Besicovitch, Almost Periodic Functions, Dover, New York, 1954. J. Favard, Fonctions presque-periodiques, Gauthier-Villars, Paris, 1933. A. M. Fink, Almost Periodic Differential Equations, Lecture Notes in Math., Vol. 377(1974), Springer-Verlag. B. M. Levitan, Almost Periodic Functions (Russian) Moscow, 1953. T. Yoshizawa, Stability Theory and the Existence of Periodic Solutions and Almost Periodic Solutions, Applied Mathematics Sciences, Vol. 14(1975), Springer-Verlag.
The books of Fink and Yoshizawa are very good presentations of the theory of almost periodic functions and differential equations. The presentation below follows these two books. Notation. ex,fj, ... will denote sequences {on}, {fjn}, ... in Rn. The notation fj C ex denotes fj is a subsequence of ex, ex + fj = {on + fjn}, -ex = {-on}. We say ex,fj are common subsequences of ex',fj' if on = ~(k)' fjn = ffn(k) for some function n(k), k = 1,2, .... If I,g are functions on R and ex is a sequence in R, th(!n Tcxf = g means limn-+-ool(t + on) exists and is equal to get) for all t E R. The type of convergence (uniform, pointwise, etc.) Will be specified when used. For' example, we will write Tal= g pointwise, Tal= guniformly, etc. All functions considered will be continuous and complex valued functions on R unless explicitly stated otherwise. Defmition 1. I is almost periodic if for every ex' there is an ex C ex' such that Ta I exists uniformly. If AP = if:1 is almost periodic}, III = suptll(t)1 if lEAP, then AP with . this norm is a normed linear space. It will be shown to be complete later. 339
ORDINARY DIFFERENTIAL EQUATIONS
340
LEMMA 1. Every periodic function is AP. PROOF. If T is the period of I and a ' is a sequence, then there exists an a C a ' such that a(mod T) = {an(mod T)} converges to a point Ii. Then TOI.l = I( + 6) uniformly. 0
LEMMA 2. Every almost periodic function is bounded. I
'
'
I
PROOF. If not, there is a sequence a 'C ,R such that I/(an) I -+- co as n -+- co. But lEAP implies there is an a C a ' such that TOI.l exists uniformly. In particular'/(an) converges which contradicts our supposition. THEOREM 1. (properties of almost periodic functions) (1) AP is an algebra (closed under addition, product and scalar multiplication). (2) If lEAP, F is uniformly continuous on the range of F, then Fo/E AP. (3) If lEAP, inftl/(t) I> 0, then 1IfEAP. (4) If l,gEAP, then 1/1,[, min (f,g), max(f,g) EAP. (5) AP is closed under uniform limits on R'. (6) (AP, 1 1) is a Banach space. (7) If lEAP and dfl dt is uniformly continuous on R, then dlldt E AP. 0
PROOF. We prove this theorem in detail since it is simple and illustrates the way subsequences are used in Definition 1. (1) If t.g E AP and a" CR, there exists a ' C a" such that Tai:! exists uniformly, 1'at'~ al exists uniformly for any complex number a. Also there exists a C a ' such that TaK exists uniformly. Thus, T 01.(1 + g) exists uniformly, Ta(lg) = (TJ)(TaK) exists uniformly. (2) If lEAP and F is uniformly continuous on the range of t. then, for any a ' C R, there e1C.ists a C a' such that Ta(Fo/) = FoTal uniformly. (3) If inft I/(t) I > 0, then liz is uniformly continuous on the range of I and 1IfEAP. (4) The same argument proves 1/1.1E AP if lEAP by using the function F(z) = Izi ,z, respectively. Since min(l,g) =
~ [(1+ g) -I/-gl],
max(/,g) =
~ [(1+ g) + 1/- gl],
the remainder of (4) is proved by using (1). (5) Suppose {In} CAP, I continuous, lIn - II-+-O as n -+- co. For any a' 13' :> :> :> such that T({'In exists unif.0rmly, n = 1,2, .... the diagonalization process to obtain'l3 C a ' such tlJit TpJn
Use
0
0
0
rr
0
0
0
exists uniformly for all n = 1,2, . .. . ;~ .~.: For any e > 0, three is a ko(e) such that, for n = 1,2, ... ,k ~ ko(e), t E R I/(t + 13n) - Ik(t + /3n) I < e13.
APPENDIX. ALMOST PERIODIC FUNCTIONS For a fIxed k
~
341
ko(e) choose N(k, e) such that, for n, m ~N(k,e), t ER
IA(t + ~n) - A(t + ~m)1 < e13. Then, for
n,m~N(k,e),
tER,
I/(t + ~n) - I(t + ~m) I < I/(t + ~n) - A(t + ~n) I
+ IA(t + ~n) - A(t + ~m)1 + IA(t + ~m) - I(t + ~)I <e. Thus, T pI exists uniformly and lEAP. (6) This is obvious from (1) and (5). (7) If In(t) = n[/(t + lin) - I(t)] , then the mean value theorem implies
!'(t) - In(t) = !'(t) - !'(t + O(t, n» for some t ~ O(t, n) ~ t + lin. Since !'(t) is uniformly continuous, we have In(t)"""* !'(t) uniformly onR. Since each In EAP, we have from (5) that I' EAP. This completes the proof of the theorem. The space AP contains all periodic functions. Theorem 1 shows that it contains all functions which are sums of periodic functions and thus all trigonometric polynomials. It also contains the uniform limit' of trigonometric polynomials. One can actually show (but the proof is not trivial) that AP consists precisely of functions which are uniform limits of trigonometric polynomials. To obtain other characterizations of almost periodic functions, we need the following concept.
DefInition 2. The hull of I,H(/), is defIned by
H(!) = {g: there is an a C R such that Tal = g uniformly} If lis periodic, then H(f) consists of all translates of I; that is,
= {fr:fr(t) = I(t + 1'), tER, l' ER} and thus, H(!) is a dosed curve in AP if I is periodic. If IE AP, then H(!) may contain elemeI}.ts which are not translates of I. In fact, if I(t) = cos t + cosV2t, H(!)
then lEAP since it is the sum of periodic functions, but is not periodic since it takes the value 2 only at t = O. Also, I(t) > -2 for all t and there is an a' = {~} C R such that I(~) ~ -2. If a C a' and Tal = guniformly, then g(O) = -2 and g cannot be a translate of I. THEOREM 2. lEAP if and only if H(!) is compact in the topology of uniform convergence on R. Furthermore, if lEAP then H(g) = H(!) for all
gEH(!).·
.
342
ORDINARY DIFFERENTIAL EQUATIONS
PROOF. If H(f) is compact, then for any 01.' C R .. there is an 01. C 01.' such that Til/exists uniformly and, thus/EAP. Conversely, if / E AP, {gn} C H(f), then there is an 01.' == {~}such that 1/(' + ~) - gnl < lin (we. have used the diagonalization process here). Also, there is an 01. C 01.' such that TIl /= g exists uniformly. If 01. = {a,,} = {OI.kn }, then 1/(' + a,,) - gknl-+ 0 as n-+ 00. Thus,
Ig- gknl
~
Ig- /(. + a,,)1 + 1/(' + a,,) -gknl-+ 0
as n -+ 00. Thus, H(f) is compact. To prove the last part of the theorem, if g E H(f) and h E H(g); that is , there is an 01. ~ {a,,} C Rsuch that Tllg = h uniformly, then there is al3 = {13n} such that 11/(' + I3n) -g(' 4- ~)I< lin. Thus, T~= h andhEH(f);that is, H(g) ~ H(f). Conversely, if g E H(f), Til/ = g, then T-a.K = / by the change of variables t + Ofl = s. Thus, IE H(g), Since H(g) is compact, this implies H(f) f H(g), Finally, we have H(f) = H(g)aiJ.d the theorem is proved. To obtain another characterization of an almost periodic function, we need the· following· defInitions. Definition 3. A subset·S C.R is relatively dense if there is an L that [a, a + L] n S =F cp for any a ER.
> 0 such
DefInition 4. If / is any continuous complex valued function on R, the e-translation set of / is defmed as T(f,e)";' {r: I/(t + r) - /(t}1 DefIniti~n
<e
for tER}.
S. A continuous complex valued function/onRis Bohr almost
periodic if for every e> 0, T(f,e) is relatively dense.
Lemma 3. A Bohr almost periodic function is bounded and uniformly continuous. If I(t) is almost periodic, then for any 7J > 0, there exist an I in every interval [t -I, t] such that
PROOF.
and a
l'
I/(t) - I(t - 1')1
< 7J
for all t in (-00, (0).
If we let t' = t -1' then 0 ~ t' ~ I and I/(t) - l(t')1 < 7J. If 11(t')1 ~ B for o ~ t' ~ I then I/(t)1 ~ B + 7J for all t in (-00, (0). This proves boundedness. Suppose 7J, I are as above. Since 1 is continuous, it is uniformly continuous on [ -I, I] and we can find a S < I such that I/(t{) - l(t~)1
< 7J
if
It{ - t~1
< S, t{,t~ E [ -I, I].
For any tl, t2 with tl < t2, It1 -:- t21 < S, consider the interval [tl, tf + I]. Let l' be an almost period relative to.7J in this interval. Then t{·= tl -1', t~ = t2 - 1 ' are
APPENDIX. ALMOST PERIODIC FUNCTIONS
343
in '[ -l, l] and If(tl) - f(ti)1 < 7J, If(t2) - f(t 2)1 < 7J. The triangle inequality shows that If(tt) -f(t2)1 <37J and this proves uniform continuity. THEOREM 3.
The following statements are equivalent:
(i) [EAP (ii) H([)"is compact in the topology of uniform convergence on R. (iii) [is Bohr almost periodic. PROOF. We have already proved (i) ~ (ii). We next prove (iii) => (i). Suppose
a' CR. We first show that, for any e> 0, there is an a C a' such that I[(t + Ctn) - [(t
+ Ctm) 1< e
for all n,m
Let L be as in the definition of relatively dense for T(f, i), and choose 0 from Lemma 1 so that I[(t) - [(s) I < e if It - sl < 20. We may write the sequence a = {Ctn} as Ctn = {3n + 'Yn, (3n E ~([,i), 0:::; 'Yn :::;L, 'Yn ~ 'Y asn~oo, 'Y - 0 < 'Yn < 'Y + o. Then . I[(t + Ctn) - [(t + Ctm) I:::; I[(t + (3n - 13m + 'Yn - 'Ym) - [(t - 13m + 'Yn - 'Ym I
+ I[(t - 13m + 'Yn - 'Ym) - [(t + 'Yn - 'Ym I + I[(t + 'Yn - 'Ym) - [(t) I < e VtER. Now use the diagonalization process to get a" Cae a' such that Ta"[exists uniformly. We now prove (ii) => (iii). H(f) compact is equivalent to H(f) totally bounded is equivalent to {f(t + T):T E R} totally bounded is equivalent to, for any e > 0, there are aI, ... ,On and a function z"(t) from R to {I, ... ,n} such that I[(t + lLj(r) - [(t + T) I < e for all t, T. This is equivalent to
I[(t) - [(t + T -tzz·(r) 1< e If L = max IlLjl, then T - L :::; T - ai(r) :::; T T(f,e) is relatively dense with constant 2L. This completes the proof of the theorem.
for all t.
+ Land
T -
ai(r) E
T(f,e).Thus,
THEOREM 4. [EAP if and only if (iv) . For ~ny a',{3' C R, there are common subsequences a C a', {3 C {3' such that pointwise. PROOF. Suppose [E AP, a', {3' C R are given and choose {3" C {3' such that uniformly. Since g E AP, for a" C a' common with {3", there is an a" C a" such that Tam g = h uniformly. Let {3'" C {3" be common with alii. Then we can find common subsequences, a C a'II, {3 C {3'11 such that Ta+ p[ = k
Tf,' [ = g
ORDINARY DIFFERENTIAL EQUATIONS
344
uniformly and also note T(JI = g, TOig = h uniformly. If e > 0 is given, then, there is a no such .that, for all n 2: no and all t,
Ik(t) - I(t + an
+ ~n) 1< e, Ig(t) - I(t + ~n)1 < e, Ih(t) - get + an)1 < e
Thus, for n 2: no,
Ih(t) - k(t) I~ Ih(t) - get + an)1
+ Ig(t + an) - I(t + an + ~n)1 + I/(t+an +~n)-k(t)I<3e
Since e is arbitrary, h = k; i.e. TOl+(JI = TOlT(JI uniformly and, in particular, pointwise. To prove (iv) implies IE AP, suppose 'Y' CR. Then there is a'Y C 'Y' such that T'YI exists pointwise. In fact, for a' = {O}, ~' = 'Y', we know there are common subsequences a C a', 'Y C 'Y' such that TOl+'Y I = TOlT'YI pointwise. Since a = {O}, this proves the assertion. Suppose that T'YI does not exist uniformly. Then there exist a' C~,~' C 'Y, r' C R, e > 0
(*) Apply condition (iv) to a',T' to obtain common subsequences a" C a', T" C T' such that T,"+0l"1= TT" TOl"1 pointwise. Choose ~" C~' common to a",T" and apply (iv) to ~",T" to find ~ C ~", i C Til such that TT+(JI= TTT(Jlpointwise. Choose a + a" common to ~,T. Then T,+01 I = TT Ta I pointwise from above. But TOll = T'YI = T(JI pointwise, since a C 'Y, ~ C "y. Thus, TT+OlI = TTTOlI = TTT(JI = TT+(JI pointwise. At t = 0, this contradicts (*) since T + ~ C T' + ~'. T + aCT' + a'. Thus, the convergence is uniform. This completes the proof of the theorem. A remarkable consequence of Theorem 3 is that almost periodicity can be decided by discussing only pointwise convergence as in (iv). This is generally easier to check than uniform convergence on R. From Proposition 1 and Theorem 1, the space AP contains all trigonometric polynomials and all uniform limits of trigonometric polynomials. The converse is also true and is stated without proof. THEOREM 5. lEAP if and only if I can be uniformly approximated by trigonometric polynomials. It is natural to investigate the problem of the existence of a theory of trigonometric series for almost periodic functions analogous to the theory of Fourier series for periodic functions. The first fundamental result is THEOREM 6.
If lEAP, then the mean value
.
Ift+TI(s)ds
M[f] = limT-->00 T t
APPENDIX. ALMOST PERIODIC FUNCTIONS
345
exists, is independent of t and the convergence is uniform in t. If lEAP, then lexp(-iA') EAP for any real number A. Define
a(A,f) = M[/e-iA'l
= limT-->oo ~
f:
l(t)e-iAtdt.
The set
A(f) = {A E R :a (A,f) 1= O} is called the set of Fourier exponents of I, the numbers a(A,f), A E A(f), the Fourier coelficients and the Fourier series of lis designated by 'At
I(t) - ~AEA(j)a(A,f)eJ . Then, one can prove THEOREM 7. A(f) is denumerable. If l,g,E AP, then 1=:= g if and o_llly if A(f) = A(g) and a(A,f) = a(A,g) for A E A(f). If I - ~AEAi(j)a(A,f)exp(iAt),. then ~AEA(j)la(A,f)12 < 00. The usual operations on Fourier series are valid. For any sequence {An} C R, the module of {An} is the set consisting of all real numbers which are finite linear combinations of the {An} with integer coefficients. We say {OJ} C R is linear independent over the rationals if, for every N ~ I, 1rkOJk = 0 for each rk rational implies rk = 0, k = 1,2, ... ,N. The sequence {OJ} is said to be an integral base for {An} if {OJ} is linearly independent over the rationals and if each element of {An} is a finite linear combination of the {OJ} with integer coefficients, If IE AP, the module m[/l of I is the module of A(f). A function lEAP is called quasiperiodic if there exist an integer N ~ I, positive real numbers Tl, , , . ,TN and a function F on RN such that F(Xl' ... ,XN) is periodic in Xi of period Tj' j = 1,2, ... ,N such that I(t) = F(t, ... ,t).
'£f=
Example 1. If lis periodic of period 2rr/w, then m[/l and an integral base for m[f] in {w}.
= {nw,n= 0, ± I, ... }
If I(t) = cos t + cos 2 t + cos ..j2t, then A(f) = + m../2,n,m= O,±l, ... } and an integral base for m[f]
Example 2.
m[/l = {n
{I, 2, ..j2}, is {1,..j2}.
This function I is quasiperiodic with
l(t)=F(t,t),F(Xl,X2) = cos xl +COS2Xl +cos..j2x2' Example 3. If ~;= liOn 1< 00, On 1= 0 for all n, then
I(t) = ~;= lOne it /n is in AP, A(f) = {I/n, n = 1,2, ... }. What is m[/l and what is an integral base for m[f]?
ORDINARY DIFFERENTIAL EQUATIONS
346
The following result is basic· for differential equations. The proof is complicated and contained in Fink, p .. 61. In the statement of the theorem, "convergence in any sense'~ may be interpreted as "pointwise-convergence," "Ul'liform convergence on compact sets," ''uniform convergence" or "mean convergence"; that is.!" converges to lin the mean if M(l/n - 112) ~ 0 as n-' 00. THEOREM 8. For l,gEAP, the following statements are equivalent:
(i) (li) (iii) (iv) (v)
m[fJ:J mrg] for any e > 0, there is a·~ > 0 such that.T(f,cS)C T(g,e) Ta/existsimplies Ta.gexists(any sense} Tal = I implies Ta.g = g (any sense) Tal = I implies there exists 0/.' C 0/. such that T«g = g (any sense)~
In the study of almost periodic differential,equations,one encounters frequently the simple equation y =I(t)where lEAP, M[f] = O.-For I periodic, this equation always has a periodic solution. However, when/EAP,M[f] = 0, this may not be the case. In fact,
I(t)is in AP, M [fJ = O. However, if
fI
~ 1 't. /n 2 "2e
k
n=ln
is in AP, then
fIN (constant) +
1:.;= 1(-z)eit/n2 .
However, this latter funtcion is not in AP since the sum of squares of the Fourier coefficients does not exist. ,The following result is known about the integral of an almost periodic function. THEOREM 9.
If lEAP, then
ItI is
in AP if and only· if it is bounded.
In the applications to differential equations, this proposition generally cannot be applied. The next result due to Bogoliubov [1] is concerned with an AP approximation to the solution of the equation y - I(t) = 0 when lis AP, hasM[fJ = 0 and does not necessarily have a bounded integral. LEMMA 4. SuppoSe I is in AP, M [fJ is a continuous scalar function {("7), 0 < "7 < the almost periodic function!1/ defined by
(1)
!1/(t) ~
t
For every 1/ > 0, there {("7) -+0 as "7 -+0 such that
= O. 00,
e-1/(t-B)!(s) dB
-00
satisfies (2)
(a)
1!1/(t>1 ~ "7-1 {("7),
(b)
Id!1/(t)/dt -:-f(t)1 ~ {("7),
347
APPENDIX. ALMOST PERIODIC FUNCTIONS
for all t in R. Also, m[fl1] C m[f]. PROOF.
Since M[f] = 0, there is a continuous nonincreasing scalal
function e( T), 0
<
<
T
00,
e(T) --* 0 as T --* 00 such that IT-l J:+ T f (s)
e(T) for all t in R. Furthermore, for any T
dsl ~
> 0,
{Xlo e- l1Sf(t-s)ds
fl1(t) =
co
I
=
e- l1kT
k=O
fkT(k+l)T f(t -
s)e-l1 (s-kT) ds.
If B is a bound for if(t)1 on (-00, (0), then the above relation yields Ifl1(t)i
~kI:,oe-l1kT co
IIkT(k+l)T f(t -s) ds I
(k+l)T e-l1 kT [ (1 - e-l1(s-kT») ds k=O JkT co
+B I ~
Le- l1kTe(T)T + B J (1 T
co
k=O
~
co
e- 118 ) ds
0
e(T)T[1 - e- l1T ]-l
I
e- l1kT
k=O
+ BT.
Since e( T) --* 0 as T --*00 and is nonincreasing, there always exists a unique solution to the equation 1 - e- l1T = e(T). Suppose Tl1 is chosen to be this solution. Since e( T) > 0 for all T, it is clear that T 11 --* 00 as 'I] --* 0 and this together with the fact that e(T) --*0 as T ~ 00 imply that 'l]Tl1 ~O as 'I] --*·0. The solution Tl1 is obviously contin)lous in '1]. If we let '('1]) = (B + 1)'I]Tl1 , then relation (2a) is proved. From (1), it follows that (3)
dfl1(t) - f(t) dt
=
-'I]fl1(t),
and therefore, (2b) follows from (2a). It remai~s, to show that f1/ E AP. Suppose (x' CR. Since f E AP, there is a sequence (X C (x' such that TOtf exists uniformly. Thus, for any e> 0, there is l11N(e) such that
If1/(t + exn) - f1/(t +am)1 ~
Looo e1/ lf(t + s + exn) - f(t + s + am)lds S
for n,m ~ N(e). This shows f1/ E AP. The fact that m[f1/1 C m[fl follows from Theorem 8. This proves the lemma.
OR DINAR Y DIFFERENTIAL EQUATIONS
348
We want to study almost periodic solutions of an almost periodic differential equation oX = [(t, X). Thus, we need to know when [(t,t/>(t» is almost periodic if 4> E AP. In general, this is not true. One can see this from the example [(t, x) = sinxt by showing that [(t, sin t) = sin(tsin t)is not uniformly continuous. The difficulty arises from the fact that the e-translation set for sinxt is not well behaved in x. An appropriate generalization of the definition of an almost periodic function depending on parameters is the following one. Defmition 6. A continuous function [:R X D ~ en, where D is open in en, is ~aid to be almost periodic in t uniformly for xED if for any e > 0,
the setir([,e) = xEK n T(f(·,x),e) is relatively dense in-R' for each compact set . . . K CD. Equivalently, if for any e > 0; and any compact setK CD, there exists L(e,K) > 0 such that any interval of length L(e,K) contains a T with I[(t + T,X) - [(t, x) 1< e
for all tER,xEK.
One can then prove the following results (see Fink or Yoshizawa). THEOREM 10. If [(t,x) is AP in t uniformly for x ED, then [is bounded and uniformly continuous on RX K for any compact set K CD. THEOREM 11. If [(t,x) is AP in t uniformly for xED, then, for any R, there is an 01. C 01.' and a function g(t: x) which is AP in t uniformly for xED such that [(t + an,x) ~ g(t,x) uniformly on R X K for any compact setKCD. Conversely, if [:R X D ~ is continuous and for any 01.' C R, there is an 01. C 01.' such that, for any compact set K C D,f(t + an,x) converges uniformly on R'X K, then [(t, x) is AP in t uniformly for xED. In general, functions AP in t uniformly for xED satisfy the same properties as before if convergence is always uniform on compact sets K CD. 01.' C
en
THEOREM 12. If [(t,x) is AP in t uniformly for xED, and a compact set ofD, then[(·»EAP.
~
E AP,
~(t)EK,
In the applications to differential equations, it is desirable to have an improvement of Lemma 4 in this more general situation. More specifically, one wants an approximating solution of the equation
ay
- - f ( t x)=O
at
'
,
which is very smooth in x. One can apply Lemma 4 and more or less classical smoothing operators to achieve this end result. If M[f(·, x)] denotes the mean value of f(t, x) with respect to t, then we have
APPENDIX. ALMOST PERIODIC FUNCTIONS
349
LEMMA 5. Suppose [(t,x) is AP in t uniformly with respect to x in a setD containing the closed ballBa = {x in Ck:lxl ~a}. IfM[[(',x)] = 0 for x in Ba then for any al < a, there are an 1}o > Oand a function w(t,X,1}) defined and continuous for t in R, x in B a " 0 < 1} < 1}o, which is almost periodic in t uniformly with respect to x in B a , and 1} in any compact set of (O,1}o), m[w(·,x,1})] C m[[(',x)], w(t,x,1}) has derivatives of any desired order with respect to x and is such that, if g(t, x, 7])
ow(t, x, 7])
=
ot
-
j(t, x),
then g(t; x, 1/), 7]w(t, x, 7]) and 7]Ow(t, x, 7])/ox approach zero as 7] ~O uniformly with respect to t in R, x in B u ,. If, in addition, j(t, x) has continuous first partial derivatives with respect to x, then og( t, x, 7])/ ox ~ 0 as 7] ~ 0 uniformly with respect to t in R, x in B u,. PROOF. From Lemma 4, it follows that there is a function jn(t, x) defined by (1) which is a.p. in t uniformly with respect to x in Bu and 7] in any compact set such that
(4)
ojn(t, x) -o-t- - j(t, x)
=
-7]jn(t, x),
where ~(7]) ~O as 7] ~o. For a fixed a > 0 and some fixed integer q ~ 1, consider the function ~a(x) defined by for for where the constant d a is determined so that
Ixl ~ a Ixl >a
fB" ~a(x) dx = l.
Define
w(t, x, 7]) by (5)
w(t, x, 7]) =
f
B"
~a(x -
y)jn(t, y) dy.
It is easy to see that w(t, x, 7]) is a.p. in t uniformly with respect to x in Bu and 7] in any compact set since jn(t, x) has the same property. The function ~a(x - y) possesses continuous partial derivatives up through order 2q - 1 with respect to x which are bounded in norm by a function G(a)/(area of integration) where G(a), 0 < a < 00, is continuous. The function G(a) may approach 00 as a~O. Therefore, from (4), the function w(t, x, 7]) defined by (5) has partial derivatives with respect to x up through order 2q -1 which are bounded by G(a)~(7])7]-l. Since q is an arbitrary integer, the number of derivatives with respect to x may be as large as desired. Choose a = an as a function of 7] in such a way that an ~O, G(an)~(7]) ~O
350
ORDINARY DIFFERENTIAL EQUATIONS
as 7] -+0. The conclusion of the lemma concerning 7]w(t, x, 7]) is therefore valid since 7]w, 7]Ow/ox are bounded by G(al1)'(7]). For any Gl < G, choose 7]0 so small that Gl + a l1 < G for 0 < 7] < 7]0.
.
From the definition of da(x), it follows that f~ fia,,(x - y) dy = 1 for every x in B u" 0 < 7] < 7]0. Therefore, .from relations (4) and (5), it follows that (6)
g(t, x, 7]) =
fB .. da,,(x -
y)[!(t, y) - !(t, x)] dy,
where g = g + 7]w. From 'the definition of fia(x); it follows that Ig(t, x, 7])1 ~
sup
I!(t, y) -!(t, x)l.
0;:;;12:-1I1;:;;a"
Since a l1 -+O as 7]-+0 and!(t, x) is uniformly continuous for tin R, x in B u" there is a continuous function 8(7]),0 <7] <7]0,8(7])-+0 as 7]-+0 such that sup
I!(t, y) - !(t, x)1 < 8(7]).
O;:;;I2:-III;:;;a"
+
Consequently, Ig(t, x, 7])1 < 8(7]) and Ig(t, x, 7]1 < 8(7]) G(al1)'(7]) and g satisfies the properties stated in the first part of the lemma. If, in addition, ! (t, x) has continuous first partial derivatives with respect to x, then an integration by parts in relation (6) yields og(t, x, 7]) oX
=
f B ..
fia (x _y)[O!(t, y) _ o!(t, X)] dy. " oX ox
The same, type of argument as before shows that this expression is bounded by a function 81 (7]) which approaches zero as 7] -+0. This completes the proof of the lemma. • If, in Lemma 5, the function!(t, x) is periodic in some of the components of x, then the fUIl,ction w(t, x, 7]) can also be chosen periodic in these components with the same period. If g is an n-vector and ~ is an r-vector, then g(~) is said to be multiply periodic in ~ with vector period W = (WI, ... , wr), Wi:> 0, j = 1, ,2, ... , T, if the function g is periodic in the lh component of ~ with periQQ. Wj. If g( ~) is multiply periodic, then the function g(cpl t, ... , ~r t) is AP in t uniformly with respect to ~. It is therefore meaningful to define the mean value of this function with· respect' to t. This concept of mean value was first used in differential equations by S. Diliberto [1]: To simplify notations, let ~ t = (~1 t, ... , ~r t) and designate the above mean value by
+
+
+
+
+
f ",_coT
1 T
(7)
Mt/>[g(cp)]
= lim -
r
g(cp+t)dt.
)
0
To understand this concept a little better, consider the case where cp is a two-vector, ~ =(~1' ~2) and
APPENDIX. ALMOST PERIODIC FUNCTIONS
g(cp) '"
351
L ajkei(kP
In usual treatises on multiply periodic functions, the mean value of g is the constant term in the Fourier series of g; that is, aoo. According to definition (7), M q,[g(cp)] =
L
ajk ei(kpq,l+jO)q,.)
j, k: k/L+jw~O
and this could be a function of cp if the frequencies /-', ware such that /-,jw is rational. Lemma 5 has an appropriate generalization to the case of functions of the form g(t, cp, x) which are multiply periodic in the vector cp, AP in t, uniformly with respect to c/>. in Cr and x in Bu and Mt,q,[g(t, cp, x)] = O. Under these conditions, for any 01 < a, there are an T)o > 0 and a function W(t, cp, x, T)), 0 < T) < T)O, multiply periodic in cp, -AP in t such that the function oW
G(t, cp, x, 7])
r
oW
= at + j~O ocpj -
g(t, cp,x)
as well as 7] W, 7]oWjocp, 7]oWjox --+0 as 7] --+0 uniformly with respect to t in R, has continuous first partials with respect to x, then the partial derivatives of G(t, ep, x, 7]) with respect to cp, x also approach zero as 7] --+0 uniformly with respect to t, cp, x. The proof of this fact is very similar to the proof of Lemma 5 and may be found in Hale [3].
ep in Cr, x in B u,. If g(t, cp, x)
REFERENCES Andre, J., and P: Seibert [1]. On after-endpoint motions of general discontinuous control systems and their stability properties. Proc. 18t Intern. IFAG Gongr. Moscow 1960, II, 919-922. Andronov, A. A.., S. E. Khaikin, and A. A. Witt [1]. Theory of 08cillator8, Pergamon Press, New York, 1966. Andronov, A. A. Leontovich, E. A., Gordon, I. I. and A. G. Maier [1], Theory of Bifurcations of Dynamic Systems on a Plane, John Wiley and Sons, 1973. Antosiewicz, H. [1]. Boundary value proplems for nonlinear ordinary differential equations. Pac. J. Math. 17 (1966), 191-197. [2]. Un analogue du principe du pointe fixe de Banach. Ann. Mat. Pura. Appl. (4) 74 (1966), 61-64. [3]. On an integral inequality. To appear. Arnold, V. 1. [1]. Small denominators. 1. Mapping the circle onto itself. Izv. Aka.k. Nauk SSSR Ser. Mat. 25 (1961), 21-86. [2]. Small denominators and problems of !ltability of motion in classical and celestial mechanics. Uspehi Mat. Nauk 18 (1963), no. 6(114), 91-192.
352
REFERENCES
" Arscott, F. M. [1]. Periodic Differential EquationB. Pergamon Press, New York, 1964. Auslander, .r. and W. H. Gottschalk [1]. Topological Dynamic8. W. A. Benjamin, New York, 1968. Bailey, H. R. and L. Cesari [1]. Boundedness of solutions of linear differential equations with periodic coefficients. Arch. Rat. Meeh. Ana. 1 (1958), 246-271. Bancroft, S., J. K. Hale and D. Sweet, [1]. Alternative problems for nonlinear functional equations. J. Differential EquationB 4 (1968), 40-56. Bartle, R. G. [1]. Singular points of functional equations. TranB. Am. Math. Soc. 75 (1953), 366-384. Bass, R. W. [1]. Equivalent linearization, nonlinear circuit analysis and the stabilization and optimization of control systems. Proc. Symp. Nonlinear Circuit Anal. 6 (1956). [2]. Mathematical legitimacy of equivalent linearization by describing functions. IFAC CQngreB8, 1959. Bellman, R. [1]. Stability Theory of Differential EquationB, McGraw-Hill, New York, 1953. [2]. Introduction to Matrix AnalysiB. McGraw-Hill, New York, 1960. Bendixson, 1. [1]. Sur les courbes definies par des equations differentielles. Acta. Math. 24 (1901), 1-88. Birkhoff, G. D. [1]. Dynamical Systems. Am. Math:. Soc. Colloq. Publ. New York, 1927. Bogoliubov, N. N. [1]. On some statistical methods in mathematical physics. Akad. Nauk. Ukr. R.S.R. 1945. Bogoliubov, N. N. and Y. A. Mitropolskii [1]. Asymptotic Methods in the Theory of Nonlinear OscillationB. Gordon and Breach, New York, 1961. [2]. The method of integral manifolds in nonlinear mechanics. ContributionB to Differential EquationB 2 (1963), 123-196. Bogoliubov, N. N., Jr., and B. 1. Sadovnikov [1]. On periodic solutions of differential equations of nth order with a small parameter. Symp. Nonlinear VibrationB, Kiev, USSR, Sept., 1961. Borg, G. [1]. On a Liapunov criteria of stability. Am. J. Math. 71 (1949), 67-70. Borges, C. [1]. Periodic solutions of nonlinear differential equations: existence and error bounds. Ph.D. dissertation. University of Michigan. 1963. Cartwright M., [1]. Forced oscillations in noniinear systems. Contributions to the Theory of Nonlinear Oscillations, 1 (1950), 149-241, Annals Math. Studies, No. 20, Princeton. Cesari, L. [1]. Asymptotic Behavior and Stability Problems. Springer, 1959; Second edition, Academic Press, New York, 1963. [2]. Proprieta asintotiche delle equazioni differenziali lineari ordinaire. Rend. Sem. Mat. Roma 3 (1939), 171-193. [3]. Sulla stabilita delle soluzioni dei sistemi di equazioni differenziali lineari a coefficienti periodici. Atti Accad. Mem. Classe Fis. Mat. e Nat., (60) 11 (1940), 633-692. [4]. Existence theorems for periodic solutions of nonlinear differential systems. Boletin Soc. Mat. M exicana 1960, 24-41. [5]. Functional analysis and periodic solutions of nonlinear differential equations. Contr. Diff. Equations 1 (1963),149-167. [6]. Functional analysis and Galerkin's method. Michigan Math. J., 11 (1964), 385-414. [7]. Existence in the large of periodic solutions of hyperbolic partial differential equa-' tions. Arch. Rat. Mech. Anal. 20, (1965), 170-190. [8]. A nonlinear problem in potential theory. Michigan Math. J., 1969. Cesari, L. and J. K. Hale [1]. A new sufficient condition for periodic solutions of weakly differential systems. Proc. Am. Math. Soc. 8 (1957), 757-764. . Cetaev, N. G. [1]. Un tMoreme sur l'instabiliM. Dokl. Akad. Nauk SSSR,[2 (1934), 529-534. Chafee, N. [1]. The bifurcation of one or more closed orbits from an equilibrium point of an autonomous differential system. J. Differential EquationB, 4 (1968), 661-679. Chen, K. T. [1]. Equivalence and decomposition of vector fieius about an elementary critical point. Amer. J. Math. 85 (1963), 693-722.
REFERENCES
353
Coddington, E. A. and N. Levinson [1]. Theory of Ordinary Differential Equationa. McGraw-Hill, New York, 1955. Conley, C. [11, Isolated invariant sets and the Morse index. CBMS number 38~ Am. Math. Soc., Providence, R. I. Coppel, W. A. [1] Stability and Asymptotic Behavior of Differential Equationa. Heath Mathematical Monographs, 1965. Coppel, W. A. and A. Howe [1]. On the stability oflinear canonical systems with periodic coefficients. J. AU8tral. Math. Soc. 5 (1965), 169-195. [2]. Corrigendum: On the stability of linear canonical systems with periodic coefficients. J. AU8tral. Math. Soc. 6 (1966), 256. Coppel, W. A. and K. J. Palmer [11, Averaging and integral manifolds. BulL Australian Math. Soc. 2(1970), 197-222. Cronin, J. [1]. Branch points of solutions of equations in a Banach space. Trana. Am. Math. Soc. 69 (1950), 208-231. Cruz, M. A. and J. K. Hale [1]. ~tability of functional differential equations of neutral type. J. Differential Equations. 7(1970),334-355. Denjoy, A. [I]. Sur les courbes definies par les equations differentielles a la surface du tore. J. de Math. 11 (1932), 333-375. Diliberto, S. P. [1]. Perturbation theorems for periodic surfaces. I, II. Rend. Gire. Mat. Palermo 9 (2) 1960,265-299: 10 (2)(1961), Ill. [2]. On stability oflinear mechanical systems. Proc. Internat. Sympos. Nonlinear Vibrations. Izdat. Akad. Nauk Ukrain. SSR, Kiev. 1, 1963, pp. 189-203. (t:l) New results on periodic surfaces and the averaging principle. U.S.-Japanese Seminar on D~fferential and Functional Equations, pp. 49-87. W. A. Benjamin, 1967. Fillipov, A. F. [1]. Differential equations with discontinuous right hanli side. Mat. Sbornik (N.S.) 51 (1960),99-128. Flugge-Lotz, 1. [1]. Discontinuous automatic control. Princeton, 1953. Friedrichs, .K. [1]. Spec'ial Topics in Analysis. Lecture Notes, New York University, 1953-1954. [1]. Advanced Ordinary Differential Equations. Lecture Notes, New York University, 1956. Gambill, R. A. and J. K. Hale [1]. Subharmonic and ultraharmonic solutions of weakly nonlinear systems. J. Rat. l\r[ech. Ana. 5 (1956), 353-398. Golomb, M. [1]. Expansion and boundedness theorems for solutions oflinear differential equations with periodic or almost periodic coefficients. Arch. Rat. Mech. Ana. 2 (1958), 284-308. Gottschalk, W. H. and G. A. Hedlund [1]. Topological dynamics. Amer. Math. Soc. Golloq. Publ. 36 (1955). Graves, L. M. [1]. Remarks on singular points of functional equations. Trana. Am. Math. Soc. 79 (1955), 150-157. Hahn, W. [1]. Theory and Application of Lyapunov's Direct Method. Prentice·Hall, New York,1963. [2]. Stability of Motion, Springer-Verlag, New York, 1967. Halanay, A. [1]. Differential Equations-Stability, Oscillation, Time Lags. Academic Press, New York, 1966. Hale, J. K. [1]. Periodic solutions of nonlinear systems of differential equations. Riv. Mat. Univ. Parma 5 (1954), 281-311. [2], Sufficient conditions for the existence of periodic solutions of first and second order differential equations. J. Math. Mech. 7 (1958), 163-172. [3]. Integral manifolds of pt'rturbed differential systems. Ann. Math. 73 (1961), 496-531. [4]. On differential equations containing a small parameter. Gontr. Diff. Equations. 1 (1962) [5]. Asymptotic behavior of the solutions of differential difference equations. Proc. Intern. Symp. Nonlinear Oscillations, Kiev. Sept. 1961, II (1963), 409-426. [6]. On the behavior of solutions of linear differential
354
ORDINARY DIFFERENTIAL EQUATIONS
equations near resonance points. Ooner. Theory Nonlinear OBcillations IS (1960). [7]. OBcillations in Nonlinear SYBtem8. McGraw.Hill, New York, 1963. [8]. Sufficient conditions for stability and instability of autonomous functional differential equa· tions. J. Diff. Equationa 1 (1965), 452-482. [9]. Periodic solutions of a class of hyperbolic equations containing a small parameter. Arch. Rat. Mech. Anal. 23 (1967), 380-398. [10], Theory of Functional Differential Equations. AppL Math. Sciences, Vol. 3, Springer-Verlag, 1977. [11], Bifurcation near families of solutions. Proc. Int. Conf. Diff. Eqns., Uppsala (1977). Hale, J. K. and A. Stokes [1]. Behavior of sol~ions near integral manifolds. Arch. Rat. Meeh. Ana. 6 (1960), 133-170. Hale, J. K. and H. M. Rodriques [1], Bifurcation in the Duffing equation with independent parameters, I. Proc. Royal Soc. Edinburgh, 78A(1977), 56-65; [2] II. Ibid 79A(1977), 317":322. Hale, J. K. and P. Z. Ta'boas [I]. Interaction of damping and forcing in a second order equation. Nonlinear Analysis, 2(1978),77-84; [2] Bifurcation near degenerate families, ,Applicable Analy~is. To appear. Hall, W. S. [1]. Periodic solutions of a class of weakly nonlinear evolution equations. Ph.D. Thesis. Brown University, June, 1968. Harris, W. A., Jr., Y. Sibuya, and L. Weinberg [I]. Holomorphic solutions of linear differential systems at singular points. To appear. Hartman, P. [1]. Ordinary Differential Equations, Wiley, New York; 1964. [2]. On stability in the large for systems of ordinary differential equations. Oanad. J. Math. 13 (1961), 480-492. Henry, -D. P1, Geometric Theory of Semilin..ar Parabolic Equations. Sprin.ger-Verlag, '1980. Hopf, E. [I], Abzweigertemer periodischen wsung von einer stationaren Li)sung eines Differentialsystems. Berich. Math. Phys. SachlischenAkad. Wursenchaften Leipzig 94(1942). 3-22. Howe, A. [I]. Linear canonical systems with periodic coefficients. Ph.D. Dissertation, University of Canberra, Australia, 1967. Infante, E. F. and M. Slemrod [1]. An invariance principle for dynamical systems on Banach spaces, Proc. IUTAM Symp. on Stability Springer, 1970. Kelley, A. [I]. The stable, center-stable, center, center-unstable, unstable manifolds. J. Differential Equations, 3 (1967), 546-570. Knobloch, H.W. [1]. Remarks on a paper of Cesari on functional analysis and nonlinear differential equations. Michigan Math. J. 10 (1963),417-430. [2]. Comparison theorems for nonlinear second order differential equations . •J. Diff. Equations 1 (1965). 1-25. Kolinogorov, A. N. [1]. Theorie generale des systemes dynamiques et mecanique classique. Proc. Int. Oong. Math. 1954, I, 315-333. Noordhoff, 1957. Krasovskii, N. N. [1]. Stability of Motion. Stanford University Press, 1963. Krein, M. G. [1]: The basic propositions in the theory of A-zones of stability of a canonical system of linear differential equations with periodic coefficients (Russian), Pamyati A. A. Andronova, Izdat. Akad. Nauk. SSSR, Moscow. 1955, pp. 413-498 .. Krylov, N. and N. N. Bogoliubov [1]. The application of methods of nonlinear mechanics ;to the theory of stationary oscillations. Publication 8 of the Ukrainian Academy of Science, Kiev, 1934. [2]. Introduction to nonlinear mechanics, Annals Math;..StudieB, No. 11, Princeton University'Press, Princeton, N.J., 1947. . l' Kruzweil, J. [1]. Invariant manifolds for flows, Proo. Sym. Diff. Equations and hilnamical SyBtemB. Academic Press, New York, 1967. [2]. The averaging principle in CArt·'l.in speoial cases of boundary problems for-partial differential equations. OaaopiB Put.
REFERENCES
355
Mat. 88 (1963), 444-456. [3]. van der Pol perturbation of the equation for a v:ibrating string. Czecho8lovak Math. J. 17 (92) (1967), 558-608. Kyner, W. T. [1]. Invariant manifolds, Rend. Cire. Mat. Palermo, 10 (2) (1961), 98-110. Laksmikantham, V. and S. Leela [1]. Differential Inequalities. Academic Press, New York, 196~. . Laloy, M. [1], On the inversion of Lagrange-Dirichlet theorem in the case of an analytic potential. Rpt. 107, December 1977, Louvain-la-Neuve •. LaSalle, J. P. [1], Relaxation oscillations. Quart. Appl. Math. 7 (1949), 1-19. [2]. An invariance principle in the theory of stability. Int. Symp. on Diff. Eq8. and Dyn. SY8., p. 277, Academic Press, New York, 1967. LaSalle, J. P. and S. Lefschetz [1]. Stability by LiaPUnov'8 Direct Method. Academic Press, New York, 1961. Lee, B. and L. Markus [1]. Optimal Control Theory. Wiley, New York, 1967. Lefschetz, S. [I]. Differential Equations; Geometric Thwry. Second Edition. Interscience, New York, 1959. [2]. Stability of Nonlinear Control SY8tem8. Academic Press, New York, 1965. Levinson, .N. [1]. Transformation theory of nonlinear differential equations of the second order. Ann. Math. 45 (1944), 723-737. [2]. Small periodic perturbations of an autonomous system with a stable orbit. Ann. Math. 52 (1950), 727-738. Lewis, D. C. [1]. On the role of first integrals in the perturbation of periodic solutions. Ann. Math. 63 (1956), 535-548. [2]. Autosynartetic solutions of differential equations. Am. J. Math. 83 (1961), 1-32. Liapunoff, A. [I]. ProbUme General de la Stabilite du Mouvement. Annal8 Math. Studie8, No. 17, Princeton University Press. 1949. Lillo, J. C. [1]. A note on the continuity of characteristic exponents. Proc. Nat. Acad. Sci. U.S.A. 46 (1960), 247-250. Locker, J. S. [1]. An existence analysis for nonlinear equations in a Hilbert space. Trans. Am. Math. Soc. 128 (1967), 403-413. Lykova, O. B. [1]. Investigation of the solution of a system of differential equations with a small parameter on a two-dimension local integral manifold in the resonance case. Ukrain. Mat. Z. 10 (1958), 365-374. Loud, W. [I]. Periodic solutions of x' + cx' + g(x) = Bf(t). Mem. Am. Math. Soc .. No. 31-, 1959, 58 pp. [2J, Branching of solutions of two-parameter boundary value problems for second order differential equations. Ingenieur-Archiv 45(1976),347-359. MaWlus, W. and S. Winkler [1J. Hill'8 Equatwn. Interscience, New York, 1966. ,Malkin, J. U. [1]. '1'neory of Stability of Motion. AEC-tr-3352. [2]. Some Problem8 in the Theory of Nonlinear Oscillations. Moscow, 1956. Markus, L. [1]. Asymptotically autonomous differential systems. Contr. Theory Nonlinear 08cillations, 3 (1956),'17-29. Markus, L. and,H. Yamabe [1]. Global stability criteria for differential systems. 08aka Math. J. 12 (1960), 305-317. Marsden, J. E_ and M. McCracken [1], The Hopf Bifurcation and its Applications. SpringerVerlag, 1976. Massera, J. L. [1]. Contributions to stability theory. Ann. Math. 64 (1956), 182206. Massera, J. L. and J. J. Schaffer [1]. Linear Differential Equations and Function Spaces. Academic Press, New York, 1966. McCarthy, J. [1]. The stability of invariant manifolds, I, Stanford University, App!. Math. Stat. Lab., ONR Tech. Rep. 36, 1956.
356
ORDINARY DIFFERENTIAL EQUATIONS
McGarvey, D. [1]. Linear differential systems V\i'ith periodic coefficients involving I large parameter. J. Differential EquationB. 2 (1966), 115-142. McLachlan, N. W: [1]. Theory and Application of Mathieu FUnct~OnB. The C1arendol ·Press, Oxford, 1947.· Minorsky, N. [1]. Nonlinear 08cillationB. van Nostrand, Princeton, N.J., 1962. Minorsky, N. [2], Introduction to Nonlinear Mechanics. J. W. Edwards, Publisher, An Arbor, 1947. Mishchenko, Z. and L. S. Pontrjagin [1]. DifferentIal equations with a small paramete attached to the higher derivativesvand some problems in the theory of oscillatioll IRE TranB. CT-7 (1960), 527-535. Mitropolski, Y. A. [1]. ProblemeB de la theorie aaymptotique deB oacillatWnB non 8tationaire.! Gauthier-Villars, Paris, 1966. . ~ntandon, B. [1], Almost periodic solutions and integral manifolds for weakly nonlinel nonconservative systems. J;Differential Eqns. 12{19721. 417-425. Morrison, J. A. [1]. An averaging scheme for some nonlinear resonance problems. SIAl. J. Appl. Math. 16 (1968), 1024-1047. Moser, J. [1]. On invariant curves of area-preserving mappings of an annulus. NOOhl Akad. WiB8. GOttingen Math.-PhY8. Kl. 11 1962, 1-20. [2]. Bistable systems 0 • differential equations with applications to tunnel diode circuits. IBM J. Re.! Develop. 5 (1961), 226-240. [3], Stable and Random Motions in Dynamical Syste1ll Princeton University Press. 1973. . _. Nemitskii, V. V. and V. V. Stepanov 1)]. Qualitative Theory of Differential EquatiOm Princeton University Press, 1960. Newhouse, S. [1], Lectures on Dynamic Systems. eIME, Summer Session on Dynamica Systems, June, 1978. Nirenberg,~, [1]. Functionaf Analy~. Lec~ure Notes, New York Univer~ity, 1960-196] Nitecki, Z:. fll ,Diffeientiable Dynamics. M;I.T. Press, cambridge, Mass., 1971. . biech, C. [1]. On global stability of an autonomous system· on the plane. Oontributiofl to Differential EquationB 1 (1963), 389-400. ~ucliic, N.- [1], Applications of the topol~gical· method ~f ·Wdewski to certain problem of asymptotic behavior in ordinary differential equations. Pacific J. Math. 11{1961] 1511-1527; [2] On the asymptotic behavior· of the solutions of functional diffeienm equationS, pp. 223-233 in Differential. Equations and Dynamical Systems. Ed. J. I< Hale andJ. p.-LaSalle. Academic Press. 196?_ . Opial, Z. [1]. Sur ]a dependance des solutions d'un systeme d'equations differentielle de leurs seconds membres. Applications aux systemes presques autonomes. An,. PoWn. Math. 8 (1960), 75-89. Palamadov, V. P. [1], ·Stability of equilibrium ht a potential field. Functional Analysis an. its Applicati~!lIUJ18), 277-289 •. _ __ _ Palmore, J.[ 1] , Instability of equilibria, Preprint. Peixoto, M. [1]. Structural stability on two-dimensional manifolds. TopOlogy 1 (1962) 101-120. Pere1l6, C~ P~:. Periodic solutions of diffe~ntial equations with time l~ containing : .small parameter. I. Diff. Equations. 4(1968),160-175 [2] .'Anote on periodic sOlutior . of nonlinear diffeiential equations with time lags, Differential Equations. and Dynamic. .. ;to . _ Systems."!t-cademic: Press, 196?,l~5-}88._. Perron, O. [1]. Die Stabilit&tsfrage bei DifferentialgleichungssysteII\ei"Math. Zeit. 3~ (1930), 703-728. j' Pliss, V. A. [1]. On the theory ofinvarian~ surfaces. /J;,fferentialrriye Umvnarriya Z (1966) 1139-1150. [2]. Nonlocal Problema oj the Theory oj OacillationB. Academic Press New York, 1966.
REFERENCES
357
Poincare, H. [1]. Sur les courbes definies par les equations differentielles. J. de Math., (3) 7 (1881), 375-422; (3) 8 (1882), 251-296; (4) 1 (1885), 167-244; (4) 2 (1886), 151-217. [2]. Les methodes nouvelles de la mecanique celeste. Paris, 3 vols., 1892, 1893, 1899. Rabinowitz, P. H. [1]. Periodic solutions of a nonlinear wave equation. Comm. Pure Appl. Math. 20 (1967), 145-205. Reuter, G. E. H. [1]. Subharmonics in a nonlinear system with unsymmetric restoring forces. Quart. J. Mech. Appl. Math. 2 (1949), 198-207. Rouche, N. Habets, P. and M. LaLoy [1], Stability Theory by Liapunov's Direct Method. Appl. Math. Sci. Vol. 22, 1977, Springer-Verlag. Sacker, R. J. [1]. A new approach to the perturbation theory of invariant surfaces. Comm. Pure Math. 18 (1965), 717-732. Sansone, G. and R. Conti [1]. Nonlinear Differential Equations, Pergamon Press, New York,1964. Schwartz, A. J. [1]. A generalization of a Poincare-Bendixson theorem to closed twodimensional manifolds. Am. J. Math. 85 (1963), 453-458; errata, ibid, 85 (1963), 753. Sethna, P. R. [1]. An extension of the method of averaging. Quart. Appl. Math. 25 (1967), 205-211. Siegel, C. L. and J. K. Moser [1], Lectures on Celestial Mechamcs, Grundlehren, Vol. -187, Springer, 1971. Smale, S. [1]. Differentiable dynamical systems. Bull. Amer. Math. Soc. 73 (1967), 74'i-817. Sotomayor, J. [IJ, Generic one parameter families ot vector fields, Publ. Math. IHES, 43(1973), 5-46. Sternberg, S. [1]. Local contractions and a theorem of Poincare. Am. J. Math. 79 (1957), 809-824. [2]. On the structure of local homeomorphisms of euclidean n-space, II. Am. J. Math. 80 (1958), 623-631. [3]. 'III, Ibid, 81 (1959), 578-604. Szarski, J. [1]. Differential Inequalities. Warszawa, 1965. Urabe, M. [1]. Nonlinear Autonomous Oscillations. Academic Press, New York, 1967. [2]. Galerkin's procedure for nonlinear periodic systems and its extension to multipoint boundary value problems for general nonlinear systems. Numerical Solutions of Nonlinear Differential Equati,!ns (Proc. Adv. Sympos. Madison, Wis., 1966), pp. 297-327. Wiley, New York, 1966. Vainberg, M. M. and V. A. Tregonin [1]. The methods of Lyapunov and Schmidt in the theory of nonlinear differential equations and their further development. Mathematical Surveys, 17 (1962), 1-60. van der Pol, B. [1]. Forced oscillations in a circuit with nonlinear resistance. Phil. Mag. 3 (1927), 65-80; Proc. Inst. Radio Eng. 22 (1934), 1051-1086. Wasow, W. [1]. Asymptotic Expansions of Ordinary Differential Equations. Interscience, New York, 1965. Wa~ewski, T. (1), Sur un principe topologique de l'examen de l'allure asymptotique des integra:les des equations differentielles ordinaires. Ann. Soc. Polon. Mat. 29(1947), 279-313. Williams, S. [1]. A connection between the Cesari and Leray-Schauder methods. Michigan Math. J., 1969. Yakubovich, V. A. [1]. Structure of the group of symplectic matrices and the unstable canonical systems with periodic coefficients. Mat. Sbornik N.S. 44 (86) (1958), 313-352. [2]. Critical frequencies of q';1asi-canonical systems. Vestnik Leningrad Univ. 13 (1958), 35-63. Yoshizawa, T. [1]. Asymptotic behavior of solutions of a system of differentiai equations. Contributions to Differential Equations 1 (1963), 371-387. [2]. Stability Theory of Liapunov's Second Method. Mathematical Society of Japan, 1966.
358
ORDINARY DIFFERENTIAL EQUATIONS
YoshizaW8, T. and J. Kato [I]. Asymptotio Behavior of Solutions near Integta.l Manifolds. DiJlerenWil EquationB and Dynamical Systems, J. K. Harle and J. P. LaSalle, Eds., Academic frese, New York, 1967, pp. 267-275. . Zubov, V. I. [1]. Methods of A. M. Lyapunov and their AppUcatiom. Noordhoff. 1964.
Index
Admissible, definition of, 154 strongly, 171 Almost periodic function, 339 e-translation set, 342 hull of, 341 module of an, 345 uniformly in x, 348 Alternative problem, 298, 304 Arc, 39 Averaging, method of, 175, 190,252
Equation, adjoint, 80, 92 autonomour,37 Duffing,.168, 199, 206, 287 Hill, 131 linear variational, 21 van der Pol, 60, 198, 254, 263 Equations, bifurcation,.258, 261, 269 determining, 258, 261, 269 Ergodic flow, 71, 72 Esadi diode, 300
Bifurcation, equation, 258, 261, 269 generie, 273, 274 Hopf,270 several parameter, 268,274,287,290
First integrals, 283 Fixed point theorem, Brouwer, 10 contraction, 5 Schauder, 10 Tychonov,11 Floquet representation, 118 Focus, 42, 104 . Fourier, coefficients, 345 exponents, 345 series, 345 Fredholm's alternative, 146 Frequency response curve, 201 Function, Liapunov, 316, 325 negative definite, 311 positive definite, 311 Fundamental matrix, 80
Canonical systems, 133, 136 Caratheodory conditions, 28 Cauchy sequence, I Center, 40,103 Characteristic exponents, 119 multipliers, 119 number, 142 Conservative, 176 Continuity in parameters, of fixed points, 7,8 of solutions, 20 Contraction mapping principle, 5 Converse theorems of Liapunov, 327 Critical point, 38 Curve, 38
Galerkin approximation, 307 Green's function, 148 Hamiltonian systems, 133 Harmonic forcing, 199 Hill's equation, 121 Homeomorphism, 3 Hopf bifurcation, 270
Derivative, Frechet, 6 Differentiability in parameters, of fixed points, 8 of solutions, 21, 22 Duffing equation, 168, 199, 206, 287 Dynamical system, 50
Implicit function theorem, 8 Inequality, differential, 30 Gronwall, 36 Instability, definition of, 26 criteria for, 116,317 in conservative systems, 177
Eigenspace, generalized, 96 Entrainment of frequency, 203
359
360 Integrals, 176, 283 . linearly independent, 285 Invariant set, 47 Jordan canonical form, 99 Hordan curve, 39 theorem, 52 J-unitary, 134 Limit cycle, 42, 54, 57 Limit set, 47 Linear systems, autonomous, 93 general,79 noncritical, 144 nonhomogeneous, 145 periodic, 1 ! 7 stability of, 83 Linearly independent, functions, 90 subspaces, 96 Lip (7), 111 Lip (7),M), 154 Manifold, integral, 230 stable, 107, 108, 159 unstable, 107,108,159 Mapping, bounded linear, 3 compact, 10 completely continuous, 10 contraction, 5 domain of, 3 linear, 3 range of, 3 uniform contraction, 6 Monodromy matrix, 119 Moving. orthonormal coordinates, 214 Multiple periodic function, 350 Node, 101, 102 Nonconservative, 184 Noncritical system, 144 Norm, vector, 1 linear mapping, 3 Oribt, definition of, 37 Orbital stability, definition of, 221 conditions for, 22.1, 224 Path,37 Path cyliner, 43 closed,44 open,43
INDEX
Path ring, 45 Pendulum, conservative, 180 with oscillating support, 208 period, 180 Period in planar systems, 180 Periodic orbit, definition of, 39 stability of, 221, 224 perturbation of, 226, 229 Point, critical, 38 equilibrium, 38 regular,38 singular, 38 Principal matrix, 80 Property (E), 280 Quasiperiodic function, 345 Reciprocal systems, 131 Regular point; 38 Rotation number, 66 Saddle point, two dimensions, 102 of type (k), 106, 107 Scalar equation, 12 Secular term, 184 Set, bounded, 2 compact, 2 convex, 10 invariant, 47 minimal,48 O!-Iimit, 4 7 w-limit, 47 Singular perturbation, 166 Singular point, 38 Solution, definition of, 12 continuation of, 16 domain of definition, 18 existence of, 14 fundamental matrix, 80 general,25 orbit of, 37 path of, 37 principal matrix, 80 trajectory of, 18 uniqueness of, 18 Space, Banach, 1 complete, 1 t:t'(D,Rn),2
normed linear, 1 phase, 37 state, 37
361
INDEX
Stability, exponential, 84 of invariant set, 54, 221,240 of solution, 26 orbital, 221 under constant disturbances, 333 Stable on (-00,00),132 strongly, 132 Structurally stable, 143 Subharmonic, 206
invariant, 250, 253 ff. Trajectory, 18 Transformation, of van der Pol, 198 Transversal, 43, 52
Torus, existence of, 231, 237 flow on, 70, 74
Wazewski, principle of, 335 Wronskian, 90
Uniform boundedness,principle of, 4 Van der Pol equation, 60, 198, 254, 263 Variation of constants formula, 81
