® INDIA EDITION DIFFERENTIAL EQUATIONS any RESTRICTED! FOR SALE ONLY IN Shepley L. Ross—__ DIFFERENTIAL —— EQUATIONS ~~ ‘Third Edition i Shepley L. Ross Unicity of New Hampshire WILEY- INDIA| | | | | DIFFERENTIAL EQUATIONS Third Eattion Copyright® 2004 by John Wiley & Sons, Inc. All rights reserved Authorized reprint by Wiley india (P.) Lid, 4435V7, Ansari Road, Daryagan), New Delhi 110 002, All rights reserved, AUTHORIZED REPRINT OF THE EDITION PUBLISHED BY JOHN WILEY & SONS INC., UK. No part ofthis book may be reproduced in any ‘orm without the writen permission ofthe publisher. Limits of Liabitty/Disclaimer of Warranty: The publisher and the author make no representation or warranties with respect to the accuracy or completeness of the ‘contenis of this work and specifically disclaim all warranties, including without livtation warranties of fitness for a particular purpose, No warranty may be ctoated or extended by sales or promotional materials. The advice and strategies contained herein may not be suitable for avery situation. This work is sold win the ‘understening that the publisher is not engaged in rendering legal, ‘accounting, or other professional services. if professional assistance is required, the services of a competent professional person should be sought, Neither the publisher nor the author shall be {iabie for damages arising hereirom, The fact ‘that an organization or Website If referred to in this work a8 a citstion andlor a potential source of further Information does nof man that the author or the ‘publisher endorses the information the organization or Website ray provide or recommendations it may make. Further, readers should be aware that internet ‘Websites listed in this work may have changed or disappeared between when ‘this work was written and when itis read, Wiley also puitishes its books in a variety of electronic formats. Some content that eppears in print may not be avaleble in electronic books. For more Information about Wiley products, visit our website at wnaw.iley.com. Reprint : 2010 Printed at : Rajiv 800k Binding House, Delhi ISBN: 978-81.265-1537-0== PREFACE———_— ‘This third edition, ike the first two, isan introduetion to the basic methods, theory, and applications of differential equations. A knowledge of elementary calculus is presupposed, ‘Thedetailed style of presentation that characterized the previous editions of the text hhas been retained. Many sections have been taken verbatim from the second edition, while others have been rewritten of cearranged with the sole intention of making them clearer and smoother. As in the earlier editions, the text contains many thoroughly worked out examples. Also, a number of new exercises have been added, and assorted ‘exercise sets rearranged to make them more useful in teaching and learning ‘The book is divided into two main parts. The first part (Chapters 1 through 9) deals: with the material usually found in a one-semester introductory course in ordinary differential equations. This par is also available separately as Introduction o Ordinary Differential Equations, Third Edition ohn Wiley & Sons, New York, 1980). The second part of the present text (Chapters 10 through 14) introduces the reader to certain specialized and more advanced methods and provides an introduction 10 ‘fundamental theory. The table of contents indicates just what topics are treated. The following additions and modifications are specifically noted. 1. Material emphasizing the second-order linear equation has been inserted at appropriate places ia Section 4.1. 2, Newillustrative examples, inciuding an especially detailed introductory one, have been written to clarify the Method of Undetermined Coefficients in Section 43, and a useful table has also been supplied, 3. Matrix multiplication and inversion have been added to the introductory material on linear algebra in Section 7.5. 4. Additional applications now appear in the text in Sections 33 and 72. 5. Section 7.6isa completely new section on the application of matrix algebra fo the solution of linear systems with constant coefficients in the special case of wo ‘equations in two unknown functions. The theory that occupied this section in theiv vestace. second edition now appears in Chapter 11 (see note 9 following). We believe that this change represents a major improvement for both intcoductory and inter- mediate courses. 6. Section 7,7extends the matrix method of Section 7.6to the case of linear systems. with constant coefficients involving m equations in n unknown functions. Several etailed examples illustrate the method for the ease m = 3, 7. Both revised and new material on the Laplace Transform of step functions, translated functions, and periodic functions now appears in Section 9.) 8 The basic existence theory for systems and higher-order equations, formerly located at the beginning of Chapter L1, has now been placed at the end of ‘Chapter 10. This minor change has resulied in better overall organization. 9 Chapter 11, the Theory of Linear Differential Equations, has been changed considerably. Sections 11.1 through 114 present the fundamental theory of linear systems. Much of this material was found in Section 7.6in the second edition, and some additional results are also included here. Sections {1.5 through 11.7 now ‘present the basic theory of the single nth-order equation, making considerable use ‘of the material of the preceding sections. Section 11 8 introduces second-order self-adjoint equations and proceeds through the fundamentals of classical Sturm ‘Theory. We believe that the finear theory is now presented more coherently than in the previous edition 10, An appendix presemts, without proof, he fundamentals of second and third order determinants, ‘The book can be used as a text in several different types of courses. The more or less traditional one-semester introductory course could be based on Chapter ? through. Section 7.4 of Chapter 7if elementary applications are to be included. An alternative fone-semester version omitting applications but including numerical methods and ‘Laplace transforms could be based on Chapters 1,2, 4,6, 7,8, and 9. An introductory ‘course designed to lead quickly to the methods of partial diflerentisl equations could be based on Chapters 1, 2(in part), 4,6, 12, and 14. “The book can also be used as a text in various intermediate courses for juniors and seniors who have already had a one-semesier introduction to the subject. An interme- diate course emphasizing further methods could be based on Chapters 8, 9, 12, 13, and 14. An intermediate course designed as an introduction to fundamental theory could be based on Chapters 10 through 14. We also note that Chapters 13 and 14 can be interchanged advantageously. T am grateful to several anonymous reviewers who made useful comments and suggestions. I thank my cotleagues Wiliam Bonnice and Robert O. Kimball for helpful advies. [also thank my son, Shepley L. Ross, Il, graduate student in mathematics, University of Rochester, Rochester, New York, for his careful reviewing and helpful suggestions, ‘Tam grateful to Solange Abbott for her excellent typing, {am pleased to record my appreciation to Editor Gary Ostedt and the Wiley staff for their constant helpfulness and cooperation. ‘As on several previous occasions, the most thanks goes to my wife who offered encouragement, understanding, patience, and help in many different ways. Thanks, Gin, Shepley L. RossPART ONE FUNDAMENTAL METHODS AND APPLICATIONS ‘One Differential Equations and Their Solutions LL Clasication of Diferential Equations; The Origin and Application LZ Solutions 1.3 Initia-Value Problems, Boundary-Valve Problems, and Existence of Solutions ‘Two First-Order Equations for Which Exact Solutions Are Obtainable 21 Exact Differential Equations and Integrating Factors 22. Separable Equations and Equations Kedveibie Ths Form ~ 23 Linear Equation and Bemoull Equations 24 Special Imegrating Fators and Transformations Three Applications of First-Order Equations 4.1. Orthogonal and Oblique Trajectories 32 Problemsin Mechanics 433. Rate Problems licit Methods of Solving Higher-Order Linear ial Equations AL Basic Theory of Linear Difeential Equacions 42. The Homogeneous Linear Equation with Constant Coeficiente —— CONTENTS —— 102 102 1543. "The Method of Undetermined Coefficients 44 Variation of Parameter: 43. The Cauchy-Euler Equation 46 Statements and Proofs of Theorems on the Second-Order Homogeneous Linear Equation ° Five Applications of Second-Order Linear Differential Equations, with Constant Coefficients 5.1 The Differential Equation ofthe Vibrations of @ Mau on a Spring 52. Free, Undamped Motion 53. Free, Damped Motion 54 Forced Motion 53. Resonance Phenomena 5.6 Blectric Cirewie Problems Six Series Solutions of Linear Differential Equations, 6.1 Power Series Solutions About an Ordinary Point 6.2 Solutions About Singular Points; The Method af Feobenivt 63 Besse’ Equation and Bessel Functions Seven Systems of Linear Differential Equations 2.1 Dilfeentiat Operators and an Operator Method 72 Applications 7.3. Basic Theory of Linear Sysems in Normal Form: Two Equations in Two Unknown Fonction: 74 Homogeneous Linear Systems with Constant Cocfisients: Two Equations in Two Unknown Fanctiont 23. Matrices and Vectors 16 The Matri Method for Homogeneous Linear Sytems with Constant Coeficiente: ‘Two Equations in Two Unknown Functions The Matri Method for Homogencous Linear Systems with Constant Coeficints: ‘Equations in e Unknown Functions Eight Approximate Methods of Solving First-Order Equations 8.1. Graphical Methods 82 Power Series Methods 83. The Medhod of Succesive Approximations 84 Numerical Metheds ‘Nine The Laplace Transform 9.1 Definition, Existence, and Basie Properties ofthe Laplace Transform 92. The Invere Transform ant the Convolution 93 Laplace ‘Transform Selition of Linear Differential Equations with Constant Cocticient 94 Laplace Transform Solution of Linear Systems 179 19 132 199 au. a 2 23 232 264 264 28 290 sot a2 346 355 377 37 se 390 394 an su a1 a 453| cones PART TWO FUNDAMENTAL THEORY AND FURTHER METHODS Ten Existence and Uniqueness Theory 461 | 101 Some Conceps fiom Real Funetion Theory 461 } 102 ‘The Fundamental Bastence and Uniqucness Theorem 473 } 105 Dependence of Solutions on Inti Conditions and an the Funeti 8 104 Extence and Uniguenes Theorems fo Systems and Higher-Order Equations 498 Eleven The Theory of Linear Differential Equations 305 HA Touroduetion 50s 12. Basie Theory ofthe Homogencous Linear Sytem 510 13. Further Theory of the Homogencous Linear System m2 14 The Nonkomogenecus Lineae System 53, 1.3. aie Theory ofthe wh-Order Homogeneous Linea Differasial Equation 48 116 Further Poperis ofthe wh-Order Homogencous Lica Diferetial Equation 558 117. "The mh-Order Nenhomogencous Linear Equation 569 1B. Sturm Theory 373 Twelve Sturm-Liowville Boundary-Value Problems and Fourier Series 588 i 1241, Sturm-Liovle Problems ses j 122. Orthogonal of Characters Factions on ‘ 123 The Bepason oa Fenton na Ser of Ontonormal Function oor { 124 “Tgpeemerc Fourier Sees 8 { Thirteen Nonlinear Dierntial Equations on \ 182. Phase Plane Pats and ii! Poi see | 182 Gaal Pons and Phe Liner Systeme bea 153. Grea Pom and Fao Nonna Sens ost 134 Limit Gees anl Peo Satan esa 133 The Meth o Key a Boga bt Pa Fourteen Partial Differential Equations 15 V4 Some Base Concepts and Examples ns 162. Treated of Sparen of Vrs m 143. Canonical Foo Second-Order Linear quai with Constant Coefiens 148 THA hn Vale bien; Chgracecs : Bt Appendices m Answers a ‘Suggested Reading 807 Index 803———PART ONE——— FUNDAMENTAL METHODS AND APPLICATIONS—— CHAPTER ONE—— Differential Equations and Their Solutions ‘The subject of diflerential equations constitutes a large and very important branch of ‘modern mathematics, From the early days of the calculus the subject has been an area of great theoretical esearch and practical applications, and it continues to beso in our day. This much stated, several questions naturally arise. Just what is a diferent equation and what does it signify? Where and how do differential equations originate and of what use are they? Confronted with a differential equation, what does one do il, how does one do it, and what are the results of such activity? These questions indicate three major aspects of the subject: theory, method, and application. The Purpose ofthis chapter is to introduce the reader tothe basic aspects ofthe subject and. atthe same time give a brief survey of the three aspects just mentioned. In the course of, ‘the chapter, we shall find answers to the general questions raised above, answers that ‘will become more and more meaningful as we proceed with the study of diferential ‘equations in the following chapters. 1.1 CLASSIFICATION OF DIFFERENTIAL EQUATIONS; THEIR ORIGIN AND ‘APPLICATION A. Differential Equations and Their Classi DEFINITION Anequation inzolving derivatives of one or more dependent variables with respect to one or ‘mare independent variables is called a differential equation.* in connection with is bac defiaon, we do ror nclade inthe class of dori equations those tsqustons that ate actly tive tdontes For example, wa excude sich expresons a, Lier = set, Law = ober ad 50 forts * os roar 34 DIFFERENTIAL EQUATIONS AND THEE SOLUTIONS D Example 1.1 For examples of differential equations we lst the following #4 (2) S+n(Z) on dts dix 1 4 58 1 se msint a a3) Pu Ou Ou ee tg a From the bref lis of diflerential equations in Example 1.1 itis clear that the various variables and derivatives involved in a diflerential equation can occur in a variety of ways, Clearly some kind of classification must be made. To begin with, we classify example 1.2 Equations (1.1) and (1.2) aré ordinary differential equations, In Equation (I.1) the variable xis the single independent variable, and yisa dependent variable. In Equation (1.2) the independent variable is ¢, whereas x is dependent DEFINITION A differential equation inoolving partial derivatives of one or more dependent variables ‘with respect to more than one independent variable is called a partial differential equation. > Example 1.3 Equations (1.3) and (1.4) are partial differential equations. In Equation (1.3) the variables sand tare independent variables and visa dependent variable. In Equation (1.4) there are three independent variables: x, y, and 2; in this equation wis dependent. We further classify differs ial equations, both ordinary and partial, according to the ‘order of the highest derivative appearing in the equation. For this purpose we give the following definition1.1 CLASSIFICATION OF DHFERENTIAL EQUATIONS; THEI ORIGIN AND ARUCATION 5 DEFINITION ‘The order of the highest ordered derivative involved ina differential equation iscalled che order of the diferental equation D Example 1.4 ‘The ordinary differential equation (1.1) is of the second order, since the highest derivative involved is a second derivative. Equation (1.2) is an ordinary differential equation of the fourth order. The partial diferentia equations 1.3) and{(1.4)areof the first and second orders, respectively. Proceeding with our study of ordinary differential equations, we now introduce the important concept of linearity applied to such equations. This concept will enable us 0 classify these equations still further. DEFINITION A Yinear ordinary differential equation of order n, in the dependent variable y and the independent variable x, is an equation that is in, or can be expressed in, the form anty de ay 7 y aa) + UD act +7 OyasCaD Gy + aly = Bla, where ag isnot identically zero, Observe (1) that the dependent variable yand its various derivatives occur to the first degree only, (2) that no products of y and/or any of its derivatives are present, and (3) that no transcendental functions of y and/or its derivatives occur. D> Example 1.5 ‘The following ordinary differential equations are both linear. In each case y is the dependent variable. Observe that y and its various derivatives occur tothe first degree only and that no products of y and/or any of its des fy 6564 bye BrtsZsoy-o, as ay Py odd oe Gate grt Zax a) DEFINITION ‘nonlinear ordinary dierent equation san rdlnar diferente equation hats not6 DirremenTiAt EQUATIONS AND THEIR SOLUTIONS. > Example 1.6 ‘The following ordinary diferential equations are all nonlinear: ay a an ey Pres as) a a | Equation (17) i nonlinear because the dependent variable y appears to the second degree in the term 6y", Equation (1.8) owes its nonlinearity to the presence of the term Sidy/dx)?, which involves the third power of the First derivative. Finally, Equation (1.9) is nonlinear because of the term Sy(dy/dx), which involves the product of the dependent variable and its first derivative Linear ordinary differential equations are further classified according to the nature ‘of the coefficients of the dependent variables and their derivatives. For example, Equation ({.5)issaid to be linear with constant coefficients, while Equation (1.6)is linear with variable coeficients B. Origin and Application of Differential Equations Having classified diferential equations in various ways, let us now consider briefly where, and how, such equations actually originate. In this way we shall obtain some indication of the great variety of subjects to which the theory and methods of differential equations may be applied Differential equations occur in connection with numerous problems that are ‘encountered inthe various branches of science and engineering. We indicate a few such problems in the following list, which could easily be extended to fill many pages. rocket, sat oF planet. 1, “The problem of determining the motion of a projecti 2. The problem of determining the charge or current in an electric circuit. 3. The problem of the conduction of heat in & rod or in a slab. 44. The problem of determining the vibrations of a wire or a membrane. 5. The study of the rate of decomposition of a radioactive substance or the rate of growth of a population. 6 The study of the reactions of chemicals. 7. The problem of the determination of curves that have certain geometrical properties. ‘The mathematical formulation of such problems give rise to differential equations. But ust how does this occur? In the situations under consideration ineach of the above problems the objects involved obey certain scientific laws. These laws involve various rates of change of one or more quantities with respect to other quantities. Let use12 sownons 7 call that such rates of change are expressed mathematically by derivatives. In the ‘mathematical formulation of each of the above situations, the various rates of change are thus expressed by various derivatives and the scientific laws themselves become ‘mathematical equations involving derivatives, tha is, differential equations. In this process of mathematical formulation, certain simplifying assumptions generally have to be made in order that the resulting differential equations be tractable. For example, ifthe actual situation ina certain aspect of the problem is ofa relatively complicated nature, we are often forced to modify this by assuming instead an approximate situation that is of a comparatively simple nature. Indeed, certain relatively unimportant aspects of the problem must often be entirely eliminated. The result of such changes from the actual nature of things means that the resulting diflerential equation is actually that of an idealized situation. Nonetheless, the information obtained from such an equation i of the gfeatest value to the scientist A nnatural question now isthe following: How does one obtain useful information froma differential equation? The answer is essentially that itis possible todo s0, one solves the differential equation to obtain a solution; if this is not possible, one uses the theory of differential equations to obtain information about the solution. To understand the meaning ofthis answer, we must discuss what is meant by a solution of 4a diflerential equation; this is done in the next section. Exercises differential Cassify each of the following differential equations as ordinary or parti tqustiossatethe orderofeachequatoncand deerme wheter heaton oder Smideration ea oc nonin 1 La ety ere. 2 Sg syesinx Boo 4 dye yarn 2 (Eee 0 (t= fn 1.2 SOLUTIONS A. Nature of Solutions We now consider the concept of a solution of the nth-order ordinary differential ‘equation,8 owreteniat eQuaTiONs AND THUR SOLUTIONS DEFINITION fondo ft =n a dy ty where Fis areal function ofits (n+ 2) arguments x, 1% oo 1. Let f be a reat function defined forall x ina real inereal I and hasing an nth derivative (and hence aso all lower ordered derivatives) forall x€ 1. The function fis called an explicit solution ofthe differential equation(J.10)on 1if i fulfill the following ‘v0 requirements: FL SOM Loe LCT “ is defined for all x € 1, and FL £9, S'.-- SE] =O 8 Jor all x 1. That is, the substitution of f(x) and its various derivations for y and ts corresponding derivatives respectcely, in (I-10) reduces (10) to an identity on 2. relation olx,») = Os caled an implicit solution of 10) this relation defies cat least one eal function f ofthe variable x on an interoal sack hat this function i an explicit solution of (1.10) on this interval 3. Botkexplcit solutions and implicit solutions will usually be called simply solutions Roughly speaking, then, we may say that a solution ofthe differential equation (1-10) isa relation explicit o implicit—between x and y, not containing derivatives, which identically satisfies (1.10) D Example 1.7 “The function f defined for all real x by fle) = 2sin x + 3008 x aan is an explicit solution ofthe differential equation ay Shtys0 ary fora ral x. First note that is defined and has.a second derivative forall real x. Next ‘observe that J") = 208 x —3 sin x, Sta) = ~2sin x— 3.08 x. ‘Upon substituting f(x) ford? y/dx? and f(x) for yin the diferental equation (1-12), it reduces to the identity (~2sin x ~ 3.08 x) + (2 sin x-+ 3.08 x) = 0,12 sorutions 9 ‘which holds for all real x. Thus the function f defined by (1.11) isan explicit solution of the differential equation (1.12) forall real x. > Example 1.8 ‘The relation x+y? 2550 13) is an implicit solution of the differential equation ay x4 yZn0 a9 ‘on the interval I defined by —5 Example 1.17 Problem. Find a solution f of the diferemtial equation ay Ram (2y such that at x = I this solution f has the value 4, Explanation, First let us be certain that we thoroughly understand this problem. We seck a real function f which fulills the two following requirements: 1. Thefonetion f must satisfy the differential equation (1.21) Thats, the function f must be such that J") = 2x fr ll real x ina ral interval 2. Thefunction f must havethe value 4atx = 1. Thatis the function f must besuch that £0) = 4 Notation. The stated problem may be expressed in the following somewhat abbreviated notation: ay Bory, ax yO) = In this notation we may regard y as representing the desired solution. Thea the differential equation itself obviously represents requirement 1, and the statement (l= 4 stands for requirement 2. More specifically, the notation y(1) = 4 states that the desired solution y must have the value 4 at x = 1; that is, y= 4 at x Solution. We observed in Example 1.9 that the difleential equation (1.21) has a ‘one-parameter family of solutions which we write as yextte, (22) where cisan arbitrary constant, and that each ofthese solutions satisfies requirement L Let us now attempt to determine the constant ¢ so that (1,22) satisfies requirement 2, thats, y = 4.at x= 1, Substituting x = 1, y = 4 into (1.22), we obtain 4 = | + ¢,and hhence ¢ = 3, Now substituting the value c= 3 thus determined back into (1.22), we obtain yan,16 { } DIFFERENTIAL EQUATIONS AND THER SOLUTIONS which is indeed a solution ofthe differential equation (1.21), which has the value 4 at In other Words, the function f defined by Soy +3, satisfies both of the requirements set forth in the problem. Comment on Requirement 2 and its Notation. Ina problem of this type, require- ment 2 is regarded as a supplementary condition that the solution of the diferential equation must also satisy. The abbreviated notation y(1) = 4, which we used to express this condition, is in some way undesirable, but ithas the advantages of being both customary and convenient. In the application of both first-and higher-order differential equations the problems most frequently encountered are similar to the above introductory problemin that they involve hoth a differential equation and one or more supplementary conditions which the solution of the given differential equation must satisfy. 1C all of the associated ‘supplementary conditions relate to one x value, the problem is called an intial-zalue ‘rable (or one-point boundary-valve problem). If the conditions relate to ro dif- ferent x values, the problem is called a two-point boundary-value problem (or simply a boundary-valve problem). We shall illustrate these concepts with examples and then ‘consider one such type of problem in detail. Concerning notation, we generally employ abbreviated notations for the supplementary conditions that are similar to the ab- breviated notation introduced in Example 1.11 > Example 1.12 0) p= -4 This problem consists in finding a solution of the differential equation Which assumes the value 3at x = | and whose first derivative assumes the value —4 at Both of these conditions relate to one x value, namely, x = I. Thus this is an initial-value problem, We shall see later that this problem has @ unique solution. Example 1.1313 IMTIALVALUE Prom, soUNDARY.vALUE PROBLEMS 17 In this problem we again seck a solution of the same differential equation, but this time the solution must assume the value 1 at x =O and the value $ at x = n/2. That is, the conditions relate to the two different x values, 0 and 1/2. This is a (two-point) boundary-value problem. This problem also has a unique solution; but the boundary value problem i: Paya, YO=I yln)= 5, has no solution at all! This simple fact may lead one to the correct conclusion that boundary-value problems are not to be taken lightly! We now turn toa mote detailed consideration of the initial-value problem fora frst- order dilferential equation. DEFINITION Consider the first-order differential equation és =f (123) where {is «continuous function of x and y in some domain® D ofthe xy plane: and let (50, 3) be a point of D. The initial-value problem associated with (7.23) is to find ‘solution ofthe differential equation (1.23) defined on some real interval containing Xp, ‘and satisfying the inital condition $600) = Yo! In the customary abbreviated notation, this inital-ualue problem may be written ‘ eo Seah Wx) = Yo. ‘osolvethis problem, wemust inda fonction that not onl satsfos the diferentiat equation (1.23) but that also satsfes the initial condition that it haste value y when x has value x. The gsometrc interpretation of th initial condition, and hence ofthe centr initalvalye problem, i casly understood. The graph of the dsied solution must pas through the pont with coordinates (x, ye). That i, interpreted geometi- cally, the initial-value problem st find an integrl curve ofthe diferntial equation (1.28) that passes theough the point (x) “Themethod of actually finding the desired solution depends upon thenatue of the differential equation ofthe problem, that i, upon the form off}, Cetain special types of iflerential equations have a one-parameter family of solutions whose equation may be found exactly by following definite procedures (ee Chapter 2). the Sifferetial equation ofthe problemi of some such special type, one st obiain the equation of ts one-parameter family of solutions and then applies the initial condition "Nema an open, Connected at For ove unarliar wih ach concept, O may be eed the interior of Some simple closed cure inthe plore.18 wrenentia. tQuaTions AND THEIR SOLUTIONS {othisequationin an attempt to obtaina“particular” solution @ that satisfies the entire initial-value problem. We shall explain this situation more precisely in the next paragraph. Before doing so, however, we point out that in general one cannot find ‘he equation of a one-parameter family of solutions of the differential equation; approximate methods must then be used (see Chapter 8) Now suppose one can determine the equation als, no) = 0 2) cof a one-parameter family of solutions of the differential equation of the problem. ‘Then, since the initial condition requires that y= yp at x= xo, we let x = Xp and 13 Vo in(1.24) and thereby obtain Bo.¥od = Solving this for cin general we obtain a pasticular value of ¢ which we denote here by x, Wenow replace the arbitrary constant c by the particular constant cy in (1.24, thus ‘obtaining the particular solution 4s, Yea) = 0. ‘The particular explicit solution satistying the two conditions (differential equation and {niial condition) of the problem is then determined from this, if possible. We have already solved one initial-value problem in Example 1.11. We now give another example in order to illustrate the concepts and provedures more thoroughly > Example 1.14 Solve the initial-value problem (125) (126) given that the differential equation(1.25)has a one-parameter family of solutions which -may be written in the form, (127) ‘The condition (1.26) means that we seek the solution of (.25)such that y = 4at x = 3 ‘Thus the pair of values (3,4) must satisy the relation (1.27) Substituting x = 3 and 4 into (1.27), we find vay 9+ 16=c? or c ‘Now substituting this value of c? into (1.27), we have xt yha 2s Solving this for y, we obtain yas VB—H ‘Obviously the positive sign must be chosen to give y the value +4 at x = 3. Thus the function f defined by fl) = JB ~Sex Example 1.15 Con inte den A alte Bovey, wl Let us apply Theorem 11. We first check the hypothesis. Here f(x, y) = x? + y* sn Ty oh of ton and fy ae ots yon of the xy plane. The initial condition y(1) = 3 means that xo = Land yp = 3,and the point (1, 3) certainly lies in some such domain D. Thus all hypotheses are satisfied and the conclusion holds. That is, there is a unique solution @ of the diferential equation dyjdx = x* + y?, defined on some interval |x — 1] Example 1.16 Consider the tio problems: ayy L Rage nr yoy 2 Bete rome Here fx Sewn zin ane ‘These functions are both continuous except for x = 0 (that is, along the y axis). In problem 1.x = 1, yo = 2. The square of side I centered about (1,2) does not contain the y axis, and so both f and df/éy satisty the required hypotheses inthis square. Is interior may thus be taken to be the domain D of Theorem 11; and (1,2) certainly les within it, Thus the conclusion of Theorem 1.1 applies to problem 1 and we know the problem has @ unique solution defined in some sufficiently small interval about xp = J Now let us turn to problem 2. Here xy = 0, yo = 2.At this point neither f nor dff2y are continuous. In other words, the point (0,2) cannot be included in a domain D where ‘the required hypotheses are satisfied. Thus we can no: conclude from Theorem 1.1 that22 overex, equations ANO THEHE SOLUTIONS problem 2 has a solution, We are not saying that it does not have one. Theorem 1.1 simply gives no information one way or the other. Exercises 1. Show that ya det +20 yo)= 6, YO) =2. Is y = 2% + 4e-* also a solution of this problem? Explain why or why not 2. Given that every solution of uae Rt ya2 may be written in the form y = (x? 4 de~*, for same choice of the arbi constant c solve the following initial-value problems: ay oF ® yee @ Pa yeres, ty Pyare, 0) = 2 wale +3. 4, Given that every solution af ey ay _ oy 8 yao ‘may be written in the form pecs tee, for some choice of the arbitrary constants c, and 3, solve the following initial- value problems: ®y @ Py _ay & -Z- nyo 10) yOr6 4. Every solution ofthe diferential equation @y Gitrno( Reece Be esc ee OSE aEnREEE 13 INITAL-VALUE PROBLEMS, ROUNDARY-VALUE Poss 23 may be written in the form y = ¢, sinx + ¢; os, for some choice of the arbitrary ‘constants c, and c. Using this information, show that boundary problems (a) and. (©) possess solutions bur that (c) does-n0:. a © Baye yO) = 0, yO) = 1, Wen/2) 41 YR) = “ © Bayeo 30) =0, 0) 1 “ Given that every solution of Py yi dy gt Eh - 3 Sh 4 6% oye ‘may be writtenintheform y = ¢,x + ¢,x7 + ex? forsome choice of the arbitrary constan's ¢,,¢3, and ¢s, solve the initial-value problem consisting of the above dlifferential equation plus the three conditions yQ)=0, y@=2 y"'A=6 . Apply Theorem 1.1 to show that each of the following initial-value problems has 2 unique solution defined on some suficently small inverval |x ~ 1| 0. (8) Carefully graph the solution for which ¢ = 0. Then, using this particular staph, also graph the solutions for which ¢ = 1,¢ = 2, and ¢ = 3.=== CHAPTER TWO—— First-Order Equations for Which Exact Solutions Are Obtainable In this chapter we consider certain basic types of first-order equations for which exact solutions may be obtained by definite procedures. The purpose ofthis chapteristo gain ability to recognize these various types and to apply the corresponding methods of solutions. OF the types considered here, the so-called exact equations considered in Section 2.1 arcin asense the most basic, while the separable equations of Section 2.2are {na sense the “easiest.” The most important, from the point of view of applications, are the separable equations of Section 2.2 and the linear equations of Section 23. The remaining types are of various very special forms, and the correponding methods of solution involve various devices. In short, we might describe this chapter as a col- lection of special “methods,” “devices,” “tricks,” oF “recipes. in descending order of kindness! EXACT DIFFERENTIAL EQUATIONS AND INTEGRATING FACTORS, A, Standard Forms of Fitst-Order Differential Equations The first-order differential equations to be studied in this chapter may be expressed in cither the derivative form Pesan ey or he iteremia form Mos hax + NO Nd =O. an 2526 masronoer EQUATIONS FOR WHICH EXACT SOLUTIONS ARE OBTAINABLE ‘An equation in one of these forms may readily be written in the other form. For example, the equation ay ety & is ofthe form (2.1). It may be written (e+ y)ax+ aay which i of the form (22). The equation (sin + yds + (4 39) y =0, which i of the frm (22), may be written in the form (21) a8, In the form (2.1) itis clear from the notation itself that is regarded as the dependent variable and x as the independent one; but in the form (2.2) we may actually regard cither variable as the dependent one and the other as the independent. However, in this text, in all differential equations of the form (2.2) in x and y, we shall regard y as dependent and x as independent, unless the contrary is specifically tated B. Exact Differential Equations DEFINITION Let F be a function of two real variables such that F has cominuous first partial derivatives in a domain D. The total dillerental dF of the function F is defined by the formula 9 = ED an 4 FLED cs {for ait(x, )¢ D. > ixampe 21 ~ Let F be the function of two teal variables defined by Fis, y)= ay? + Bey for al real (x, 3). Then BFC sg gery, SFE : is 675, SRD e aey 4 20% ‘and the total differential dF is defined by F(x, 9) = (9? + 6x7y) do + Oxy + 2x) dy for all real (x)| | I | } { i | | i | 2.4 WACK DIFFERENTIAL EQUATIONS AND INTEGRATING FACTORS 27 DEFINITION The expression M(x, y) dx + NOs y) dy @3) is called an exact differential ina domain D if there exists a function F of two real vari- ‘ables such that this expression equals the total differential dF(x,y) for all (x,y) €D. That is, expression (2.3) is an exact differential in D if there exists @ function F such that 2PM es.) and FED 9 aces fue yal a cg cated ope aa age ce M(x, y)dx + N(x, ypdy = 0 seated ana trent eution > Example 2.2 The differential equation yi dx + 2xpdy=0 es is an exact diferential equation, since the expression y? dx + 2xy dy is an exact Example 23 We apply the exactness criterion (2.7) to Equations (2.4) and (2.5), introduced in Example 22. For the equation y? dx + 2xydy=0 (24) we have Mey) N(s,9) = 29, OMG 9) _ 9, _ ANU 9) ey a for all(x,»). Thus Equation (24) is exact in every rectangular domain D. On the other hand, for the equa pdx + 2xdy=0, as we have Mex y= NOs 9) = 2x aMix 9) antx,») a jon (2.8) isnot exact in any rectangular domain D, 142 for all(x, 9). Thus Equi12.1. BRACT DIFERENTIAL EQUATIONS ANO WwTEGaATING FACTORS 3 > example 24 Consider the differential equation Qxsin y+ ye") de + (x? cosy + 3y%e dy = 0. Here M(x y)= 2x sin y + ye N(x y)= x? 60s y+ 39%e, ames 9) ents.) ay a inevey rectangular domain D Ths his diferent equation is exc in vty such = 2x cos y + 3yte* = ‘These examples illustrate the use ofthe test given by (2-7)for determining whether or not anequation of the form M(x, y) dx + N(x, y) dy = Oisexact.Itshould be observed ‘that the equation must be in the standard form M(x, y) dx + N(x, y) dy = Oin order to Use the exactness test (27). Note this carefully: an equation may be encountered inthe nonstandard form M(x, ») dx = N(x, »)dy, and in this form the test (2.7) does not apply. C. The Solution of Exact Differential Equations ‘Now that we havea test with which to determine exactness let us proceed to solve exact differential equations. If the equation M(x, ypde + N(x, »)dy =O is exact in a reétangular domain D, then there exists a fonction F such that FFL») _ sayy oF») i SR Messy and 7 « Nix 9) forall ye. ‘Then the equation may be written BOD gy POD yay oxsingy dF 91= ae ae + y dy=0 orsimply dF (x, y) = 0. ‘The celation F(x, #} = cis obviously a solution of this, wherecis an arbitrary constant ‘We summarize this observation in the following theorem. ‘THEOREM 2.2 ‘Suppose the differential equation M(x, y) dx + NIx, y) dy = 0 satisfies the differentia. Dilley requirements of Theorem 2.1 and s exact wn a rectangular domain D. Then a one- parameter family of solutions of this differential equation is gioen by F(x, ») = ¢, where F isa function such that aF i, Ft FEDS Mex, 9 and : 2D Nix,y) for ail (x ye. ‘and ets an arbitrary constant.Fns1-0806R EQUATIONS FOR WHICH EXACT SOLUTIONS ARE OBTAINABLE Referring to Theorem 2.1, we observe that F(x, y) is given by formula (2.13) However, in solving exact differential equations its neither necessary nar desirable to Use this formule. Instead one obtains F(x, y) either by proceeding asin the proof of Theorem 2.1, Part 2,orby the so-called “method of grouping,” which willbe explained inthe following examples > Example 25 Solve the equation Bx? + dy) dx + Ox? + Iy)dy = 0. ‘Our first éuty is to determine whether or not the equation is exact, Here M(x, y) = 3x? + 4p, NOx 9) = 28? +29, Mix, 9) ant 9) a x for all real (x, yj and so the equation is exact in every rectangular domain D. Thus We must find F such that 66,9) oF») ey = Misys) = 38! + 44y- and = Nu yb = 208 429 From the fist of these, Fo = futon ax + 00)= foe 4+ 4ny) dx + 00) Bey FOOL, 259) _ aya , HOU? eur dy ‘Buc we must have Thus ‘Thus #(y) = ¥? + co, where cp is an arbitrary constant, and so : Fix, ax + Bet ty? +00ws AND intecRATING Factors 33. 24 Exact oUrimeNALEQUAN Hence a one-parameter family of solution is F(x, y) a4 ety + yb eg me} Combining the constants ¢y and cy we may write this solution as ety ts Pac, wheree = in generality by taking cy = 0 and writing $y) procedure, ois anarbitrary constant. Thestudent will observe that there sno oss y3, We now consider an alternative Method of Grouping. We shall now solve the differential equation of this exam ple by grouping the terms in such a way that its let member appears as the sum of certain exact differentials. We write the diferential equation Gxt + dxyl de + Ox? 4 2y)dy = 0 in the form 3x? dx + (dy dx + 2x? dy) + 2y dv = 0. We now recognize this as dix?) + d(2x?y) + diy?) where ¢ is an arbitrary constant, or dls? + Ixty + y?) = dle. a, From this we have at once xtadty tyme, Clearly this procedure is much quicker, but it requires a good “working knowledge” of differentials and a certain amount of ingenuity to determine just how the terms should be grouped. The standard method may require more "work and take longer, but its perfectly straightforward. [tis recommended for those who like to follow a pattern and for those who have a tendency to jump at conculsions. Just to make certain that we have both procedures well n hand, we shall consider an initial-value problem involving an exact differential equation, > Example 2.6 Solve the initial-value problem (De cos y+ Ixty) dx + (x? — x3 sin y= yhdy = 0, 0) = 2 ‘We fist observe thatthe equation is exact in every rectangular domain D, since aM») iny + 3x2 = ON) Spe = anny 4 act = Rs forall rel (x, 9)34. emst-onote EQUATIONS FOR WHICH EXACT SOLUTIONS ARE OBTAINABLE Standard Method. We must find F such that Fon 9) a) a My) = 2ec08 y+ 3x4 and ree (ea ee Bo Mag) =~ x7sin yp Then Fe,y)= f Mix, ix +6) = fexcosy +349 2x4 649 = eosy4yt 60), 2FO ys >, dal) ssiny tt OU But also FOI) _ Nyy yy ex =x gin FED Ns yaw yoy and so and hence oun B00, Thus Fes, y) = x eos y +29 2 + oo Hence a one-paraineter family of solutions is Fix, 9) = ¢,, which may be expressed as sPeosy + ty Applying the initial condition y = 2 when x ~ 0, we find ¢ = —2 Thus the solution of ‘he given initial-value problem is sPoos yey Method of Grouping, We group the terms as follows: ede = xtsin yds) + Bx4y dy #8 dy vey21 PRACT DUFERENTIAL EQUATIONS AND INTEGRATING FACTORS 35. ‘Thus we have atx? os ») + ax*y) — (5) = dos and so ew yeey ee sx one-parameter family of solutions of the differential equation. Of course the intial condition y(0) = 2 again yields the particular solution already obtained D. Integrating Factors Given the differential equation M(x 3) dx + Nex y)dy =0, aNtxy) M(x. 9) v then the equation is exact and we can obtain a one-parameter family of solutions by fone of the procedures explained above, But if M(x, 9), ONG 39) 4 ONE ey ex then the equation is nor exact and the above procedures do not apply. What shall we do ‘in such a case? Perhaps we can multiply (he nonexact equation by some expression that will ransform it into an essentially equivalent exact equation. If so, we can proceed to * solve the resulting exact equation by one of the above procedures. Let us consider ‘again the equation yax + 2xdy =0, es which was introduced in Example 2.2. In that example we observed that thisequation is not exact. However, if we multiply Equation (2.5) by y, itis transformed into the ‘essentially equivalent equation vy dx + xy dy =0, e4 hich is exact (see Example 2.2). Since this resulting exact equation (2.4) is integrable, wwe call y an integrating factor of Equation (2.5). In general, we have the following definition: DEFINITION If the diferencia equation Mex. yh dx + Nlxs)dy = 0 1a36 FIRST.ORDUR EQUATIONS FO WHHCH EAACT SOLUTIONS ARE OBTAINABLE is not exact in a domain D but the differential equation HG Mx, 9) dx + ls SINGH WD dy 15) isexact nD, then x, ybis called an integrating factor of the differential equation (2.14). > Example 2.7 Consider the diferent ‘equation Gy + day?) de + (2x 4 3x y dy ‘This equation is ofthe form (2.14), where 2.16) Mix, y) = 3y + 4xy?, N(x, y) = 2x + 3x7y, sie aoa aM by ox 0, Equation (2.16) is not exact in any rectangular exc for (such that 2x9 + omain D. Let u(x, ») x2y, Then the corresponding diflerential equation of the form (2.15) is Gx2y? $ay?) de + Qx°y + ety) dy, ‘This equation is exact in every rectangular domain D, since Ante, YMOx 9) Peau oy ice ay ney? = Mis for al real (x,y), Hence u(x, y) = x?y isan integrating factor of Equation (2.16), Multiplication of @ nonexact diferential equation by an integrating factor thus transforms the nonexact equation into an exact one, We have referred to this result- ing exact equation as “essentially equivalent” to the original. This so-called essentially equivalent exact equation has the same one-parameter family of solutions asthe non- exact original. However, the multiplication of the original equation by the integrating factor may resultin ether (I) the loss of (one oF more) solutions ofthe origina, or (2}the ‘ain of (one or more) functions which are solutions ofthe “new” equation but nar of the Criginal, or (3) both of these phenomena, Hence, whenever we transform a nonexact {equation into an exact one by multiplication by an integrating factor, we should check carefully to determine whether any solutions may have been lost or gained. We shall illustrate an important special case of these phenomena when we consider separable equations in Section 22. See also Exercise 22 atthe end of this section. ‘The question now arises: How isan integrating factor found? We shall not attempt to answer this question at this time. Instead we shall proceed to a study of the important ‘lass of separable equations in Section 2.2and linear equations in Section 2.3, We shall See that separable equations always possess integrating factors that are perleetly{LY ERACT DIFFERENTIAL EQUATIONS AND INTEGRATING Factors 377 ‘obvious, while linear equations always have integrating factors of a certain special orm, We shall return to the question raised above in Section 2.4. Our object here has ‘been merely to introduce the concept of an integrating factor. Exercises In Exercises 1-10 determine whether or not each ofthe given equations is exact; solve those that are exact. Gx + 29)dx4 Qe Ady =O. (07 + 3)de + Oxy — 4) dy =. (xy + Ide +O + ay) dy =O. Gey sad — Wr td (6xy + 29? de + Bx? + dxy - .dy=O. (0+ Neos dr + 20sin rd? = 0. (yse2? x + see xtan x) dx + (tan x +29) dy = 0 G)aGinjond » (eter (SP)enn reat este 10 WL a atts! —yayea Solve he nivale problems in Entei 11-16 1. Qxy-Ddx +7 +4)dy=0, yl =2 12. Gx4y? =p 42a) dx Rey — Say? + Ddy=O, -2)= 13. Qysin xcosx + y*sins)dx + (sin? x—27c0s x) dy m0, y(0) 3 Get b2e + yde He HI HO, O=E aan ene 15, Jee (FSR ayeo nan 20 ett ga 16, Ede BO yn, y=8 = y 17. In each of the following equations determine the constant A such that the equation is exact, and solve the resulting exact equation: fa) (x? + 3xyide + (Ax? + 4y) dy =0, wy (deed)ane(SH)ayu38 {RST-ORDER EQUATIONS FOR WHICH EXACT SOLLIIONS ARE OBTAINABLE 18. In each of the following equations determine the constant A such that the equation is exact, and solve the resulting exact equation: (a) (Axty + 2y*) de + 60° + dry) dy = 0. © (S+h)e+(S-Lo 19, Incach ofthe following equations determine the most general function N(x, ) such that the equation exact: 0. (a) (2 +29") de + NG Way =O. (b) (x#y"F + xy) dx + NG yb dy =O. 20, In each of the following equations determine the most general function M(x, ») such that the equation is exact: (a) Moxy) dx + xy? + xtyhdy (0) M(x, y) dx + Qye" + ye dy 21, Consider the differential equation (4x + 3y?)dx + Day dy =0. (a) Show that this equation is not exact (©) Find an integrating factor of the form x", where n is a postive integer. (€) Multiply the given equation through by the integrating factor found in (6) and solve the resulting exact equation. 22, Consider the differential equation (2 + 2x) de = x? dy =0, (a) Show that this equation is not exact. (b) Multiply the given equation through by y*, where m isan integer, and then determine m so that y" is an integrating factor of the given equation. (6) Multiply the given equation through by the integrating factor found in (b) and solve the resulting exact equation. (4) Show that y = 0s a solution of the original nonexact equation but is not a Solution of the essentially equivalent exact equation found in step (c (¢) Graph several integral curves of the original equation, including all those ‘whose equations are (or can be written) in some “special” form. 23. Consider a differential equation ofthe form Ly xf? + y?)) dx + ofl? + 94) — x) dy = (2). Show that an equation of this form is not exact. (©). Show that 1/(x? + y?)is an integrating factor of an equation ofthis form. 24, Use the result of Exercise 2(b) to solve the equation [yb x08 + 927] de + O96? + 92) ad dy 2.122 SEPARABLE EQUATIONS AND EQUATIONS REDUCIELE TO ms FORM 39. 2.2 SEPARABLE EQUATIONS AND EQUATIONS REDUCIBLE TO THIS FORM ‘A. Separable Equations DEFINITION An equation of the form FO)G(y) dx + Sls) dy = 0 can 7 with variables separable or simply a separable equ is called an equat For example, the equation (x — 4)y4 de — xy? ~ 3) dy = O is a separable equa tion {In general the separable equation (2.17) is not exact, but it possesses an obvious integrating factor, namely 1/f(2)G(y). For if we multiply Equation (217) by this expression, we separate the variables, reducing (2.17) to the essentially equivalent equation FO) te 4 3) ay m0 as) Fey" 6) ‘This equation is exact, since ofa ]-0- 2 [a oy | FO). ax LG), Denoting F(x)/f(x) by M(x) and g(y)/G(y) by N(y) Equation (2.18) takes the form Mats NOj tp = 0 Shee iss funchon ef tony Sods uncon o yon acerca tet Sones way uno fuera [aoa 19 here cis the arbitrary constant. Thus the problem of finding such a family of solutions of the separable equation (2.17) has reduced to that of performing the integrations indicated in Equation 2.19). Its in this sense that separable equations ae the simplest firstorder differential equations. Since we obtained the separated exact equation (218) from the nonexact equation (2.17) by multiplying (2.17) by the integrating factor 1/(x)G4)), solutions may have ‘been Jost or gained in this process. We now consider this more carefully. In formally ‘multiplying by the integrating factor 1/(s)G(y), we actually divided by /(2)G(y). We did this under the tacit assumption that neither /(x) nor G(y) is zero; and, under this assumption, we proceeded to obtain the one-parameter family of solutions given by (2.19), Now, we should investigate the possible loss or gain of solutions that may have ‘curred in this formal process. In particular, regarding yas the dependent variable as usual, we consider the situation that occurs if Gt) is zero. Writing the original dif- ferential equation 2.17) in the derivative form Flag a + FIG) = 0,40 "RST-ORDER EQUATIONS FOR WHICH EXACT SOLUTIONS ARE DBTANABLE we immediately note the following: If yy is any real number such that G(y) = 0, then = Yo i a (constant solution of the original differential equation; and this solution ‘may (or may not) have been lost in the formal separation process. Im finding a one-parameter family of solutions a separable equation, we shall always make the assumption that any factors by which we divide in the formal ‘separation process are not zero. Then we must find the solutions y = yp of the equation G(s) = 0 and determine whether any of these are solutions of the original equation which were lost inthe formal separation process. D Example 2.8, Solve the equation (= dy dx? dy =0. ‘The equation is separable; separating the variables by dividing by x*y*, we obtain Fax dx O77 = 3 dy = 0. Integrating, we have the one-parameter family of solutions where isthe arbitrary constant. In dividing by x?y* in the separation process, we assumed that x° ¥ O and y* # 0, Wenow consider the solution y = Oof y* = 0. Itisnota member of the one-parameter family of solutions which we obtained, However, writing the original diferential equation of the problem in the derivative form dy _ @—4y* & SG? ‘obvious that y = 0 sa solution ofthe original equation. We concifde that itis 2 solution which was lost in the separation process. > Example 2.9 Solve the initial-value problem that consists ofthe differential equation sin yde + (x4 + 1) cos ydy =0 (229) and the initial condition vt) =5. ean We first obtain a one-parameter family of solutions ofthe differential equation (2.20),212. SEPARABLE EQUATIONS AND FQUANONS sEDLCBLE TO THs roms 41 Separating the variables by dividing by (x? + 1) sin y, we obtain x cory ayn de 4 EP ay m0, Fatt Say? Thus xéx | (oor ys, Pa? | siny Om where co is an arbitrary constant, Recall that ft emuiee an win tt Then, carrying out the integrations, we find Fln(x? +1) + Injsin y] =e, (2.22) We could leave the family of solutions inthis form, but we can putt ina neater form in the following way. Since each term of the let member of this equation involves the logatithm of a function, it would seem reasonable that something might be accom- plished by writing the arbitrary constant cy in the form ln |c,|. This we do, obtaining Hints? +1) + In [sin ») = tn Ley) Multiplying by 2, we have In(x? + 1) +2 In|sin y| = 2m fe. Since 24m sin yl = In (sin »)?, and 2imiey| =Incf = Ine, where c=d20, we now have In (x? 41) + Insin? Since In A 4+ In B = In AB, this equation may be written In(x? + sin? y = Ine. From this we have at once (+ Dsin? y Clearly (2.23) is ofa neater form than (2.2) In dividing by (x? + I)sin y in the separation process, we assumed that sin y # 0 Now consider the solvtions of sin y=0, These are given by y= nr (n=O, 21, £2,...), Writing the original differential equation (220) in the derivative'form, iis lear that each ofthese solutions y = n(n = 0, #1, $2,...)of sin y = Disa constant solution of the original differential equation, Now, each of these constant solutions 223)42 | FST-ORDER EQUATIONS FOR WHICH EXACT SOLUTIONS ARE OBTAINABLE y= mis a member of the one-parameter family (2.23) of solutions of (20) for ¢ = 0, ‘Thus none of these solutions was lost in the separation process. ‘We now apply the initia! condition (2.2) to the family of solutions (2.23). We have (1 + Hsin? land so.c = 2. Therefore the solution of the initial-value problem under consideration is Ge? + Dsin? y = 2, B. Homogeneous Equations We now consider a class of differential equations that can be reduced to separable ‘equations by a change of variables. DEFINITION The first-order differential equation M(x, 9) dx + N(x, y) dy = 0 ie said to be homo: _peneous when written inthe derivative form (dy dx) = f(s, )) there exist function 9 Such that f(y) an be expressed inthe form g( 3/3). Example 2.10 ‘The differential equation (x* ~ 3y*) dx + 2xy dy = Ois homogencous. To soe this, We first write this equation in the derivative form Now observing that veneneey ESQ 2xy Qe dy 2x) 20, wie se that the diferentil equation under consideration may be written as dy _3/y\_ 11) #-3(:) 2G) in which the right member is of the form g(y/'x) for a certain function g Example 2.11 The equation + SPF F de edy 0 is homogeneous: When writen inthe form ye VPTP am +122. SEPARABLE EQUATIONS AND EQUATIONS REDUCIBLE TO THIS FoRM 43, the right member may be expressed as 14 vEar ve depending on the sign of x. This is obviously of the form g(y/2). Before proceeding to the actual solution of homogeneous equations we shall consider slightly different procedure for recognizing such equations. A function F is called homogeneous of degree nif Fits, 3) = F(x, 3) This means that if tx and ty are substituted for x and 4, respectively, in F(x ), and if ris then factored out, the other factor that remains isthe original expression F(x, 9) tse For example the function F given by F(x, y) = x? + y? is homogeneous of degree 2, since F(x, 19) = (9)? + (0)? = FG? + 97) = PFO). "Now suppose the functions M and NV in the differential equation M(x, y)dx + N(x, )dy = 0 are both homogencous ofthe same degree n. Then since M(¢x, (9) = eMC, 9) if we let t= 1s, we hive 1 6} Mex) u(bsbs) ule and from this we at once obtain ses () mld nesre(’)"n(u2) Now writing the differential equation M(x, y) dx + N(x») dy = On the form ay _ Mts») a” NEY)" Likewise, we find we find re Clearly the expression on the right is of the form g(/), and so the equation M(x, ») dx + N(x, y) dy = 0 is homogeneous in the sense of the original definition of homo-44 sinst-onven eQuaTIONS FOR WHUCH EXACT SOLUTIONS ARE OBTAINABLE geneity, Thus we conclude that if M and N in M(x, y)dx + N(x, y}dy = 0 are both homogeneous functions of the same degree n, then the diferential equation is & homogeneous differential equation. Let us now look back at Examples 2.10 and 2.11 in this light. In Example 2.10, M(x, ») = x? ~ 3y? and N(x, ») = 2xy. Both M and N are homogeneous of degree 2. ‘Thus we know at once that the equation (x ~ 3y7) dx + 2xy dy = Ois a homogeneous equation. In Example 2.11, M(x, y)= y+ /37 + yFand N(x, 9) = —x. Clearly Nis homogeneous of degree 1. Since Mlix.te) = t+ JOP FOP = dy + fe we see that M is also homogeneous of degree 1. Thus we conclude that the equation (7+ JP + VF) dx = xdy = 0 Mea is indeed homogeneous. We now show that every homogencous equation can be reduced to a separable ‘equation by proving the following theorem. ‘THEOREM 2.3 y M(x, y) de + NG ¥) dy = 0 224 is a homogeneous equation, then the change of variables y = wx transforms (2.24) into a separable equation in the variables v and x, Proof. Since M(x, y) dx + N(x, ») dy = Ois homogeneous, it may be written in the ‘on Let y= ox Then ay att ee and (2.24 becomes ae oat =o fo ~ of0)] dx + x dv = 0. “This equation is separable. Separating the variables we obtain de (225) OED.22, SEPARABLE EQUATIONS AND EQUATIONS REDUCIBLE TO THIS FORM 45; Thus to solve a homogeneous differential equation of the form (2.24), we let p = vx ‘and transform the homogeneous equation into a separable equation of the form (2.25). From this we have de fax : feta JE where cis an arbitrary constant, Letting F(v) denote Sim ‘and returning to the original dependent variable y, the solution takes the form #(2) +9 11c > Example 2.12 Solve the equation (t= 3p) de + Day ‘We have already observed that this equation is homogeneous, Writing it in the form #242 ay and letting y = vx, we obtain or, finally, “This guavon is separable. Separating the variable, we obtain Dodo de Pl x Iotegratng, we fot Info? — 1] = In |x| + Intel, snd hence lo? = = text,[RST ORDUR EQUATIONS FOR WHICH FRACT SOLUTIONS ARE OBTANARLE where cis an arbitrary constant, The reader should observe that no solutions were lost in the separation process. Now, replacing v by y/x we obtain the solutions in the fom be- |y? — x7 |= Leela? If y > x > 0, then this may be expressed somewhat more simply as potent. D Example 2.13 Solve the initial-value problem 4 JF) de — xay = 0, yy =0, ‘We have seen that the differential equation ishomogencous. As before, we write it in the form Since the initial x value is 1, we consider x > Oand take x = J and obtain ay Bats f-Q) Wellet y = ox and obtain Separating variables, we find wae feet x ‘Using tables, we perform the required integrations to obtain Injo+ Jo FI] = In jx} + Ine}, + JPET22. SEPARABLE EQUATIONS AND EQUATIONS aeDUCILE TO THIS FORM 47 Now replacing v by y/x, we obtain the general solution ofthe differential equation in the form yo be Ea yt (OEP = ox? ‘The intial condition requires that y = 0 when x = 1. This gives c = 1,and hence yt VF from which it follows that yoo? Solve each of the differential equations in Exercises 1-14 1 dye tot + Day bey tet yt Dede $ OF 420 dy WO. 2s? + 1dr +64 + Ids = a 3 4. ose yd + seoxdy = 0. 5. tanOdr-+ 2rd0 =0, 6. (e+ Hoos udu + eMlsinu + Ded 2. 8 9. Gh AY + Ne + WO? + Be Day =O. a Oxy +397) dx - Oxy + x4) dy = 0. 10. 9 du (u? — we) do = 0. (stan2s y)ax—ady = 12, Qs 42m + eA) ds 4 (6? + 2st Dd 13. (249 RE Pde — xy SPE dy 14 (Vet yt Jem ax t (lem y~ Jet dy =0. Solve the initial-value problems in Exercises 15-20, In (yt adet etd =~ (et devas 13) 74 1. Gx + By? + Ade Aye? + 5x4 Ody =O, y(t) =2. i teotyteredtetyen (5)=%48 FIRST-ORDER EQUATIONS FOR WHICH XACT SOLUTIONS ARE OBTAINABLE 18, (a? +3y))dx—2xydy=0, 2m 6 19, (Qx—Sy)dx+f4x—yidy= 0, yl. 20. x? + Sy 4 Sy?) dx — (6x? + 43) dy =0, 21. (a) Show thatthe homogeneous equation y0Q)= 6 (Ax + By) de + (Cx + Dy) dy =0 is exact if and only if B= C. (©). Show that the homogeneous equation (Ax? + Bay + Cy?) dx + (Dx? + Exy + Fy) dy = 0 is exact if and only if B = 2D and E = 2C. 22. Solve each of the following by two methods (se Exercise 21(a) @ &+2)éx+x-ydy =O. 0) Gx—y)ax— (4 Ydy=O. 23. Solve each ofthe following by two methods (see Exercise 21(b)} () (2 + 2y%)de + xy y*)dy =0. (0) Gx? + Dey + 9?) dx + 697 + Dap) dy 24. (a) Provethatif M dx + N dy = Oisa homogencous equation, then the change of variables x = uy transforms this equation intoa separable equation in the variables w and x. (b) Use the resut of (a) to solve the equation of Example 2.12 of the text (6) Use the result of (a) to solve the equation of Example 2.13 of th text 25. Suppose the equation M dy +N dy = 0 is homogeneous. Show that the trans- formation x = r¢0s 0, y =r sin 8 reduces this equation to a separable equation in the variables r and 8, 26. (@) Use the method of Exercise 25 to solve Exercise 8 (©) Use the method of Exercise 25 to solve Exercise 9 27. Suppose the equation Mdx+Ndy=0 a is homogeneous. (@) Show that Equation (A) is invariant under the transformation weak’ 8 where kis a constant. (8) Show that the general solution of Equation (A) can be written inthe form aneo(2). © whore cis an arbitrary constant.f-S ee aee a egeeeee CeCe 23 LINEAR EQUATIONS AND BERNOULII Equations 49 (6) Use the result of) to show that the solution (C)is also iny transformation (B), (@)_ Interpret geometrically the results proved in (a) and (0). jant under the 2.3 LINEAR EQUATIONS AND BERNOULLI EQUATIONS, ‘A. Linear Equations In Chapter 1 we gave the definition ofthe linear ordinary differential equation of order 1m; we now consider the linear ordinary differential equation of the fist order. DEFINITION A first-order ordinary differential equation is linear in the dependent variable y and the independent variable x if iti, oF can be, written in the form 4 2 mioy = 063 29 For example, the equation ¥ . Ett Dyas isa Srstorder liner diferential equation, for it can be written as dy ( Lh eer) dea(is pews ‘which is of the form (2.26) with P(x) = 1 + (1/x) and Ql Let us write Equation (2.26 in the form [Posy ~ Qtx)] dx + dy = 0. e2n Equation (2.27) is of the form M(x, y) dx + N(x, 9) dy = 0, where M(x, 3) = PO)y~ OG) and Mx,» Since M(x, 9) a Equation (2.27) is nor exact unless P(x) = 0, in which case Equation (2.26) degenerates into a simple separable equation. However, Equation (2.27) possesses an integrating factor that depends on x only and may easily be found, Let us proceed to find i. Let us multiply Equation (2.27) by p(x), obtaining Ca PCy — Ho O(8I] dx + nls) dy = 0. 228) aN.) Pts) ang SMG?50 FRST.ORDER EQUATIONS FOR WHICH EXACT SOLUTIONS ARE OBTAINABLE By definition, u(x) is an integrating factor of Equation (228) if and only if Equation (2.28) is exact; that i, if and only if a a rr Cals) Py = nX)QG)) = & Tue) “This condition edues to a MLpP(x) = a Cat). (229) Jn 229), Pisa boown function ofthe independent variable x, bt isan unknown function of x that we are trying to determine. Thus we write (2.29) as the differential vaoaton HP(x) = x in the dependent variable j: and the independent variable x, where P is a known function of x. This differential equation is separable; separating the variables, we have Pis)de. Integrating, we obtain the particular solition nia) ~ freee aa erne 2.30) where itis clear that 4 > 0. Thus the linear equation (2.26) possesses an integrating factor of the form (2.30). Multiplying (2.26) by (2.30) gives inne dd gare my = gee, which is precisely 4 peitondeyy = Phd leroy] = Dee! Jegrating tha we obtain the solution of Equation (2.26) in the form eirenty = f ATMO de + 6 where cis an arbitrary constant Summarizing this discussion, we have the following theorem THEOREM 2.4 The near diferental equation 2 + PCO = Ole (226)23. UNEAR EQUATIONS AND ataNOULL Equations 51 hasan megan factor of the form A one-parameter family of solutions of this equation is yet fanongey acs tha, sarimal farm ee +] Furthermore itcan be shown that this one-parameter foil of solutions ofthe linear equation (2.26 includes all solutions of 2.26) ‘We consider several examples > Example 2.14 2.74). a3) ty)! ‘and hence an integrating factor is on, f Pe | = on fi ons \ae] expQx + In|x)) = exp(2x) exp(in [xl) = x exp(2s)* ng Equation (231) through by this integrating factor, we obtai xem Ds ears nyex 4 ety) = Roeax Integrating, we obtain the solutions xery Ste yah eS where cis an arbitrary constant. The expresions reap # 80 dene52 esr-onvex EQUATIONS FOR WHICH EXACT SOLUTIONS ARE OBTAINABLE > Bxample 2.15 Solve the initial-value problem that consist ofthe differential equation 24% CDR drye x 32) ‘and the initial condition y@)=1. 033) The differential equation (2.32)is not inthe form (2.26) We therefore divide by x? + 1 toobiain ay de x ate a oo Equation (234) is inthe standard form (2.26), where Am integrating factors esol 200d] = exe( f 2) —expance + net = 02+ 1% Multiplying Equation (2.34) through by this integrating factor, we have ct eZ sare s Ny axe? +) or te + yee + Applying the initial condition (2.33), we have S=6 +e. ‘Thus ¢ = 19 and the solution of the initial-value problem under consideration is ox (2 + Dy = +19. > Example 2.16 Consider the differential equation Wax + (xy — Ndy=0. (2.35)123 LINEAR EQUATIONS AND seeNoutet equations 53 Solving for dy/dx, this becomes a iy which is clearly not linear in y. Also, Equation (2.35) is not exact, separable, or ‘homogencous. [t appears fo be of @ type that we have not yet encountered; but let us oak a litle closer. In Section 21, we pointed out that in the differential form ofa first- ‘order differential equation the roles of x and y are interchangeable, in the sense that either variable may be regarded as the dependent variable and the other as the independent variable. Considering differential equation (2.35) with this in mind, let us now regard x as the dependent variable and y as the independent v interpretation, we now write (235) in the derivative form a or a3 t Felee sy 236) ayy ‘Now observe that Equation (2.36) of the form ax Ft Poe = Ov) and 50 is linear in x Thus the theory developed in this section may be applied to Equation (2.36) merely by interchanging the roles played by x and y. Thus an inte- erating factor is os f Py) ]- eo( [2 4) = expla [y?) = 9° Muluipying (2.36 by y* we obtain 24s ys, PB sayteny ton furaes Integrating, we find the solutions in the form vee where cis an arbitrary constant,54 FIRST-ORDER EQUATIONS FOR WHICH EXACT SOLUTIONS ARE OBTAINABLE B. Bernoulli Equations ‘We now consider a rather special type of equation that can be reduced 1 a linear ‘equation by an appropriate transformation. This isthe so-called Bernoulli equation. DEFINITION An equation of the form 4 Get Pony = Oy" an is called a Bernoulli differential equation, We observe that ifn = O0r 1, then the Bernoulli equation (2:37) actually a linear equation ands therefore readily solvable as such. However, inthe general casein which #0 ot t this simple situation does aot hold and we raust proceed in a different ‘manner, We now state and prove Theorem 75, which gives a method of solution in the general case. THEOREM 2.5 Suppose n #0 or 1, Then the transformation v = y'~" reduces the Bernoulli equation 24 Pun our ean 10. linear equation in Peoot. We first maltiply Equation (2.37) by y~*, thereby expressing it in the equivalent form + Poay'-* = O60. 23 it welete = yt°% then and Equation (2.8) transforms into ett + Pe = 00) oF, equivalently, E+ (1 — mPlato = (1 — n QC. # a= Pine , Letting PA) = — 9 Pe)23, LINEAR EQUATIONS AND seeNOU EQUATIONS 55, and 2,08) 1 = mow, this may be written de Fe + Pleo Quen ‘which is linear in QED. > Example 2.17 dy 3 Byars 039) ‘This is a Bernoulli differential equation, where m = 3. We first multiply the equation through by y~?, thereby expressing it in the equivalent form 3B yo yieytex 98 then do/dx = ~2y"(dy/d) and the preceding diferent If we let» = y' equation transforms into the linear equ: (240) ‘We see that an integrating factor for this equation is ‘Multiplying 2.40) by e°*, we find Integrating, we find eMpm de MOx + N+, paxtdtce™, Where cis an arbitrary constant. But56. rst-onnen EQUATIONS FOR WHICH EXACT SOLUTIONS ARE OBTAINABLE Thus we abtain the solutions of (2.39) in the form 1 pret to Note. Consider the equation Ody - dpi * POM) = Obs, ean where f is known function of y, Letting » = f(y), we have dv _dudy _ afty) dy Bade dy ae and Equation (2.41) becomes & E+ Pode = Ot), ‘which is linear in v. We now observe that the Bernoulli differential equation (2.37) is special case of Equation (241), Weiting (2.37) in the form “y 2 + Poay'-*= 060 ‘and then multiplying through by (1 ~ n), we have = my + Roast = O00, = (1 ~ 1) PC) and Q,(2) = (1 ~ n)Q(2). This is ofthe form (241), where "setting v = py! it becomes é Ft Pulao = O40), which is linear in v. For other special cases of (2.41), see Exercise 37. Exercises ‘Solve the given differential equations in Exercises 1-18, ty yg «dy 1 BBL xt Bs rete B+ = oe 2 xt Ba etyat y age ay 3. Pa syearte 4 Badr = be. ax Pee ieaeeectes 5 Ee 6 W + DF + dw =u dy eeteee ee eeee tn neeeaa 23, UNEAR EQUATIONS AND BEENOULU EQUATIONS 57 8 wax sae ny 9. xdy tbe ty Dax=0, 10. yde + (xy? + x—y)dy=0. ar Spt tan @ + 00s 8. 12 cos r+ (rind ~ cost 40 =0 13, (eos? x — y cos x) dx — (1 + sin x) dy 14, (ysin 2x 00s x) de + (1+ sins) dy =0 yyy ay i, Be 6 xBey dented 1. dy Ay 8y Yd = eee IT dy + (4p — 8y"*)x dx = 0. 18. apt ar Solve the initial-value problems in Exercises 19-30. 19. xB apart, yQ) =8. i Haseyest yee 2. eLy-He + MDa t(er+ Ndy=0, Hh Bayt Ndx—GP 4 Ydy= y= 5. a Sermonow'a (§)=t seers af rt Bayete, gle 2. Bare sea, woae fea- FP 9555 sone a Baye sia, woe sean fh OSES oye Be Baya soa, we sin {Oe 235% yo 2x, O0 every solution approaches 0 as x -» co Consider the differential equ dy i Ft Poly=0. (2) Show that if f and g are two solutions of this equation and c, and e, are eee rset, ec Pa is also a solution of this equation. Consider the diferent! equation 2s royna, “ where P is continuous on a real interval I. (8) Show that the function f such that f(x) = 0 forall x€ i solution of this ‘equation. (&) Show that if fis a solution of (A) such that f(x) = Ofor some xg € J, then F(8) = Ofor all xe I (0 Show thatif f and gare two solutions of (A)such that f(s) = g(Xo)for some Xq€ f then f(x) = g(x) for all x € 1 (a) Prove that iff and g are two different solutions of 2 + Poy = 00. “ then f — gis a soltion of the equation 24 Poyy=o. {0} Tausshow thatiff and gare two different solutions of Equation(A)and cis an arbitrary constant, then ats js one-parameter family of solutions of (A. (@) Let f, bea solution of * Fi+ Poly = 0409)123 LINEAR EQUATIONS ANO seRNoULL! EQuarions 59. and f, be a solution of ay Zt Pedy = Ole), where P, Q,, and Q, are all defined on the same real interval J. Prove that Si + fis a solution of 2+ Peoy=ouy+ O00 onl. (0) Use the result of (a) to solve the equation 84 yezsina tssin2e 36, (@) Extend the result of Exercise 35(a) to cover the case ofthe equation 4 a Ft Poay= F oven, where P, Qx(k = 1,2,....0) are all defined on the same real interval I. (6) Use the result obtained in (a) to solve the equation dy Zey 37. Solve each of the following equations of the form (2.41): dy Jt (0) cosy 2 4 siny = ‘ © OFZ + x07 +29 38, The equation 2% ate? + Bey + C0) w is called Riccat's equation. (@)_ Show that if A(x) =0 for all x, then Equation (A) is a linear equation, ‘whereas if C(x) = 0 forall x, then Equation (A) isa Bernoulli equation, (®) Show that iff is any solution of Equation (A), then the transformation 1 yeasty reduces (A) to a linear equation in. In each of Exercises 39-41, use the result of Exercise 38(b) and the given solution to find a one-parameter family of solutions of the given Riceati equation: ” 2 a 39, Bad 2)? +x — yy — xigiven souton f69)= 160 'RST-ORDER EQUATIONS FOR WHICH EXACT SOLUTIONS ARE OBTAINABLE ay 40, _y? + xy + I; piven solution f(x) 41 Bay? + eld + Dy —(Be? + 4x? — 2): given solution f(x) = x Exercises: Miscellaneous Review Solve each ofthe differential equations in Exercises 1-14, Several can be solved by at least wo different methods. 1. 6xtpde~( + aya. 2 Gxly? — a)dy + Oxy? ~y)de #0. 3B (y= Dade tafe + dy = 4, 7 = 2y)dx—xdy=0. 5. Gx-Sy)dx tet dy =O. 6 ey dx + (ey — 2) dy =O. 1 Oxt'y— 2x) dx + ot + dy =0. 8 Ox bay 4 y)de +28? dy = 0. dy _4xty? —302y Sag a iey dy 0 writs a u. 2 a Itsy? & Bye Solve the inital value problems in Exercises 15-24, 1, G+ yTdx=2xydy =O, HD 16 2? H4de + —x Awd =O, y)=0. 17. (ey = 2x)dxteMydy=0, (=. 18 Ge? +20") dx + Qxy4 6y")dy=0, =? B tw Baten @=| 224 src wreceaTinG FACTORS AND Teansronuarions 61 y= 2. nT 92 2 Dyas, where f= { 1, OSx<2, , >, 30) dx, O Example 2.20 (e+ 2y43)de4 Ox tay Nay =0. 256)414 SPECIAL INTEGRATING FACTORS AND TRANSFORMATIONS 67 Here, = 1,6, = 24 =2,D, = 4and y/o =bs/b, “Theorem 27, We therefore let Therefore, this is Case 2 of pax42y, and Equation (2.56) transforms into t+ nderar0(54 =o Tdx + (22 I)de=0, Which is separable, Integrating, we have Jett Replacing z by x + 2y we obtain the solution of Equation (2.56) in the form Txt e+ DP (e+ IW Re XP day + dy? 4 6x = Dy C. Other Special Types and Methods; An Important Reference “Many other special types of first-order equations exist for which corresponding special ‘methods of solution are known, We shall nat go into such highly specialized types in this book. Instead we refer the reader to the book Differentialgleichungen: Losungsmethoden und Losungen, by E. Kamke (Chelsea, New York, 1948). This re ‘markable volume contains discussions of a large number of special types of equations and their solutions. We strongly suggest that the reader consult this book whenever he ‘encounters an unfamiliar type of equation. Of course one may encounter an equation {or which no exact method of solution is known. In such a case one must resort to ‘various methods of approximation. We shall consider some of these general methods in Chapter & Exercises Solve each differential equation in Exercises 1-4 by fist finding an integrating factor 1 Gay + 4y* + 1) dx + (2? + Dey) dy =O, 2% Qe ttany) de +(x — x? tan y)dy = 0. 3. [e+ I+ y)dx + Oxy + Ody 4 Gy + yd + Oy 9d In each of Exercises $ and 6 find an integrating factor of tue form x) and solve 5. (xy? + Oy) dx + (Sxy + Bx) dy =0, 6 (xy? —2y4) dx + (5x29? — Buy) dy =O.HRST-ORDER KQUATIONS FOR WHICH EXACT SOLUTIONS ARE OBTAINABLE Solve each differentia equation in Exercises 7-10 by making suitable transformation, 71. Gx42y+ Ndx4Qxty + dy =0, 8 Gx-y+ Ide (6x -2y- dy =0. 9. 2 de + Ox +y— dy 10, (10x ~4y + 12)dx— (e+ Sy dy =O. Solve the initial-value problems in Exercises 11-14 LL (6x4 4y + dx + Gx + 2y + dy vi) 12 Gro y-Oded(x+y42dy=0, yQ)=-2. 1B Ort 3yt Ddx txt Ot dy 0, y(-2)—2 1 Gxt apt de diet ys Ddy=O, yO)= 4 IS, Prove Theorem 26. 16, Prove Theorem 2.7. 17. Show that if (x,y) and o(, y) are integrating factors of M(x») dx + Nx, 9) dy w such that u(x, y/o(s, 9 is not constant, then bs, 9) = cobs») is a solution of Equation (A) for every constant ¢ 18. Show that if the equation Mos, 9) dx + Nx yb dy “ay is homogeneous and M(x, y)x + N(x, y)y #0, then 1/[M(x, yx + N(x yh] is an integrating factor of (A). 19, Show that ifthe equation M(x, »)dx + N(x, y) dy = is both homogeneous and exact and if M(x, y)x + Nix, )y is not a constant, them the solution of this, equation is M(x, y)x + NIX, yy = €, where cis an arbitrary constant. 20. An squation that is of the form ya px> flv “ay where p = dy/dx and / is a given function, is called a Clairaut equation. Given ‘such an equation, proceed as follows: 1. Differentiate (A) with respect to x and simplify to obtain t+ son 20 (8) Observe that (B) is fstorder diferental equation in sand p 2, Assume x + /"(p) # 0, divide through by this factor, and solve the resulting ‘equation to obtain, pee © where cis an arbitrary constant\ | | i 21 124 SPECIAL INTEGRATING FACTORS AND TRANSFORMATIONS 69 3. Eliminate p between (A) and (C) to obtain peer 4 fle o ‘Note that (D) is one-parameter family of solutions of (A) and compare the form of dilferential equation (A) withthe form of the family of solutions (D) 4. Remark Assuming x + /"(p) = Oand then eliminating p between (A) and x-+ #0) = O may lead to an “extra” solution that is not a member of the one parameter family of solutions of the form (D). Such an extra solution is usually called a singular soluion, For a specific example, see Exercise 21. Consider the Clairaut equation eas & yopet py whee po. (a) Find a one-parameter family of solutions of this equation. {b) Proceed asin the Remark of Exercise 0 and find an “extra” solution that is, ‘not a member of the one-parameter family found in part (a). ph the integral curves corresponding to several members of the one- parameter family of part (a); graph the integral curve corresponding to the “extra” solution of part (bj, and describe the geometric relationship between the graphs of the members ofthe one-parameter family and the graph of the “extra” solution. ©== CHAPTER THREE—— 3a Applications of First-Order Equations {In Chapter 1 we pointed out that differential equations originate from the mathemati~ cal formulation of a great variety of problems in science and engineering. In this ‘chapter we consider problems that give rise to some ofthe types of first-order ordinary differential equations studied 18 Chapter 2. Furst, we formulate the problem mathemat- icaily,ithereby obtaining a differential equation. Then we solve the equation and attempt to interpret the solution ia terms of the quantities involved in the original problem. ORTHOGONAL AND OBLIQUE TRAJECTORIES A. Orthogonal Trajectories DEFINITION Let Fone) =0 on be a given one-parameter fanily of curces Inthe xy plane. A curee that intersects the ‘eurves ofthe family (3.1) at right angles is ealled an orthogonal trajectory ofthe given family D xample 34 Consider the family of circles way 2)| | | | | i \ | i 21 ORTHOGONAL AND OsLIQue reawcrowes 71 with center at the origin and radius c. Bach straight line through the origin, yok, 63) isan orthogonal trajectory ofthe family of circle (32). Conversely, each circle of the family @2}isan orthogonal trajectory of the family of straight lines (3.3). The families (G2) and (.3)are orthogonal trajectories ofeach other. In Figure 3.) several members of the family of circle (3.2), drawn solidly, and several members of the family of straight lines (3.3), drawn with dashes, are shown, The problem of finding the orthogonal trajectories ofa given family of curves arises inmany physical situations. For example, ina two-dimensional lect ied the ines of force ux lines) and the equipotental curves are orthogonal trajectories ofeach other. ‘We now proceed to find the orthogonal trajectories of a family of curves Fix, =0. en ‘We obtain the differential equation of the family (3.1) by irs differentiating Equation (3.1) irmplicitly with respect to” and then eliminating the parameter c between the derived equation so obtained and the given equation (3.1) itself. We assume that the resulting diferential equation of the family (3.1) ean be expressed in the form 4 Fl 9h G4) ae Thus the curve € of the given family (31) which passes through the point (x ) has the slope f(xs9) there, Since an orthogonal trajectory of the sven family intrsets cach curve of the family at ight angles, the slope ofthe orthogonal trajectory to Cat bss easel Fey)APPCATIONS OF FRST.ORDER EQUATIONS Thus the differential equation of the family of orthogonal trajectories is aw 4 ci ees 5 a” Few te A one-parameter family Gx, ya=0 Te) of solutions of the differential equation (3.5) represents the family of orthogonal trajectories ofthe original family (3.1) except possibly for certain trajectories that are vertical lines. ‘We summarize procedure as follows: Procedure fo Finding the Orthogonal Trajectories of a Given Family of Curves Step 1. From the equation Fou y=) en of the given family of curves, find the differential equation 4 Zo joy 6a) ofthis family, Suep 2. In the differential equation dy/dx = f(x, y) so found in Step 1, replace f(s») by its negative reciprocal ~1/ fx,» This gives the differential equation aye 2-75 69 of the orthogonal trajectories, Step 3. Obtain a one-parameter family G{x,y0)=0 of y= Foxe) of solutions of the differential equation (3.5), thus obtaining the desired family of ‘orthogonal trajectories (except possibly for certain trajectories that are vertical lites and must be determined separately) Caution. In Step 1, in finding the differential equation (3.4) ofthe given family, be sure to eliminate the parameter c during the process. > beample 3.2 Tn Example 3.1 we stated that the set of orthogonal trajectories ofthe family of circles 6a31 ORTHOGONAL AND OBLIQUE TRAKCIORKS 73 isthe family of straight lines yoke. 63 Let us verify this using the procédure outlined above Step 1. Differentiating the equation Sayiad 2) of the given family, we obtain sifleo From this we obtain the differential equation ay ky of the given family (3.2) (Note thatthe parameter cwas automatically eliminated in this case) 66) Step 2. We replace —x/y by its negative reciprocal y/x inthe differential equation (G.6)t0 obtain the differential equation dy vy zr an (of the orthogonal trajectories. Step 3. We now solve the differential equation (3.7). Separating variables, we have dy _ ae y integrating, we obtain yaks, 63) ‘his is @ one-parameter family of solutions of the differential equation (3.7) and thus represents the family of orthogonal trajectories of the given family of circles (32) (except for the single trajectory that is the vertical line x =O and this may be deter- ‘mined by inspection) > Example 3.3 Find the orthogonal trajectories of the family of parabolas y = ex? Step 1. We first find the differential equation of the given family yee. 63) Differentiating, we obtain 69)74 armucarions 0F siest-0RDER EQUATIONS Figure 3.2 Eliminating the parameter c between Equations (3.8) and (3.9), we obtain the differ. ‘ential equation of the family (2.8) in the form ay ik 6.10) Step 2. We now find the differential equation of the orthogonal trajectories by replacing 2y/x in (3.10) by its negative reciprocal, obtaining, dy x ky Ga ‘Step 3. Wenow solve the differential equation (2.1). Separating variables, we have 2y dy = ~x dx. Integrating, we obtain the one-parameter family of solutions of (3.11) in the form. ewer, where kis an arbitrary constant. Thisis the family of orthogonal trajectories of (3.8); it is clearly a family of ellipses with centers at the origin and major axes along the x axis. ‘Some members of the original family of parabolas and some of the orthogonal {tajectories (the ellipses) are shown in Figure 3.2. B. Oblique Trajectories 7 DEFINITION Let Fix,y.)=0 3.12)I a 3.1. ORTHOGONAL AND OBLIQUE TRAECTORES 75 ‘eaone-parameter family of eurves. A curce that intersects the curves of the family(3.12) ‘ata consrant angle 2 # 99° is called an oblique trajectory of the given family. ‘Suppose the differential equation of a family is “y Besos 61) ‘Then the curve of the family (3.13) through the point (x, y)has slope f(x, y) at (x, y)and hence its tangent ine has angle of inclination tan~*L/(x, 9] there. The tangent line of an obique trajectory that intersects this curve at the angle awl thus have angle of incination tan7'Lflx y+ a the point (x) Hence the slope of this oblique trajectory is given by (uy) + tana 7% yan @ ‘Thus the differential equation of such a family of oblique trajectories i given by ay | flowy) rtana de 1 fix pian ‘Thus to obtain a family of oblique trajectories intersecting a given family of curves at the constant angle a # 90°, we may follow the three steps inthe above procedure (page 72) for finding the orthogonal trajectories, except that we replace Step 2 by the fol owing step: tan{in*C sts. +3} Step 2. In the diferental equation dy/dx = fxs) of the given family, replace FG, 9) by the expression Sou y)+ tan 17 han 44) T= fle jana 14) > Example 3.4 Find a family of oblique trajectories that intersect the family of straight lines y = cx at angle 45" Step 1, From y = cx, we find dy/dx = c, Eliminating ¢, we obtain the differential equation oy gaz cus of the given family of straight lines. Step 2, We replace f(x») = 9) Eauation (15) by Se yittana | yixt+l_xty ‘Fix stan76 [APPUICATIONS OF FIRST-ORDER EQUATIONS (tan a= tan 45° = 1 here), Thus the differential equation of the desired oblique trajectories is dy _xty ae x=) 16 Step 3. We now solve the differential equation (3.16). Observing that it is a homogeneous differential equation, we let y = vx to obtain Integrating we obtain 4 lao? + 1) — arctan » = —In |x} ~In Je} In c2x?(o? + 1) ~ 2 arctan v = 0. Replacing v by y/x, we obtain the family of oblique trajectories in the form 7 In c%(x? + y?) —2 arctan = = 0. Exercises In Exercises 1-9 find the orthogonal trajectories of each given family of curves. In each case sketch several members of the family and several of the orthogonal trajectories on the same set of axes. 2 4 6 8 10. Find the orthogonal trajectories of the family of ellipses having center at the ‘origin, a focus atthe point (c, 0), and semimajor axis of length 2c 11, Find the orthogonal trajectories of the family of circles which are tangent tothe y axis atthe origi, 12, Find the value of K such that the parabolas y = ¢,x? + K ate the orthogonal trajectories of the Family of ellipses x? + 2)? — y = cy32 promws in mecHanics 77 13, Find the value of m such that the curves x" + y"= cy are the orthogonal tuajectories of the family 14. A given family of curves is said to be seiforthogonal iit family of orthogonal trajectories is th same asthe piven family. Show that the family of parabolas yi = 2ex + cf isselForthogonal. 15, Find a family of oblique trajectories that intersect the family of circles x*+y* at angle 4 16, Finda family of oblique trajectories that intersect the family of parabolasy? = ex at angle 60° 17, Find a family of oblique trajectories that intersect the family of eurves x+y ‘ex? atfangleasuch that tan a = 2 3.2 PROBLEMS IN MECHANICS A. Introduction Before we apply our knowledge of differential equations to certain problems in mechanics, let us briefly recall certain principles of that subject. The momentum of a ‘body is defined to be the product mv of its mass m and its velocity v. The velocity v and hence the momentum are vector quantities. We now state the following basic aw of mechanics: Newton's Second Law. The time rate of change of momentum of a body is proportional to the resultant force acting on the body and is in the direction of this resultant force. Tn mathematical langage, this law states that KF, a Slo) = KF, ‘where mis the mass of the body, visits velocity, Fis the resultant force acting upon it, and K is a constant of proportionality. If the mass m is considered constant, this reduces to “ eakE, om Pete ous78 serucanons oF naSToROER CQUATIONS where k = 1/K and a = dvjdtis the acccleration of the body. The form 3.17} sa direct statement of the manner in which Newton's second law is usually expressed in words, the mass being considered constant. However, we shall make use of ‘the equivalent form (3.18). The magnitude of the constant of proportionality kdepends ‘upon the units employed for force, mass, and acceleration. Obviously (he simplest systems of units are those for which k = 1, When such a system is used (3.18) reduces to Fema G19) Itisin this fort that we shall use Newton's second law, Observe that Equation (3.19)is 4 Yector equation. Several systems of unitsfor which k= 1 are inuse In this text we shall use only three the British gravitational system (British), the centimeter-gram-second system (eg), and the meter-kilogram-second system mks) We summarize the various units of these three systems in Table 3. Recall that the force of gravitational attraction that the earth exerts on a body is called the weight ofthe body. The weight, being a force, is expressed inforce units. Thus the British system the weight is measured in pounds: in the ces system, in dynes; and the mks system, in newtons. Let us now apply Newtons’s second Taw to a freely falling body (a body falling twward the earth the absence of air resistance), Let the mass ofthe body be m an let w denote its weight. The only force acting on the body i its weight and so this is the resultant force. ‘The acceleration is that due to gravity, denoted by g, which is approximately 32 t/sec? in the British system, 980 em/sec* in the cas system, and 98 m/sec? in the mks system (for points near the earth's surface). Newton's second law F = ma thus reduces to w = mg. Thus ae (2) 2 iF (3.20) 4 relation that we shall requenty employ. Letus now considera body Bin rectilinear mation, that iin motion along straight line On Ewe choose afte reference point as origin O, a fied direction as pos tiveanda unto distance. Then the coordinate xf the postion of Brom the origin O tcllsus the citance or displacement of 8. (Ste Fighre 3.3) The instantaneous velocity of Bis the time rate of change of » ds a TABLE 3.1 Brivish System —__egs System mks System force pound ‘dye newion mass slug gram Kilogram distance foot centimeter meter second second second Rsec? ‘emjsec? m/sec?32 prostims in mechanics 79 Figure 3.3 and the instantaneous acceleration of Bis the time rate of change ofc: @x aa ‘Note that x,t, and a are vector quantities. All forces, displacements, velocities, and ‘accelerations in the positive direction on L ate positive quantities; while those in the negative direction are negative quantities. I we now apply Newtnn's second law F = ma to the motion of B along L, noting, that de do ds. ae aaa ‘we may express the aw in any of the following three forms: do mGnk 620 os ni 0.23) a mf = F, 823) where F isthe resultant force acting an the body. The form to use depends upon the way in which F is expressed. For example, i F isa function of time ¢ only and we desire to ‘obtain the velocity vasa function of 1, we would use (3.21); whereasif Fisexpressed asa fonetion of the displacement x and we wish to find » as a function of x, we would ‘employ (3.23). 1, Falling Body Problems We shall now consider someexamples of a body falling through air toward the earth. In such s ciccumstance the body encounters air resistance as it falls. The amount of air ‘resistance depends upon the velocity of the body, but no general law exactly expressing this dependence is known. In some instances the law R = ko appears to be quite satisfactory, while in others R = ko* appears to be more exact. In any case, the con- * stant of proportionality & in turn depends on several circumstances, In the ex- amples that follow we shall assume certain reasonable resistance laws in each case. ‘Thus we shall actually be dealing with idealized problems in which the true resis- tance law is approximated and in which certain comparatively negligible factors are distegarded.BO srrucarions oF rmst-oRDeR EQUATIONS > Example 35 A body weighing 8 Ib falls from rest toward the earth from a great height. Asit fall, air resistance act upon it, and we shall assume that this resistence (in pounds) is ‘numerically equal to 2c, where vis the velocity (in feet per second), Find the velocity and distance fallen at time 1 seconds, Formulation. We choose the positive x axis vertically downward along the path of the body Band the origin atthe point from which the body fell. The forees acting on the body are: 1. Fits weight, 8 1b, which acts downward and hence is positive. 2. Fy the air resistance, numerically equal to 2c, which acts upward and hence is the negative quantity ~ 20 ‘Soe Figure 34, where these forces are indicated, Newton's second law, F = ma, becomes 3.24 Since the body was initially at rest, we have the initial condition (0) = 0. 025) a | Ren a A 7 anh Figure 3.4\ } | a2 pronews in mecnawes 81 Solution. Equation (3.24 is separable. Separating variables, we have do ort Integrating we find —}1n|8 20] = 4+ e6, which reduces to 8-2 ce Applying the condition (3.25) we find c, = 8. Thus the velocity at time ¢is given by =a, 629 ‘Now to determine the distance fallen at time , we write (3.26) in the form ‘and note that x(0) = 0. Integrating the above equation, we obtain (e+ be") +e, Since x = O when « = 0, we find cy = ~} and henoe the distance fallen is given by atthe ®— 0). G27 Interpretation of Results. Equation (3:26) shows us that as t+ co, the velocity o approaches the limiting velocity 4ft/sec. We also observe that this limiting velocity is approximately attained ina very short time. Equation (3:27) states that as ¢~ 0, x also = 20, Does this imply that the body will plow through the earth and continue forever? ‘Of course not; for when the body reaches the earth’s surface its motion will certainly cease. How then do we reconcile this obvious end to the motion with the statement of Equation (327) Itis simple; when the body reaches the earths surface, the differential equation (324) and hence Equation (3.27) no longer apply? Example 3.6 A skydiver equipped with parachute and other essential equipment falls from rest toward the earth. The total weight of the man plus the equipment is 160 Ib. Before the parachute opens, the air resistance (in pounds) is numerically equal to», where vis the velocity (in feet per second). The parachute opens $ sec after the fall begins; after it ‘opens, the air resistance (in pounds) is numerically equal to fv’, where vis the velocity {in feet por second). Find the velocity of the skydiver (A) before the parachute opens, and (B) after the parachute opens. Formulation. We again choose the positive x axis vertically downward with the ‘origin atthe point where the fall began. The statement of the problem suggests that We break it into two parts: (A) before the parachute opens; (B) after it opens. We first consider problem (A). Before the parachute opens, the frees acting upon the skydiver are: 1, F;, the weight, 160 Ib, which acts downward and hence is positive.82 arrucanions oF nest-oxpen FQUATIONS 2. Fa the air resistance, numerically equal tof which acs upward and hence i the nogative quantity —d. We use Newton's second law F v= ma, where F = F, + Fy let m= w/a, and take 2. We obtain Since the skydiver was initially at rest, = 0 when ¢ = 0, Thus, problem (A), concerned ‘with the time before the parachute opens, is formulated as follows: 028) G29) ‘We now turn to the formulation of problem (B). Reasoning as before, we see that after the parachute opens, the forces acting upon the skydiver are: 1. F, = 160, exactly as before. 2. Fy = —de? (instead of ~40), ‘Thus, proceeding as above, we obtain the differential equation a Syn WO ie% Since the parachute opens 5 sec after the fll begins, we have v = v, when t = 5, where +, is the velocity attained when the parachute opened. Thus, problem (B), concerned with the time afer the parachute opens, is formulated as follows: ao 5G = 100-40, 630) (5) = 0 @3n Solution. We shall frst consider problem (A). We find a one-parameter family of solution of do 55 = 160 ~ fo. 628) Separating variables, we obtain Integration yields Info ~ 320) = —Het +o, which readily simplifies to the form 320 + ce"?——_—— 32 proniens In mecnanics. 83 Applying the initial condition (3.28) that » = Oat ¢ = 0, we find that c = ~320, Hence the solution to problem (A) is v= 32001 — 2°), G32) which is valid for 0 < ¢ < 5. In particular, where ¢ 2) = 126, 833) ‘which is the velocity when the parachute opens. ‘Now let us consider problem (B) We ist find a one-parameter family of solutions of the differential equation dy 522 160 — ge 630) ‘Simplifying and separating variables, we obtain de ae Integration yields 1 e=16 Borie ate neato ay nets ee This readily simplifies tothe form on 16 ee om) and solving this for # we obtain aa 35) — ee Applying the initial condition (3.31) that v = v, at ¢ = 5, where n is given by (3.33) and is approximately 126, to (3.34), we obtain He Substituting this into (3.35) we obtain w= lier 0) 639 Interpretation of Results. Let us fist consider the solution of problem (A), given bby Equation (3.32), According to ths, as ¢ +00, v approaches the limiting velocity 320 ffs, Thus if the parachute never opened, the velocity would have been approx imately 320 see atthe time when the unfortunate skydiver would have stuck the ‘earth! But, according othe statement of the problem, the parachute does open 5 sec which is void for 2 5.84 APPLICATIONS OF FASEORDER EQUATIONS after the fall begins (we tacitly and thoughtfully assume $< TT, where TT is the time when the earth is reached!). Then, referring to the solution of problem (B), Equation. (6.36), weseethat ast + 20, v approaches the limiting velocity 16 ft/sec. Thus, assuming, that the parachute opens at a considerable distance above the earth, the velocity is approximately 16 ft/sec when the earth is fnally reached, We thus obtain the-well- ‘known fact that the velocity of impact with the open parachute is @ small raction of the impact velocity that would have occured if the parachute had aot opened. The calculations in this problem arc somewhat complicated, buc the moral is clear: Make certain thatthe parachute opens! . Frictional Forces Ha body moves on a rough surface, it will encounter not only air resistance but also another resistance force due to the roughness of the surface. This additional force is called friccion. Its shown in physics thatthe friction is given by pN, where 1, risa constant of proportionality called the coeficient of friction, which depends ‘pon the roughness of the given surface; and 2. Nisthe normal that is, perpendicular force which the surface exerts on the body. We now apply Newton's second law to a problem in which friction is involved, > eample 37 ‘An object weighing 48 [bis released from rest at the top of a plane metal slide that is inclined 30° to the horizontal. Air resistance (in pounds) is numerically equal to one half the velocity (in feet per second), and the coefficient of friction is one-quarter. |A. What isthe velocity of the object 2 sec after itis released? BI the side i 24 ft long, what is the velocity when the object reaches the bottom? Formulation. The line of motion is along the slide. We choose the origin atthe top and the positive xdirection down theslide. If we temporarily neglect the rition and air resistance, the forces acting upon the object A ate: 1. Ts weight 481, which ats vertically downward: and 2. The normal force, N, exerted by the slide which gets in an upward direction perpendicular tothe slide, See Figure 35} ‘The components of the weight parallel and perpendicular tothe slide have magnitude 48 sin 30° = 24 and 48 c08 3 wi respectively, The components perpendicular fo the slide are in equilibrium and hence the normal force N has magnitude 24/3.a2 mosis i mecranes 85 o Figure 3.5 Now, taking into consideration the friction and air resistance, we see that the forces acting on the abject as it moves along the slide are the following: 1. Fythe component on the weight parallel othe plane, having numerical valu 24. Since this fore act in the positive (downward) direction along the slide, we have A 2. Fthefrctiona fore, having numerical value y= 4(25,/3) negative (upward) direction along the side, we have B= 65. 3. Fy the ar resistance, having numerical value $e: Since» > Oand this also aes in the negative direction, we have 2. 12 thisactsin the We apply Newton's second law P= ma, Here P= +A, + y= 4-6/3 Je and m = w/g = $$ = 4. Thus we have the differential equation =~ 6/3— $0. (337) Since the object is released from rest, the intial condition is (0) = 0. 0.38) Solution. Equation (3.37) is separable; separating variables we have a dr ae 27s Integrating and simplifying, we find vn 48— 25 ~ ce “The condition (3.38) gives e, = 48 — 12/3. Thus we obtain (48 — 12 /3\Q1 =e"), 3.39) Question Ais thus answered by letting 1 = 2in Equation (239), We nd (2) = 48 —12,/3M1 — €°*9) x 1020s)BG armications oF rinst-oRDeR EQUATIONS In order to answer question B, we integrate (3.39) o obtain x48 — 2/3[t 4 30°) + 6, (48 ~12,/3)3). Thus the distance covered at time + is given by = (48 — 12 /FMe + 3 — 3) Since the slide is 24 ft long, the object reaches the bottom atthe time T determined from, the transcendental equation 26 = (48 = 12VINT + 30-7 — 3), which may be writen as Since x(0) = 0,¢5 ate The value of T that satisfies this equation is approximately 2.6. Thus from Equation {.39) the velocity of the object when it reaches the bottom is given approximately by (48 ~ 12/31 — e°) = 123 (sea), Exercises 1. A.stone weighing 4 Ib falls from rest toward the earth from a great height. As it falls itis acted upon by air resistance that is numercally equal to 4v (in pounds}. where vis the velocity (in feet per second), (a) Find the velocity and distance fallen at time ¢ see. (b) Find the velocity and distance fallen at the end of 5 sec. 2 Atal wishing bisthrown vera dowaward toward hearth om aiht Sino wih anata scocly ofc Ass sated pom 9 resistance that is numerically equal to 3n(in pounds), where-vis the velocity (in feet persed (0) Whats the slot and iste fen atthe nd of oe minute? (8) With what vlc des he al sie he art? 3. A ball weighing } Ib is thrown vertically upward from a point 6 ft above the Site of thar than nal voy of 2 /ne utes ced spon by air resistance that is numerically equal to ayo (in pounds), where » is the ‘clon inet per pond), How hgh wl he al sie | 4. A ship which eigh 32000 tons tart om etude he feof constant | propel hut of 00,0 Ib The cance in pooner ea 0 Behr osm et pe sco (0) Find the vey ofthe ship as foto ofthe time (0) Find theming velo (hati thei of vas = +2) (6) Find bow tong aes the ship to atin a velo of 0% ofthe ing | wei | 5. A body of mass 10 gi dropped fom et toward the earth rom a hight of | Too0 mA al ai ven cs upon and ths esas Reto)32 promens in mecHanics 87 proportional to the velocity «(in meters per second). Suppose the limiting velocity is 245 m/sec. (a) Find the velocity and distance fallen at time r secs. (®) Find the time at which the velocity is one-fith of the limiting velocity. 6, Anobject of mass 100 gis thrown vertically upward from 2 point 60 cm above the ‘earth's surface with an intial velocity of 150.em/sec. It rises briefly and then falls, vertically to the earth, all of which time it is acted oa by air resistance that is ‘numerically equal to 2000 (in dynes), where vis the velocity (in em/sec) (a) Find the velocity 0.1 see after the object is thrown, (0) Find the velocity 0.1 sec after the object stops rising and starts falling. 7. Two people ate riding in a motorboat and the combined weight of individuals, ‘motor, beat, and equipment is 640 Ib. The motor exerts a constant force of 20 1b fon the boat in the direction of motion, while the resistance (in pounds) is ‘numerically equal to one and one-half times the velocity (in feet per second). Ifthe boat started from rest, ind the velocity of the boat after (s) 20 sec, b) { min. I ( \ | seeuenaen ena { (a) Find the velocity of the boat 15 sec after the tow rope was cast off. [ \ } i \ I | 9. A bullet weighing 102 is fired vertically downward from a stationary helicopter witha muzzle velocity of 1200 ft/sec. The ar resistance (in pounds) is numerically equal to 167 x, where vis the velocity (in feet per secon), Find the velocity of the billet asa function of the time, 10. A shell weighing 1 Ib is fred vertically upward from the earth’s surface with a ‘muzzle velocity of 1000 ft/sec. The air resistance (in pounds) is numerically equal to 10° *c%, where vis the velocity (in fet per second). (0) Find the velocity ofthe rising shell asa function of the time, (0) How long will the shell rise? 11, An object weighing 16 Ibis dropped from rest on the surface of a calm lake and thereafter starts to sink. While its weight tends to foree it downward, the buoyancy of the object tends to forceit back upward. If this buoyancy force is one of 6 Ib and the resistance ofthe water (in pounds) is numerically equal to twice the square of the velocity (in fei per second), find the formula forthe velocity of the sinking object asa function of the time. 12. An object weighing 12b is placed beneath the surface of a calm lake. The buoyancy of the object is 30 Ib; because of this the object begins to rise. If the resistance of the water (in pounds) is numerically equal to the square of the velocity in fet per second) and the object surfaces in 5 sec, find the velocity of the object at the instant when it reaches the surface.8B szpucaTions oF nstoRDER tQUATIONS 3. 16. ‘A man is pushing a loaded sled across a level field of ice atthe constant speed of 10 ft/sec. When the man is halfway across the ice field, he stops pushing and lets the loaded sied continue on. The combined weight ofthe sled and its load is 80 Ib; the ar resistance (in pounds) is numerically equal to 4, where vis the velocity of the sled (in fet per second); and the coefficient of friction of the runners on the ice is 004. How far will the sled continue to move after the man stops pushing? A girl ondee sled has just slid down a hill onto a level eld of ice and is starting to slow down, AC the instant when their speed is 5 ft/sec, the gir!’s father runs up and begins to push the sled forward, exerting a constant force of 15 Ibn the direction cof motion. The combined weight of the giel and the sled is 96 Ib, the air resistance (in pounds) is numerically equal to one-half the velocity (in feet pet second), and the coefficient of frietion of the runners on the ice is 0.05. How fast isthe sled ‘moving 10 sec after the father begins pushing? ‘A case of canned milk weighing 24 Ibis released from rest at the top of a plane ‘metal slide which is 30 ft fong and inclined 45" to the horizontal. Air resistance (in pounds) is numerically equal to one-third the velocity (in fet per second) and the Coeficent of friction is 0.4 (a) What is the velocity of the moving case 1 sec after its released? (b) What isthe velocity when the case reaches the bottom of the slide? ‘A boy goes sledding down a long 30° slope. The combined weight of the boy and his sled is 721b and the air resistance (in pounds}is numerically equal to twice their velocity (in fet per second). If they started from rest and their velocity atthe end of 5 sec is 10 fi/sec, what is the coefficient of friction of the sled runners on the ‘An object weighing 32 Ibis released fom rest 50 ft above the surface of a calm lake, Before the object reaches the surface ofthe lake, the air resistance (in pounds) is given by 2r, where v isthe velocity (in feet per second). After the object passes beneath the surface, the water resistance (in pounds) is given by 6v. Further, the object is then buoyed up by a buoyancy force of 8b. Find the velocity of the object 2 sec aftr it passes beneath the surface of the lake, AA rocket of mass m is fred vertically upward from the surface of the earth with initial velocity » = v. The only force on the rocket that we consider is the gravitational attraction of the earth. Then, according to Newton's law of ‘gravitation, the acceleration a of the rocket is given by a = —K/x?, where k > 0 §s a constant of proportionality and x is the distance “upward” from the center of the earth along the line of motion. At time ¢ = 0, x = R (where R isthe radius of the earth), a = —g (where g is the acceleration due to gravity), and Express a= dod asin Equation (323), apply the appropriate initial data, and note that v satisfies the differential equation Solve this diferent equation, apply the appropriate inital condition, and thas express vasa function of x In particular, show that the minimum valve of t for which the racket wil escape from the earth ix /29R. This isthe soled tele33 wre rromens 89 ity of escape; and using R= 4000 miles, g = 32 t/sec?, one finds that this is approximately 25,000 mph (or 7 mi/sec. 19. A body of mass mis in rectilinear motion along a horizontal axis, The resultant force acting on the body is given by —kx, where k>0 is a constant of propoctionaity and xis the distance along the axis rom axed point O. The body has initial velocity 0 = v9 when x = xo. Apply Newton's second law in the form (6.23) and thus write the differential equation of motion in the form ao me = hs Solve the differential equation, apply the initial condition, and thus express the ‘square of the velocity «asa function of the distance x, Recalling that v = dx/dt, ‘show that the relation’ between v and x thus obtained is satisfied for alltime ¢ by where ¢ is a constant. 3.3 RATE PROBLEMS In certain problems the rate at which a quantity changes is a known function of the amount present and/or the time, andit is desired to find the quantity itself. Fx denotes the amount of the quantity present at times, then dx/dt denotes the rate at which the quantity changes and we are at once led to differential equation I this section we consider certain problems ofthis ype ‘A, Rate of Growth and Decay > Example 3.8 ‘The rateat which radioactive nuclei decay is proportional (othe number of such nuclei that are present in a given sample, Half of the original number of radioactive nuclei ‘have undergone disintegration in a period of 1500 years 1. What percentage of the original radidactive nuclei will remain after 4500 years? 2. Imhow many years will only one-tenth of the original number remain? Mathematical Formulation. Let x be the amount of radioactive nuclei present “after t years. Then ddr represents the rate at which the nuclei decay. Since the nuclei decay at a rate proportional fo the amount present, we have Gnks G40) ‘where K isa constant of proportionality. The amount xis clearly positive; further, since xisdecreasing, dx/dt < 0. Thus, from Equation (3.40), we must have K <0. Inorderto90 arpucaTions oF riesroanes EQuarions ‘emphasize that x is decreasing, we prefer to replace K by a positive constant preceded —K > Oand write the differential equation (3.40) in ax a Gan Letting xo denote the amount initially present, we also have the initial condition (0) = Xo, 647) We know that we shall need such a condition in order to determine the arbitrary constant that wll apgear in a one-parameter family of solutions of the differential equation (341), However, we shall apparently need something else for Equation (3.41) contains an unknown constant of proportionality k. This “something else" appears in the statement of the problem, for we are told that half of the original number disintegrate in 1500 years. Thus half aso remain at that time, and this at once gives the condition : (1500) = 4x9 643) Solution, The differential equation (3.41) is clearly separable; separating variables, integrating, and simplifying, we have at once Applying the initial condition (3.42), x = xy when t= O, we find that c= x, and hence ‘we obtain 44) We have not yet determined k. Thus we now apply condition (3.43), x = 4xp when = 1500, to Equation (3.44), We find dx = xe, or (erty wt, or finally ete qyinsee, (45) From this equation we could determine k explicily and substitute the result into Equation (344). However, we see from Equation (3.4) that we actually do not need k itself but rather only e~*, which we have just obtained in Equation (345). Thus we substitute e~# from (3.45) into (344) to obtain x= xofe“*Y = xoL(d)8]' xo xo(AV5°0, 46) Equation (3.46) gives the number x of radioactive nuclei that are present at time t ‘Question Tasks us what percentage of the original number will remain after 4500 years. ‘We thus let: = 4500 in Equation (3.46) and find old? = #Xo.aa rare mrontens 91 “Thus, one-eighth or 125% ofthe orginal number remain after $500 years. Question 2 asks us when only one-tenth wll remain. Thus we let x = 4px in Equation 3.46) and solve fort We have wayne Using logarithms, we then obtain In) = Igy” = 5 tn). From this it follows at once that at _ ini 1500 ~ in 1500 in 10 150010 10 3585 years B, Population Growth ‘We next consider the growth of a population (for example, human, an animal species, ‘ora bacteria colony) as a function of time. Note that a population actually increases discontinuously by whole number amounts. However, ifthe population is very large, such individual increases init are essentially negligible compared to the entire pop” ulation itself In other words, the population increase is approximately continuous. We shall therefore assume that this increase is indeed continuous and in fact that the population isa continuous and differentiable fucntion of time. Given a population, we let x be the number of individuals in it at time ¢.1f we assume thatthe rate of change of the population is proportional to the number of individuals init at any time, we are led to the diferential equation ds a where kis @ constant of proportionality. The population xis positive and is increasing and hence dxjdx > 0. Therefore, froi(3:47), we must have k > 0. Now suppose that at time t the population is x9. Then, in addition to the differential equation (3.47), we hhave the initial condition % ean Alto) = Xo. 3.48) ‘The differential equation (3.47) is separable. Separating variables, integrating, and simplifying, we obtain xa cet Applying the intial condition (3.48), x = xpat = to, to this, wehave x9 = ce, From this we al once find e = xye"*® and hence obtain the unique solution x= xqett (3.49) of the differentia equation (3.47), which satisfies the initia condition (3.48)92 arpucations oF ninst-onDER EQUATIONS From (3.49) wesee tha a population governed bythe dilferential equation (347) with k > Oandinitial condition (348) is one that increases exponentially with ime. This aw ‘of population growth is called the Malthusian law, We should now inquire whether or not there are cases in which such a model for population growth is indeed realistic. In ‘answer to this, tan be shown tha this ode), with suitable value of kis remarkably accurate in the case of the human population of the earth during the last several decades (see Problem 8(b)}. It is aso known to be outstandingly accurate for certain mammalian species, with suitable k, under certain realizable conditions and for certain time periods. On the other hand, turning back to the case of the human population of the earth, i can be shown thatthe Malohusian law turns out to be quite unteasonable when applied tothe distant future (se Problem 8()).Itisalso completely unrealistic for other populations (or example, bacteria colonies) when applied over suicientl long periods of time. The reason for ths is not hard to see. For, according to (3.49), a Population modeled by this law always increases and indeed does so at an ever increasing rate; whereas observation shows that a given population simply does not row indefinitely. Population growth s represented more realistically in many cases by assuming that the number of individuals x in the population at time tis described by a diferential ‘equation ofthe form ax : Fm bax, 50) where k > O and 4 > 0 are constants. The additional term ~ x? isthe result of some ‘cause that tends to imit the ultimate growth of the population. For example, such a ‘cause could be insuficient living space or food supply, when the population becomes suificientl large. Concerning the choice of — ix? forthe term representing the effect of the cause, one can argue as follows: Assuming thecause affects theentire population of -xmembers then the effect on any one individuals proportional x. Thus the effect on all x individuals in the population would be proportional to x x = 2°. ‘We thus assume that a population is described by a differential equation of the form {€.50), with constants k > 0 and > 0, and an initial condition of the form (348) In ‘most such cases it turns out thatthe constant 2is very small compared to the constant Thus for sufficiently small x the term kx predominates, and so the population grows very rapidly fora time. However, when x becomes sufficientlylarge, the term —Ax*is of ‘comparatively greater influence, nd the result ofthis is a decrease inthe rapid growth rate. We note thatthe differential equation (3.50) is both a separable equation and a Bernoulli equation. The law of population growth so described is called the logistic law of growth, We now considera specific example of this type of growth. > Example3.9 ‘The population x of a certain.city satisfies the logistic law ee a7 300" ~ Gor as ‘where time ¢is measured in years. Given that the population of this city is 100,000 in 1980, determine the population as a function of time fort > 1980. In particular, answer33 mate rtooums 93. the following questions: (a) What will be the population in 20007 (&) In what year does the 1980 population double? (©) Assuming the differential equation (3.51) applies forall > 1980, how large will the population ultimately be? Solution. We must solve the separable differential equation (3.51) subject to the ini solution ° - x(1980) = 100,000, 3.52) ‘Separating variables in (3.51), we obtain as oreo and hence. dx Worst GH Using partial fractions, this becomes 1 oy" § wales) Integrating, assuming 0 < x < 10°, we obtain: Loon x ~ Ia = (0) I} w+ and hence of apag)eoiete. Thus we find T= 9% ~ Solving this for x, we finally obtain cet00 =a 0s) Now applying the initial condition (3.52) to this, we have 00 = eager i coy? aos | FFT TOT HPF Substituting this value for ¢ back into (3.53) and simplifying, we obtain the solution in94 aerucanions oF restonDes EQUATIONS the form G4) This gies he poptation x as function of ime fort > 1980. ‘We now consider the questions (a), (b), and (c) of the problem, Question (a} asks for the population in he year 2010 Ths we let 2000 in (54 aad obtain it 5085 Question (b) asks for the year in which the population doubles. Thus we let 2 = 200,100 = 2(10)? in G.54) and solve for ¢. We have 200) = yeaa from whic ‘and hence 1 2061 Question 6) asks how large the population will ultimately be, assuming the iferenial ‘equation (3.51) applies forall > 1980, To answer this, weevaluate im x ast» 20 using the solution (3.54 of (3.51). We find 0s Him x = fim atresia = (10% = 1,000,000. C. Mixture Problems We now consider rate problems involving mixtures. A substance $i allowed to flow into acertain mixture ina container ata certain rate, and the mixture is kept uniform by stirring. Further, in one such situation, this uniform mixture simultaneously flows out of the container at another (generally different) rate; in another situation this may not be the case. In either case we seek to determine the quantity of the substance S present in the mixture at time t Letting x denote the amount of $ present at time 1, the derivative dx/dt denotes the rate of change of x with respect 10. IF IN denotes the rate at which Senters the mixture and OUT the rate at which it leaves, we have at once the basic equation ae a from which to determine the amount x of $ at time «, We now consider examples. IN- OUT. 855) D> Example 3.10 ‘A tank initially contains $0 gal of pure water. Starting at time t = 0 a brine containing 2b dissolved salt per gallon flows into the tank atthe rate of 3 gal/min, The mixture33 ware raosiems 95. is kept uniform by stirring and the wellstired mixture simultaneously flows out of the tank at the same rate, L. How much salt isin the tank at any time ¢ > 0? 2 How much salts present at the end of 25 min? 3. How much salt is present after along time? Mathematical Formulation. Let x denote the amount of sat in the tank at time ‘We apply the basic equation (3.55), dx a IN — OUT. ‘The brine flows in atthe rate of 3 gal/min, and each gallon contains 2 Ib of salt. Thus IN = (2 Ibjeal)3 gal/min) = 6 tb/min Since the rate of outflow equals the rate of inflow, the tank contains $0 gal of the ‘mixturest any time t. This 50 gal contains x Ib of sal t time t,and so the concentration cof salt at time tis yx Ib/gal. Thus, since the mixture lows out atthe rate of3 gal/min, we have ouT= & Ia!) ‘eal/min a 5 b/min ‘Thus the diferential equation for x as a function of ris de ok G-5-% 356) Since initially there was no salt in the tank, we also have the initial condition (0) = 0. 657) Solution. Equation (3.56) is both linear and separable. Separating variables, we have dx 3 Wo et Integrating and simplifying, we obtain = 100 + e079, Applying the condition (357), x = Oat « = 0, we find that c= ~ 100. Thus we have x = 100(1 — e°99) 638) ‘This is the answer to question 1. As for question 2, at the end of 25 min, ¢ = 25, and Equation (3.58) gives (25) = 100(1 = e° 4) = 760). ‘Question 3 essentially asks us how much salts present as ¢-+ co. To answer this we let 11+ op in Equation (3.58) and observe that x + 100.96 ArmucaTions OF FinsT-onDeR EQUATIONS, > Example 3.11 ‘A large tank initially contains $0 gal of brine in which there is dissolved 10 1b of salt. Brine containing 2 Ib of dissolved salt per gallon flows into the tank at the rate of 5 gal/min. The mixture is kept uniform by stirring, and the stirred mixture simulta neously lows out atthe slower rate of 3 gal/min. How much salt isin the tank at any time 1 > 0? ‘Mathematical Formulation. Let x =the amount of salt t times, Again we shall use Equation (3.55): & = IN - OUT. Proceeding as in Example 3.10, / IN = (2 Ib/gal(5 gal/min) = 10 {b/min; Aso, once again OUT =(C Ib/gal)(3 gal/min), ‘where C Ib/gal denotes the concentration. But here since the rate of outflow is different from that of inflow, the concentration is not quite so simple. At time ¢ = 0, the tank contains 50 gal of brine. Since brine flows in at the rate of $ gal/min but lows out atthe slower rate of 3 gal/min, there isa net gain of 2 gal/min of brine in the tank ‘Thus at the end of 1 minutes the amount of brine in the tank is 50 +2 gal Hence the concentration at time ¢ minutes is 50+ 2 ‘b/eal, anise ovrag’’ yin aa sie dient cnatonBsomes #20 gy re) Sie here sini 101 of an he nk, we oe ini eondon 0-10 ow Solution. The fferential equation (3.59) is not separable but it is linear. Putting it in standard form,a3 sare reonens 97 we find the integrating factor 3 2 ool feta) + 50)", Multiplying through by this, we have Qt sop $32 + SOx = LOgee + 509 4 t+ 50)%%x] = 10 + 50)" Thus (21 + 50% = 200¢ + 50)°7 +6 or xe A+) 4 Soe Applying condition (3.60), x = 10at t = 0, we find Wo 1004 + ope or 2,00 /3. ‘Thus the amount of salt at any time ¢ > Ois given by e 22,5004/2 x 44 100 — Exercises 1. Assume that the cate at which radioactive nuclei decay is proportional to the ‘numberof such nuclei that are present ina given sample. Ina certain sample 10% of the original number of radioactive nuclei have undergone disintegration in 8 period of 100 years. (a) What percentage of the original radioactive nuclei will remain after 1900 years? (b)_ In how many years will only one-fourth of the original number remain? 2. Acertain chemical is converted into another chemical by chemical reaction. The rate at which the first chemical is converted is proportional to the amount of this ‘chemical present at any instant. Ten percent of the original amount of the first ‘chemical has been converted in 5 min, {@) What percent of the first chemical will have been converted in 20 min? (©) Inhow many minutes will 607% of the frst chemical have been convested?9B armuications oF finsT-oRDER EQUATIONS 3 ‘A chemical reaction converts a certain chemical into another chemical, and the rate at which the first chemical is converted is proportional to the amount of this ‘chemical present at any time. At the end of one kour, 50 gm of the first chemical remain; while atthe end of three hours, only 25 gm remain. @ O) © ‘How many grams of the first chemical were present initially? How many grams of the first chemical will remain atthe end of five hours? In how many hours will only 2 gm ofthe frst chemical remain? ‘A chemical reaction converts & certain chemical into another chemical, and the rate at which the first chemical is converted is proportional to the amount of this chemical present at any time. At the end of one hour, two-thirds kg of the frst chemical remains, while at the end of four hours, only one-third kg remains. @ @) ‘What fraction of the first chemical remains at the end of seven hours? ‘When will only one-tenth of the first chemical remain? Assume that the population of a certain city increases at arate proportional 10 the ‘number of inhabitants at any time. I the population doubles in 40 years, in how ‘many years will it triple? ‘The popula of the city of Bingville increases at a rate proportional to the ‘number of is inhabitants present at any time. If the population of Bingvlle was 30000 in 1970 and 35000 in 1980, what will be the population of Bingvile in 19907 In a certain bacteria culture the rate of increase in the number of bacteri proportional to the number present «® 0 If the number triples in 5 hr, how many will be present in 10 hy? ‘When will the number present be 10 times the number initially present? ‘Assume that the rate of change of the human population of the earth is pro- portional to the number of people on earth at any time, and suppose that this population s increasing atthe rate of 2% per year. The 1979 World Almanac gives the 1978 world population estimate as 4,219 million; assume this figure isin fact correct, @ © @ @ © Using this data, express the human population of the earth asa function of time. ‘According tothe formula of part (a), what was the population of the earth in 19507 The 1979 World Almanae gives the 1950 world population estimate as, 2,510 million. Assuming this estimate is very nearly correct, comment on the ‘accuracy of the formula of part (a) in checking such past populations. According othe formula of part (a, what willbe the population of the earth jn 2000? Does this seem reasonable? ‘According to the formula of par (a), what was the population ofthe earth in 19002 The 1979 World Almanac gives the 1900 world population estimate as, 1,600 million. Ascuming this estimate is very nearly correct, comment on the accuracy of the formula of part (a) n checking such past populations. According tothe formula of part (a), what will be the population of the earth in 2100? Does this seem reasonable?10. 2 14 aa ertmonns 99 ‘The human population of a certain island satisfies the logistic law (3.50) with k= 003, A= 3(10)-*, and time t measured in years. (2) If the population in 1980 is 200,000, find a formula for the population in future years {b) According to the formula of part a), what will be the population in the year 20007 (6) What isthe limiting value of the population as t+ co? ‘This isa genetal problem about the logistic law of growth. A population satisfies the logistic law (3.50) and has xo members at time fo. (a) Solve the differentia equation (3.90) and thus express the population x asa function of. (b) Show that as -+ co, the population x approaches the limiting value kj. (6) Show that dx/dt is increasing if x < k/22 and decreasing ifx > k/2A. (@) Graph x asa function of t for > ty {e) Interpret the results of parts (b), (c), and (d). ‘The human population of a certain small island would satisly the logistic law (3.50), with & = fg, 2 = (10)"%, and ¢ measured in years, provided the annual emigration from the island is neglected. However, the fact is that every year 100, people become disenchanted with island life and move from the island to the mainland, Modify the logistic differential equation (350) with the given kand 2s0 4 fo include the stated annual emigration. Assuming that the population in 1980, is 20,000, solve the resulting initial-value problem and thus find the pop- lation of the istand as a function of time. Under natural circumstances the population of mice on a certain island would increase at a rate proportional to the number of mice present at any time, provided the island had no cats. There were no cats on the island from the beginning of. 1970 to the beginning of 1980, and during this time the mouse population doubled, reaching an all-time high of 100,000 at the beginning of 1980. At this time the people of the island, alarmed by the increasing number of mice, imported a number of cats to kill the mice. Ifthe indicated natural cate of increase of mice was thereafter offset by the work of the cats, who killed 1000 mice a month, how many mice remained at the beginning of 1981? ‘An amount of invested money ssid to draw interest compounded continuously it the amount of money increases ata rate proportional to the amount present Suppose $1000 is invested and draws interest compounded continuously, where the annual interest rates 6 (@) How much money’ will be present 10 years afer the orginal amount was invested? (©) How long will it take the original amount of money to double? ‘Suppose a certain amount of money is invested and draws interest compounded continuously. (a) Ifthe original amount doubles in (wo years, then what isthe annual interest rate?100 APPLICATIONS OF FIRST-ORDER EQUATIONS 1s. Yn. 18 (6) Ifthe original amount increases 50% in six months, then how long willit take the original amount to double? ‘A tank intially contains 100 gal of brine in which there is dissolved 20 ib of salt. Starting at time ¢ = 0, brine containing 3 Ib of dissolved salt per gallon flows into the tank at the tate of 4 gal/min. The mixture is kept uniform by stirring and the ‘well-stired mixture simultaneously flows out of the tank at the satne rate, (a) How much satis in the tank at the end of 10 min? (6) When is there 160 tb of salt in the tank? AA large tank initially contains 100 gal of brine in which 10 Ib of satis dissolved. Starting at ¢ = 0, pure water flows into the tank at the rate of 5 gal/min, The ‘mixture is kept uniform by stirring and the wellstirred mixture simultaneously flows out at the slower rate of 2 gal/min, (a) How much salt is in the tank at the end of 15min and what is the concentration at that time? (b} IF the capacity of the tank is 250 gal, what isthe concentration at the instant the tank overflows? |A tank initially contains 100 gal of pure water. Starting at r= 0, a brine containing 4 Ib of salt per gallon flows into the tank at the rate of 5 gal/min. The ‘mixture is kept uniform by stirring and the well-stirred mixture flows out at the slower rate of 3 gal/min, (a) How much satis in the tank at the end of 20 min? o Seeartamne eaere: ‘A large tank initially contains 200 gal of brine in which 15 tb of salt is dissolved. Starting att = 0, brine containing 4 Ib of salt per gallon flows into the tank at the rate of 35 gal/min. The mixture is kept uniform by stirring and the well-stirred mixture leaves the tank atthe rate of 4 gal/min (a) How much salt is in thé tank at the end of one hour? (b)_How much salt isin the tank when the tank contains only $0 gal of brine? ‘A 500 liter tank initially contains 300 liters of fuid in which there is dissolved 50 gm of a certain chemical. Fluid containing 30 gm per liter of the dissolved chemical flows into the tank at the rate of fiters/min. The mixture is kept uniform by stirring, and the stirred mixture simultaneously lows out atthe rate of 25 lters/min, How much ofthe chemical isin the tank at the instant it overflows? ‘A 200 liter tank is initially full of uid in which there is dissolved 40 gm of a certain chemical. Fluid containing 50 gm per liter ofthis chemical flows into the tank atthe rate of 5 liters/min, The mixture is kept uniform by stirring, and the stirred mixture simultaneously lows out atthe rat of 7 liters/min, How much of ‘the chemica? isin the tank whem itis only half ull? ‘The air im a room whose volume is 10,000 eu ft tests 0.15% carbon dioxide. Starting at = 0, outside air testing 005%, carbon dioxide is admitted atthe rate cof $000 cu ft/min. (a) What isthe percentage of carbon dioxide inthe ar in the room after 3 in? carbon dioxide? (©) When does the air in the room test 0.1%| | | | 2. 2B, 24, 25 33 mate'rrogiens 101, ‘The air ina room 50 ft by 20 ftby 8 ft tests 0.27 carbon dioxide. Startingat ¢ = 0, ‘outside air testing 0.05% carhon dioxide is admitted to the room. How many ‘cubic feet of this outside air must be admitted per minute in order that the air in the room test 0.1% atthe end of 30 min? Newton's law of cooling states that the rate at which a body cools proportional to the difference between the temperature of the body and that of the medium in which it is situated. A body of temperature 80 °F is placed at time 1=0 in a ‘medium the temperature of which is maintained at 50°F, At the end of S min, the body has cooled to a temperature of 70 °F. (2) What isthe temperature of the body at the end of 10 min? (b) When will the teroperature of the body be 60°F? ‘A body cools from 60 °C to $0°C in 15 min in air whieh is maintained to 30°C. How long will it take this body to cool from 100 °C to 80 °C in air that is main- {tained at 50 °C? Assume Newton's iaw of cooling (Exercise 23). ‘The rate at which a certain substance dissolves in water is proportional at the product of the amount undissolved and the difference e, — ¢,, where ¢, is the concentration in the saturated solution and ¢, isthe concentration in the actual solution. If saturated, $0 gm of water would dissolve 20 gm of the substance. IF 10 gm of the substance is placed in 50 gm of water and half of the substance is then dissolved in 90 min, how much will be dissolved in 3 hr?—— CHAPTER FOUR—— 44 Explicit Methods of Solving Higher-Order Linear Differential Equations ‘The subject of ordinary linear differential equations is one of great theoretical and practical importance. Theoretically, the subject is one of simplicity and elegance. Practically, linear differential equations originate in a variety of applications to science and engineering. Fortunately many of the linear differential equations that thus occur are of a special type, linear with constant coeficients, for which explicit methods of solution are available. The main purpose of this chapter is to study certain of these ‘methods. First, however, we need to consider certain basic theorems that will be used throughout the chapter. These theorems are stated and illustrated in Section 4.1, but proofs are omitted in thisintroductory section. By far the most important cases that of the second-order linear differential equation, and we shall explicitly consider and illustrate this case for each important concept and resnlt presented, In the Final section ‘of the chapter we return to this fundamental theory and present theorems and proofs in ‘this important special case. Proofs in the general case are given ia Chapter 11 BASIC THEORY OF LINEAR DIFFERENTIAL EQUATIONS. A. Definition and Basic Existencé Theorem DEFINITION A linear ordinary differential equation of ‘order n in the dependent variable y and the independent tariable x is an equation that is in; or canbe expressed in, the form ta Z+atdy= Fe, 4D41 BASIC THEORY OF LINEAR DifFERENTIAL EQUATIONS 103, sohere a snot identically zero, We shall assume that ay, ..,d, and F are continuous real functions ona real ntercala Example 4.1 The equation #y Be ‘sa linear ordinary ditferential equation of the second order. Y yee tarde cyee > example 4.2 ‘The equation ey. dy ae aE {is a'finear ordinary differential equation ofthe third order. dy a — sy asin + y ‘We now state the basic existence theorem for initial-value problems associated with and nth-order linear ordinary differential equation: ‘THEOREM 4.1 Hypothesis 1. Consider the nth-order linear differential equation anty 2 4 ay 0) 2 + aay = (4 als) Fy + lS +o Z + aye Fo) CN)104 PXPUCT METHODS OF SOLVING HIGHER ORDER UNEAR DAFERENTIAL EQUATIONS where doy... and F are continuous real functions on areal interval a Bample 43 ‘Consider the initial-value problem Py 5d ys, Grr wZrvyne, y(t) = 2, y@=—s, ‘The coefficients 1, 3x, and x?, as well as the nonhomogeneous term e*, in this second- order diflerential equation are all continuous forall values of x, ~00 < x < ao, The point x, here is the point 1, which certainly belongs to this interval; and the real numbers cq and c, are 2 and —5, respectively. Thus Theorem 4.1 assures us that a solution of the given problem exists is unique, and is defined for all x, —co < x < «@.41 BASIC THEORY OF LineaR DirreneNMiAL equations 105 D> Example'4.4 ‘Consider the initial-value problem ey ey ay 2S5exGr tie Psy 4) = 3, Y=, y@=-F Here we have a third-order problem, The coeflicients 2,x,3x7, and —5, as well as the nonhomogeneous term sin x, are all continuous for all x,—co Example 45 ‘The unique solution f of the third-order homogeneous equation @y pia Oro 28e p otha tyna,106 [EXPLICIT METHODS OF SOLVING HIGHER-OROER LINEAR OIFERENTIL EQUATIONS ‘which is such that JO) = $12) = fiy=0, is the trivial solution f such that (2) = O forall x. 8. The Homogeneous Equation ‘We now consider the fundamental results concerning the homogeneous equation arty we ‘We first state the following basic theorem: ey ay (0) $2 + asx) +4409 + ay =0. (42) THEOREM 4.2 BASIC THEOREM ON LINEAR HOMOGENEOUS: DIFFERENTIAL EQUATIONS Hypothesis. Let fu.fas---ufu be any m solutions of the homogeneous linear differ- ‘ential equation (4.2). Conclusion. Then cy fy +2 fot °*+ CwJae 18 also a solution of (4.2), where easly are m arbitrary constants Theorem 42 states that if m known solutions of (42) are each multiplied by an arbitrary constant and the resulting products are then added together, the resulting sums alo a solution of (42). Wemay pot this theorem ina very simple form by means of the concept of linear combination, which we now introduce. DEFINITION I SisSave-n fu ar@ m given functions, and ¢4,025---46q are m constants, then the x- pression ah teah te + eda is called a linear combination of f,sas--s Sux In terms ofthis concept, Theorem 4.2 may be stated as follows: THEOREM 4.2. (RESTATED) Any linear combination of solutions of the homogeneous linear differential equation (4.2) is also a solution of (4.2). In particular, any linear combination CA teaha tet ondeasic THEORY oF LINEAR onrexenriat equATions 107 of m solutions fiafz.---nfa of the second-order homogeneous linear differential equation 2 4 240) £3 + 04092 + ay00y =0 (4a) is also a solution of (44), > example 46 “The student will readily verily that sn x and cos x are solutions of 4, getyn ‘Theorem 42 states that the linear combination ¢, sin x + ¢; cos xisalso a solution for any constants c, and cs. For example, the particular linear combination Ssinx + 6008 x isa solution, D> sxample 4.7 ‘The student may verify that e, "4, and e* ae solutions of ay ty ay Gir eto. Theorem 42 states tha the linear combination ¢, e* + ¢,e"* + c3e**isalsoasolation for any constants cy, cand ¢. For example, the particular linear combination Det — 30° + Fe is a solution, We now consider what constitutes the so-called general solution of (42). To understand this we first introduce the concepts of linear dependence and linear independence. DEFINITION The m functions ffs fy are called linearly dependent on a x Example 4.8 ‘We observe that x and 2x are linearly dependent on the interval 0 x <1. For there ‘exist constants ¢, and ¢;, not both zero, uch that yx + (28) for all x on the interval 0 Example 4.9 We observe that sin x, 3sin x, and —sin x are linearly dependent on the interval 11s x:52.For there exist constants c,,c5.¢), nt all ro, such that ex sin x + 6,(3sin x) +¢4(~sin) = 0 for all x on the interval ~1 < x <2. For example let ¢, = I,¢a = 1ycy = 4. DEFINITION ‘Then functions sf.» f,are called linearly independent on the interval a < x < bif they are not linearly dependent there, That ithe functions fi fayn-f, ar linearly independent on a's x < bij the relation CAFO) + Oafabe) +--+ eh) = 0 forall x such that ax Example 4.11 We have observed that sin x and cos x ae solutions of ey grtys0 an for all x, ~20 < x < ep. Further, one can show that these two solutions are linearly independent. Now suppose f is any solution of (4.7). Then by Theorem 43 f can be expressed as a certain linear combination ¢, sin x +c, cos x of the two linearly independent solutions sin x and cos x by proper choice ofc, and ¢;. Thatis, there exist ‘wo particular constants ¢, and c, such that Sos) ey sinx + €3c08-x or)110 [EUCTT METHODS OF SOLVING HIGHER-ORDER LINEAR DIFFERENTIAL EQUATIONS forall x, ~c0 Example 4.12 We have observed that sin x and cos x are solutions of - {hy p00 forall x, ~20 < x < co, Firther, one ean show that these two solutions are linearly independent. Thus, they constitute a fundamental set of solutions of the given differential equation, and its general solution may be expressed as the linear combination €; sin x + ¢, 608%, where c, and cz are arbitrary constants. We write this as y = cy sin x + ¢, 605 x > Example 4.13 and e** of ®y fy ye $5-205 Ba ryao say be shown to belinearly independent for allx, — co < x < oo, Thus,e',e"% and constitute fundamental set of the given differential equation, and its general solution may be expressed as the linear combination ‘The solutions ee cet tee tee where ¢,,¢2, and ¢, are arbitrary constants. We write this as pager baet tose®. ‘The next theorem gives a simple criterion for determining whether or not n solutions of (4.2) are linearly independent. We first introduce another concept. DEFINITION Let fy Fanon fy be m real functions each of which has an (nm — {)st derivative on a reat interval a'< x beample 4.14 We apply Theotem 44 to show that the solutions sin x and cos x of dy aetyne441, SIC THEORY OF UNeAR DarERENTIAL EQUAMIONS 113 are linearly independent, We find that sinx cos. Wisin x,c08 x) = = Example 4.15 ‘The solutions e*, e€*, and ¢?* of ay a are linearly independent on every real interval for | jtoay Were eyelet et reac -1 |= 60 let e* ae ron al for all real x Exercises 1, Theorem 4.1 applics to one of the following problems but not to the other. Determine to which ofthe problems the theorem applis and state precisely the conclusion which can be drawn in this case. Explain why the theorem does not ‘apply tothe remaining problem. Py say 7 ‘ @ GrtsZtone O=8 yO=7. #y dy © Sesbrgne, yO=5 v0 2. Answer orally: Whi Tear Bertyno, 0 yo. he solution ofthe following initial-value problem? Why? 3. Prove Theorem 42 for the case m = = 2. That i, prove that if fu(x} and f(%) are two solutions of ay dy g(a) Fe + AULA) FE ala) y =O, then o, fe) + eft) is also a solution of this equation, where ¢, and cy are arbitrary constants114 cxruicrr merno0s oF soLvING HIGHER-ORDER LINEAR DIFFERENTIAL EQUATIONS. 4. Consider the atferential equation Py yay ae fag tO e { (a) Show that each of the functions * and eis a solution of differential equation (A) on the interval a < x fer sea ‘au(x) Ine tcp? — [SUS a +t THF ‘This is the generat solution of Equation (4.14); choosing the particular solution for which e = 1, recalling that do/dx = w, and integrating again, we now obtain ats) | - 109 gy 2000) ce f OF Finally from (411), we obtain oso [3e8h¢<] yoo [hae wn ‘Tiefanetiondedin the ight member of 1) which we stall hnefocthdenotby ‘,is actually a Solution ofthe original second-order equation (4,10). Furthermore, this ‘new solution g and the origina! known solution f are linearly independent, since fo) ata] [foo fede Se) geo “LF Slaw + Feo} = yon onl - [28 ae] 20 ‘Thus integrating, we obtain WUS at 40x) ‘Thus the linear combination afte js the general solution of Equation (4.10). We now summarize this diseussion in the following theorem,118 {EXPLICIT METHODS OF SOLVING HIGHER.ORDEE LINEAR DIFFERENTIAL EQUATIONS THEOREM 4.7 Hypothesis. Let fhe a nomricial solution of the second-order homogeneous Iinear dferential equation 2560) + est + etey <0 (419) Conclusion 1. The transformation y = f(x)o reduces Equation (10) t0 the first= order homogeneous linear differential equation 6300) + Baybnf'0) + (9) /CxNw = 0 (ag) in the dependent variable w, where w = dé Conclusion 2. The particular solution sol - [ae (-[254] TOF of Equation (4.14) gives rise to a 1b, where f ut) aoe (-[a ats} «], Ter The function g defined by g(x) = f{x)lx) is then a solution of the second-order equation 10), Conclusion 3. The original known solution f and the “nen” solution g are linearly independent solutions of 4.10), and hence the general solution of (4.10) may be expressed as the linear combination caf t cag! Letus emphasize the utility of this theorem and atthe same time clearly recognize its limitations, Certainly its utility is by now obvious. [tells us that ifone solution of the. second-order equation (4.10)is known, then we can reduce the order to obtain linearly independent solution and thereby obtain the general solution of (4.10). But the limitations of the theorem are equally obvious. One solution of Equation (4.10) must already be known to usin order to apply the theorem. How does one “already know” a solution? In general one does not, In some cases the form of the equation itself or related physical considerations suggest that there may be a solution ofa certain special form:for example, an exponential solution or alinear solution. However, such cases are not too common and if no solution at all can be so ascertained, then the theorem will not aid us. ‘We now illustrate the method of reduction of order by means of the following ‘example443, asic THEORY OF UNEAR DurrReNTAL quaions 119) D Example 4.16 Given that y = xis a solution of 24 fh a (8+ 0G5 B+ 2y=0, (416) find a linearly independent solution by reducing the order. Solution. First observe that y = x does satisfy Equation (4.16), Then let yam ‘Then @e | do 8 eo oand Poesy 28. ‘Substituting the expressions for y, dy/dx, and dy/dx? into Equation (4.16), we obtain ver o( example 4.17 Observe that y = xs solution of the nonhomogeneous equation ay at and that y = sin xis a solution of the corresponding homogeneous equation @y Seta “Then by Theorem 48 the sum sinx +x is also solution of the given nonhomogeneous equation ‘The student should check that ths is indeed true.41 aAsic THEORY OF LuNeaR ourreneNTiAt equanions 121 ‘Now let us apply Theorem 48 in the case where o is a given solution y, of the ‘nonhomogencous equation (4.1) involving no arbitrary constants, and ws the general solution Vere t Cavs b+ Oe of the corresponding homogencous equation (4.2). Then by this theorem, et Ye {salso.a solution of the nonhomogeneous equation (41), and it is2 solution involving n arbitrary constants ¢,, cy...» 6: Concerning the significance of such a solution, we ‘now state the following result ‘THEOREM 4.9 Hypothesis (i) Let y, be a given solution ofthe nth-order nonhomogeneous linear equation (4.1) ‘nvoloing no arbitrary constants. (2) Let Yer eV beava beet CMe ‘be the general solution of the corresponding homogeneous equation (4.2), Conclusion. Then every solution 6 of the nth-order nonhomogeneous equation (4.1) ‘can be expressed in the form Yet Ye that is, cats + Cavs Hoo + eae te ‘for suitable choice of the n arbitrary constants Cy, C4y---s Cu: ‘This result suggests that we calla solution of Equation (6.1) of the form y. + yp,a general solution of (4.1), in accordance with the following definition: DEFINITION Consider the nth-order (nonhomogeneous) linear differential equation ay 4400) tao +a0oy= Fe) (4) a 4068) and the corresponding homogeneous equation fy eos + aslaly = ants) Fr + 040) +61 E + alady = 0. 42) 1. The general solution of(4.2)s called the complementary nction of Equation (4. We shall denote this by y.122 _exrucrt metH00s OF SOLVING HIGHER ORDER LINEAR DIFFEREXTAL EQUATIONS 2. Any particular solution of 4.1) involving no arbitrary constants i called a particular integral of (4.1) We shall denote his by yp. 3. The solution y, + yy of (9.1), where y, isthe complementary function and y, is a articular integral of (4.1), is called the general solution of (4.1). ‘Thus to find the general solution of (4.1), we need merely find 1. The complementary function, that is, a “general” linear combination of m linearly independent solutions of the corresponding homogeneous equation (42), and 2 A particular integral, that is, any particular solution of (4.1) involving no arbitrary constants, > Example 4.18 ‘Consider the differential equation ay ae ‘The complementary function is the general solution Bayae ‘Thus the general solution of the given equation may be written, y= yet yp se Sin x + ep C08 x +x, In the remaining sections of this chapter we shall proceed to study methods of obtaining the two constituent parts of the general soluti We point out that if the nonhomogeneous member F(x) of the linear differential ‘equation (4.1)is expressed as'a linear combination of two or more functions, then the following theorem may often be used to advantage in finding a particular integra THEOREM 4.10 Hypothesis J, Let f, bea particular integral of é amty dy : A660) 3 + a0) Fay ot IE A ledy = Rls. (IT)44 BASIC THEORY OF UNEAR DiFreRENTIAL EQUATIONS 123, 2, Let fy bea particular integral of 4 ta +aidy= Fis) (618) Conclusion. Then kf, + kf 18 a particular integral of “ a 4 aol8) Gon + 10) Feast + aye SE CY = By Fil) + hail), 419) vohere andy are constants D> Example 4.19 Suppose we seek a particular integral of Pha y odes Seanx (e209) ‘We may then consider the two equations @y artyn* (20 and #y ye Gi tya tan (4.22) We have already noted in Example 4.18 that a particular integral of Equation (4.21) is given by yer, Further, we can verify (by direct substitution) that a particular integral of Equation (4.22) is given by y= [00s x)In |sec x + tan x]. ‘Therefore, applying Theorem 4,10, a particular integral of Equation (4.22) is 3x ~ So0s x)In fsec x + tan x] ‘Thisexamplemakesthe utility of Theorem4.1Oapparent. The particularintegral y of (4.21) can be quickly determined by the method of Section 43 (or by direct inspection’), whereas the particular integral y of (422) must be determined by the method of Section 44, and this requires ‘considerably greater computation, (cos x)In |soo x + tan x]eee eee 124 [FRPUCIT METHODS OF SOLVING HIGHER ORDER LINEAR DIFERENTIAL EQUATIONS Exercises L. Given that y = x sa solution of 28 4D ye PGE eZ ety, find a linearly independent solution by reducing the order. Write the general solution, 2 Given that y = x + Lisa solution of @& 4 et PERM Zt y= find 2 linearly independent solution by reducing the order. Write the general solution, 3. Given that y = xis a solution of find a linearly independent solution by reducing the order. Write the general solution. tne tt epte nl ie ptarenthiteaBeteeiraa find a linearly independent solution by reducing the order. Write the general 5. Gheutn y= nanlanet acs merneeyne tn ney eee vtion by on re Wit eel oa 6 Given that y = xis a solution of “ ee oh (2 x4 0 428-298 4 ext 42x29 00, 2. Prove Theorem 48 forthe case n= 2, Thats, prove that if wi any solation of 2) oo retny ot) $3 + ae) 2 + asin 20 and vis any solution of ate) 23 + 0,60) 2 + an(0)y = FOO, 9 £3 + a6) 2 + exlny = Fos, then u + via also a solution of this latter nonhomogeneous equation,442. ‘TE HOMOGENEOUS UNEAR EQUATION WeTH constANT comncENTs 125 8. Consider the nonhomogencous differential equation #y_ 3a a Bihan ae, (@) Show that e* and e** are linearly independent solutions ofthe corresponding, homogeneous equation (©) What is the complementary function of the given nonhomogeneous equation? (6) Show that 2x? + 6x + 7s particular integral ofthe given equation. (@) Whats the general solution ofthe given equation? 9. Given that a particular integral of yay faa Gi-sZe oer is yas 4 particular integral of ay x 3 Fro sB reyes in ytd, and a particular integral of ‘use Theorem 4.10 to find a particular integral of oe ee : Get get FAD Ie + 6 4.2 THE HOMOGENEOUS LINEAR EQUATION WITH CONSTANT COEFFICIENTS A, tntroduction In this section we consider the special case of the nth-order homogeneous linear differential equation in which all of the coefficients are real constants. That i, we shall ‘be concerned with the equation eyed ie 09 2 4a, OF hag tay =O (a2 where ag. dy, 11 04876 Fea constants. We shall show that the general solution of tis equation can be found explicil. In an attempt to find solutions of a differential equation we would naturally inquire Whether or not any familiar type of function might possibly have the properties that ‘would enable it to be a solution. The differential equation (4.23) requires a function f126 [APUICIT METHODS OF SOLVING HICMEE.ORDER LINEAR DIFFIENTTAL EQUATIONS having the property such that if i and its various derivatives are each multiplied by certain constants, the a, and the resulting products, a,/®, are then added, the result will equal zero forall values of x for which ths result is defined. For this to be the ease we need a function such that its derivatives are constant multiples of itself. Do we know ‘of functions f having this property that e qa Ul = fle) {forall x? The answer is“yes,"for the exponential function f such that f(x) =e", where ‘mis.a constant, is such that ao Seen men ‘Thus we shal seek solutions of (423) ofthe form y = e, where the constant m will be ‘chosen such that e™* does satisfy the equation. Assuming then that y = e**isa solution for certain m, we have: ‘Substituting in (423), we obtain Aymte™ + asm 1 4 + ay. mem + aye" = 0 eM agmt + apm boot am + a) = 0. Since e™ 0, we obtain the polynomial equation in the unknown m: amt + aym! + aya gm + dy =O. 424) ‘This equation is called the auxiliary equation or the characteristic equation ofthe ven differential equation (423), If y= e* is a solution of (4.23) thon we see than the ‘constant m must satisfy (4.24) Hence, to salve (4.23), we write the auxiliary equation (4.24) and solve it for m. Observe that (4.24) is formally obtained from (4.23) by merely replacing the kb derivative in (423) by mM(k-=0,1,2,..ym). Three cases aris, according asthe roots of (424) are real and distinct, ral and repeated, or complex. B. Case 1. Distinct Real Roots ‘Suppose the roots of (4.24) are the m distinct real numbers42 THE HOMOGENEOUS LINEAR EQUATION Wik Constant coEFFICENTS 127 Then, omen setet are n distinct solutidns of (4.23), Further, using the Wronskian determinant one may show that these n solutions are linearly independent. Thus we have the following result THEOREM 4.11 Consider the nth-order homogeneous linear diferential equation (4.28) with constant coeficents. If the auxiliary equation (4.24) has then distinct real 0! mMs--sMa ‘then the general solution of (23) is Yo cy ce bone bce where cys €3.--.s€q are arbitrary constants D> Example 4.20 Consider the differential equation ‘The auxiliary equation is Henoe (m= Ym — =O, m= to m= ‘The roots are real and distinc. Thos e* and e are solutions and the general solution may be written ym eye + ene, We verify that e* and e* are indeed linearly independent. Their Wronskian is = et Jer 20 Thus by Theorem 4.4 we are assured of theit linear independence, Wee’) mer e0. > Example 4.21 Consider the diferential equation @y yy dy Aa as har ta Fm The awry equation is 4m? +m +6=0.128 exrucir metHovs oF soLvine MIGHER-ORDER LINEAR DIFFERENTIAL EQUATIONS We observe that m = ~1 isa root of this equation. By synthetic division we obtain the factorization (m+ 1m? — 5m + 6) = 0 (m+ 1m ~ 240m ~ 3) ‘Thus the roots are the distinct real numbers mee m ms =3, and the general solution is yrqet tee toe . Case 2. Repeated Real Roots ‘We shall begin our study of this case by considering a simple example. > Example 4.22: introductory Example ‘Consider the differential equation Py 6 #9 a0, 425 ie ae ‘The auxiliary equation is m? 6m 49 = or (m— 3 =O. ‘The roots of this equation are me3 med (real but not cited. | ‘Corresponding othe root m, we have the solution e™, and corresponding tom, we have the same solution e*. The linar combination ce + ¢,e™ of these “two” so- Jutions is clearly not the general solution of the differential equation (25) for itisnot a linear combination of two linearly independent solutions. Indeed we may write the combination e% + ¢ye% assimply coe™, wherecy = cy + cyiand clearly y= cye™, involving one arbitrary constant, is not the general solution ofthe given second-order equation ‘We must finda linearly independent solution; but how shall we proceed to do so? Since we already know the one solution ewe may apply Theorem 4.7 and reduce the order. We let» 42 THE HOMOGENEOUS LINEAR EQUATION WITH ConsTANT comncieNTs 129. where is to be determined, Then a en Bact se, Py rede 4 ger dd 4 ogss, Fea E+ oe 4 9c ‘Substituting into Equation (4.25) we have aa Ft, gaan dd. gerry) _ 6g 2 4 93m 3 (erty resect) 5000 pedo i Ee Letting w = do/dx, we have the first-order equation 0 de vet og acme or simply aw wo a ‘The solutions of this first-order equation are simply w = c, where is an arbitrary constant. Choosing the particular solution w = 1 and recalling that do/dx = w, we find Wo) = x Heo, where ¢y ian arbitrary constant, By Theorem 47 we know that for any choice of the constant cy, os}e2" = (x + cg}e™ isa solution of the given second-order equation (425). Further, by Theorem 47, we know that this solution and the previously known solution e* are linearly independent. Choosing cy = 0 we obtain the solution vax, ‘and thus corresponding to the double root 3 we find the linearly independent solutions eo and xe" ‘of Equation (425) ‘Thus the general solution of Equation (4.28) may be written yee bene 426 yl tee 427) With this example asa guide, let us return to the general nth-order equation (423).1f the auxiliary equation (4.24) has the double real root m, we would surely expect that e™* and xe would be the corresponding linearly independent solutions. Ths is indeed the ‘ease. Specifically, suppose the roots of (4.24) are the double real root mand the ( — 2) Aistinet real roots130 [RPC METHODS OF SOLVING HIGHER-ORDEE LINEAR DIFFERENTIAL CQUATIONS ‘Then linearly independent solutions of (423) are x0, 0 0 eens, and the general solution may be written YE Ce CHM Cet ge bot cyemen™ Ym (ey Heaxde™ 4 exe eye $f ge In like manner, if the auxiliary equation (424) has the triple real root m, ‘corresponding linearly independent solutions are xe™, and x22" ‘The corresponding part of the general solution may be written (er tegs teaxtem Proceeding further in like manner, we summarize Case 2 in the following theorem: THEOREM 4.12 1. Consider the nth-order homogeneous linear differential equation (4.23) with constant coefficients. Jf the auxiliary equation (4.24) has the real root m occurring ktimes, then the part of the general solution of (4.23) corresponding to this k-fold repeated root is (ey eax begs? to teat em 2. If further the remaining roots of the auxiliary equation (4.24) ae the distinct real smuanbers my 1-+-sMlqs then the general solution of (4.23) is PRC HEX ECA R eNO ey eRe Oe 4. Uf, however, any of the remaining roots are also repeated, then the parts of the general solution of (4.23) corresponding to each of these other repeated roots are expressions similar (0 that corresponding to m in part 1 We now consider several examples, > Example 4.23, Find the general solution of Py dy 4d BBs wy ‘The auxiliary equation m3 = dn? 23m + 18 has the roots, 3,3, -2. The general solution is, ym t eax + ge142 TE HOMOGENEOUS UNEAR EQUATION waTH constant Cosmicunts 131 Val teaxle oe, D> Example 4.24 Find the general solution of yy, gv, gd wea * aa tae ‘The auailiary equation is int — Sm? + 6m? + 4m —8 = 0, with roots 2,2,2, —1. The part ofthe general solution corresponding tothe threefold root 2is mle tex + eye and that corresponding tothe simple root —1 is simply y= ee™ Thus the general solution is y = yy + yay that is, Vm (ey bepx tex e™ + ge D. Case 3, Conjugate Complex Roots Now suppose thatthe auxiliary equation has the complex aumber a + bia, eal, = = 1,6 #0)as a nonrepeated root Then, since the cofliients are real, the conjugate complex number a — bi is also a nonrepeated root. The corresponding part of the general solution is Ayelet gle, where k, and k, are arbitrary constants. The solutions defined by e**¥ and e~¥* are complex functions ofthe real variable x. Iis desirable to replace these by two real linearly independent solutions. This can be accomplished by using Euler's formula, om cos + isin 8 Which holds forall real 6, Using this we have: belt hye kyetten te meet + ye) = e2[ky(cos bx + isin x) + ky(cos bx ~ isin bs)] Uk, + Koos bx + iy — ka)sin bz) = ettfey sin be + ¢, 608 bx], We borrow this basic ident rom complex varie theory, 25 wel asthe fact that = holds fr complex exponen.132 [EXPLICIT METHODS OF SOLVING HIGHER ORDER LINEAR DIFFERENTIAL EQUATIONS where ¢, = ilk, — yh es = ky + ky are two new arbitrary constants, Thus the part of the general solution corresponding to the nonrepeated conjugate complex roots a bi [ey sin bx +c, cos bx], Combining this with the results of Case 2, we have the following theorem covering Case 3 THEOREM 4.13 1. Consider the nth-order homogencous linear differential equation (4.23) with constant coaficiens. Ifthe auxiliary equation (4.24) has the conjugate complex rots @ + biand a ~ bi, neither repeated, then the corresponding part of the general solution of (6.23) may be written y= e%(c, sim bx +c 605 bx). 2. Ifphowever,a + biand a — blare each k-fold roots of the auxiliary equation (4.24), ‘then the corresponding part of the general solution of (4.23) may be written ye elles + eax + 6 4 Cean + cue eae? boot enaxt Neos bx) ‘We now give severat examples. > Example 4.25 Find the general solution of ey ae ty=% Wehave already used thisequationtllustrate the theorems of Section I. Letusnow ‘obtain its solution using Theorem 4.13, The auxiliary equation m? + 1 = 0 has the foots ni = sh. These ae the pure imaginary complex numbers abl, where a= 0, bb 1. The general solution is thus y He, sin Vx + €, 608 1x), ‘which is simply yc sin x + ¢; 608% > Example 4.26 Find the general solution of142 THe HOMOGENEOUS UNEAR EQUATION wiTH CONSTANT cotFHCENTS 133 The auniliary equation ism? — 6m + 25 = 0. Solving it, we find 62 JOT _ 648i 2 2 Here the roots are the conjugate complex numbers a + bi, where a= 3, b= 4, The general solution may be written £4 ymetey sin de + ¢,c08 40) > Example 4.27 Find the general solution of Ys osye 20 + a5y=0. ‘The auxiliary equation is m* = din! + 14m? — 20m + 25 = 0, ‘The solution ofthis equation presents some ingenuity and labor. Since our purpose in this example is not to display our mastery of the solution of algebraic equations but rather to illustrate the above principles of determining the general solution of dif- ‘erential equations, we unblushingly list the roots without further apologies. They are 142 1-2) 142, 1-2 Since each pair of conjugate complex roots is double, the general solution is Y= ele, + cax)sin 2x + (C5 + c4x}605 2x] = eye" sin 2x + egx0* Sin 2x + cye* 008 2x + cg xe" cos 2x ‘We now apply the results concerning the general solution of a homogeneous linear equation with constant coefficients to an initial-value problem involving such an equation. > Example 4.28 Solve the initial-value problem dy _ gay Get 8 gy t=O. (4.28) wo (429) y= -1 430)134 exrucir memos oF SOLVING HIGHER ORDER LINEAR DIFERENTIAL EQUATIONS First let us note that by Theorem 4.1 this problem has a unique solution defined for all x, 99 Example 4.20 Examples of UC functions of the four basic types (i), (i), (o} of the preceeding definition are those defined respectively by xe, SinBx/2},—cos(2K + 2/4), Examples of UC functions defined as finite produets of two or more of these four basic types are those defined respectively by Pe, xeds2e, eM sin 3x, sin 2x-c0s 3x, x7e** sin Sx The method of undetermined coefficients applies when the nonhomogencous function F in the differential equation is a finite linear combination of UC functions.142 [XPUCT METHODS OF SOLVING MIGHER-OROEE LINEAR DIFFERENTIAL EQUATIONS ‘Observe that given a UC function f each successive derivative of fis ether itself a ‘constant multiple of a UC function or else a linear combination of UC functions. DEFINITION Consider a UC function f. The set of functions consisting of J itself and all linearly independent UC functions of which the successive derivatives of f are either constant multiples or linear combinations will be called the UC set of J > Example 4.31 ‘The function f defined for all real x by fix) = x? is a UC function. Computing derivatives of J, we find PEI HIG, FH E=E 1, FMAI=O for n>3. ‘The linearly independent UC functions of which the successive derivatives of f are either constant multiples or linear combinations are those given by woox Thus the UC set of x? isthe set $ = {x*.x?,x,1) D> Example 4.32 ‘The function f defined for all real x by f(%) = sin 2x is a UC function. Computing Aeivatives of f, we find se) cos 2x, "(x)= 4 sin 2x, ‘The only linearly independent UC function of which the successive derivatives of fare Constant multiples or linear combinations is that given by ¢9s 2x, Thus the UC set of sin 2x is the set S = (sin 2x, €08 2x}. ‘These and similar examples of the four basic types of UC functions lead tothe eesuts listed as numbers 1, 2, and 3 of Table 4. > Bxample 4.33 ‘The fonction f defined for all real x by f(x) = x? sin xis the product of the two UC functions defined by x? and sin x Hence f is itself a UC function, Computing derivatives of fe find foo Sf") = Dsin x + dx 08 x — 7 sin x, ne sin x + x4 608 x, 7 (x) = 605 x — 6x sin x ~ x? 608 x, [No “new” types of functions will eccur from further differentiation. Bach derivative of {8 linear combination of certain of the six UC functions given by x? sinx, x7 c0s3x,43. THe Memo OF UNDETERMINED comments 143 UC function UC set ile ETc I) fae {eo} 3 | sindx + 00° {sin(bx + 0, costbx + 3} | costbx +€) 4 | xe (ote xt Feet t Fe xe 5 | xsinox+eor | fx"sin(@x +0, x*cox(bx +), x coslbx +c} xt sin(bx +9, x7"4 cos(bx + 0), sea ¥ sin(bx + 0, x costbx +0, sin(bx +), c08(bx + €)) 6 | etsindx+ oor | fe sin(bx + oe conlbx +0) | er costbe +0) 7 | setaade sor | (eerste saree ante + xe cosibx +c) | et eM sin(bx +0, x* Ve coslbx +0) et sinlbx + 0), xe™ cos(bx + 0), * sin(bx + o), e% coslox + 6} -xsinx, x00sx, sin x, and c0s.x. Thus the st s isthe UC set of x? sin x, Note carefully that x*,x, and | are not members ofthis UCset. ix? sin x, x? €05 x, x sin x, x 608 x, sin x, c0s x) ‘Observe thatthe UC set ofthe product x sin xis the set ofall products obtained by ‘multiplying the various members of the UC set (x?,x, 1} of x? by the various members of the UC set {sin x, cos x} of sin x. This observation illustrates the general situation regarding the UC set of a UC function defined as a finite product of two or more UC functions of the four basic types. In particular, suppose bisa UC function defined as the product of two basic UC functions f and g. Then the UCsset of the product function ‘his the st of all the products obtained by multiplying the various members of the UC set of / by the various members of the UC set of g. Results of this type are listed as numbers 4, §,and 6 of Table 4.1 and a specifcillustration is presented in Example 4.34 > Example 4.34 “The function defied for all el x by f(s) = x ca 2x isthe product ofthe two UC functions defined byx® and cbs 2x. Using the result sated inthe preceding paragraph, the Ue set of this product x os 2x ithe sto al products abixned by mliplyng ‘the various members of the UC set of x? by the various members of the UC set of 082. Using the definition of UCset or the appropriate numbers Table 4, we ind that the UC set of (2x8 at144 [EXPLICIT METHODS OF SOLVING HIGHER ORDER UNEAR DIFEEENTIAL EQUATIONS ‘and that of cos 2x is {sini 2x, c0s 2x}, ‘Thus the UC set of the product x? cos 2xis the set of all products ofeach of x2, x2, x, and 1 by each of sin 2x and cos 2x, and so itis fx? sin 2x, x 608 2x, x2 sin 2x, x? cos 2, x sin 2x, x 608 2 sin 2x, €08 2x). ‘Observe that this can be found directly from Table 1, number 5, with m = 3, = 2,and e=0. ‘We now outline the method of undetermined coefficients for finding a part integral y, of ay ay ao Ga tee Fame tt ay z Flo), +a, where F is finite linear combination Aus + Aatts ++ Aatly fof UC funetions uy, 3,.-.;tgy the A, being known constants. Assuming the comple mentary function y, has already been obtained, we proceed as follows: 1. For each of the UC functions of which Fis linear combination, form the corresponding UC set, thus obtaining the respective sets 85, S2y0 9S 2. Suppose that one of the UC sets so formed, say S, is identical with or completely included in another, ay S,. In this ease, we omit the (identical or smaller) set S, from further consideration (retaining the set S). 3, We now consider in turn each of the UC sets which stil remain after Step 2 Suppose naw that ane ofthese UC sets, say S, includes one or more members which are solutions of the corresponding homogeneous differential equation. If this isthe case, wwe muluply each member of S; by the lowest positive integral power of x so that the resulting revised set will contain no members that are solutions of the corresponding homogeneous dilferential equation. We now replace S, by this revised set, so obtained. Note that here we consider one UC set at a time and perform the indicated multi plication, if needed, only upon the members of the one UC set under consideration at the moment, 4. In general there now remains: certain of the original UC sets, which were neither omitted in Step 2 nor needed revision in Step 3, and (ii) cettain sevised sets resulting from the needed revision in Step 3. Now forma linear combination ofall of the sets of these two categories, with unknown constant coefficients (underermined coefficients) ‘4. Determine these unknown coefficients by substituting the linear combination formed in Step 4 into the differential equation and demanding that it identically satisfy the differential equation (that is, that it be » particular solation)42 THE METHOD OF UNDETEEMNED councENTS 145, This outline of procedure at once covers al of the various special eases to which the method of undetermined coefficients applies, thereby freeing one from the need of considering separately each ofthese special cases Before going on tothe illuscative examples of Part C following, let us look back and observe that we actually followed this procedure in solving the dilerental equations (436) and (439) ofthe Introductory Example 4.29. In each of those equations, the nonhomogeneous member consisted of a single term that was constant multiple of 2 ‘UC function; and ineach case we followed the outline procedure step by step, as farasit applied For the differential equation (4.36, the UC function involved wase* and we formed its UC et, which was simply {e** (Step 1, Step Zobviously did not apply. Nor did Step 3, for as we noted later, e** was not a solution of the corresponding homogeneous ‘equation (4.4). Thus we assumed y, = Ae“ (Step 4) substituted in differential equation (436) and found 4 and hence yp (Step 5) For the differential equation (439), the UC function involved wase™; and we formed its UC set, which was simply (e*} (Step 1) Step 2 did not apply here cither. But Step'3 ‘was very much needed, for e was a solution of the corresponding homogeneous equation (441), Thus we applied Step 3 and multiplied ein the UC set {e**) by x, ‘obtaining the revised UC set {xe"*}, whose single member was nota solution of (441) Thus we assumed y, = Axe* (Step 4), substituted in the differential equation (39), and found A and hence y, (Step 5) ‘The outline generalizes what the procedure for the differential equation of Introductory Example 429 suggested. Equation (439) of that example has already brought out the necessity for the revision described in Step 3 when it applies. We give here a brief itustration involving this ertical step. DP Example 4.35 Consider the two equations (44a) and as) ‘The UC set of x7e* is The homogeneous equation corresponding (o (444) has linearly independent solutions e* and e**, and so the complementary function of (4.44)is y, = ce + ce. Sinct member e*of UCset Sis qsolutonof the homogeneous enuation corresponding to(td) we multiply each member of UC by te owest postive integral power of ‘20 thatthe resling eved ge wl contain no members ha ae solution of the homogeneous equation corespondng to (440) Ths tums out to be ise forthe towed et S = fete 20h xe") has no members that satisfy the homogencous equation corresponding to (6.48),146 [RPLICIT METHODS OF SOLVING HICHER ORDER LINEAR DIFFERENTIAL EQUATIONS The homogeneous equation corresponding to (445) has linearly independent solutions e* and xe", and so the complementary function of (4.45)is y, = cye* + exe Since the two merabers e* and: xe* of UC set $ are solutions of the homogeneous equation corresponding to (445), we must modify S here also. But now x itself will not do, for we would get $°, which still contains xe%. Thus we must here multiply each member of S by x# to obtain the revised set So fate et eet, Which has no member that satisfies the homogencous equation corresponding to (4.49). . Examples AA fow illustrative examples, with reference to the above outline, should make the ‘procedure lear. Our first example will be simple one in which the situations of Steps 2 and 3 do not occur. > Example 4.36 2" — 10sin x. and the complementary function is nee ene ‘The nonhomogenous term is the linear combination 2e* — 10 sin x of the two UC Janetions given by e* and sin x 1. Form the UC set foreach ofthese two functions. We fing S.= le}, 5, = fsin x, 008 x} 2. Note that neither of thes sets is identical with nor included inthe other; hence both ate retained. 3. Furthermore, by examining the complementary function, we see that none of the functions ein x, €08 «in ether of these sets is a Solution ofthe corresponding homogeneous equation, Hence neither set nezds to be revised. 44. Thus the original sels S, and S, remain intact in this problem, and we form the linear combination Act + Bsinx + Ccos x of the three elements ¢*, sin x, cos x of S\ and S,, with the undetermined coeficents ABC.43. Te MeMUOD OF uNDKTERNED comnts 147 5, We determine these unknown coefficients by substituting the linear combination formed in Step 4 into the differential equation and demanding that it satisfy the dif- ferential equation identically. That is, we take 9p = Ae + Bsin x + Coos x 5 a particular solution. Then yy = Alt + Beos x Csin x, Y= det ~ Bsin x — Coos x. Actually substituting, we find (Act — B sin x — C.cos x) — 2(de" + Bos x ~ C sin x) : ~ MAe* + Bsinx + Cos x) = 2e" ~ 10sinx or Ade" + (4B + 20)sin x + (AC ~ 2805 x = Det — 10 sin x Since the solution isto satisfy the diferential equation identically forall x on some real interval, this relation must be an identity forall such x and hence the coeficients of like terms on both sides must be respectively equal. Equating coefficients ofthese like terms, we obtain the equations —44=2, -4B42C=-10, ~4C-2B=0. From these equations, we find that and hence we obtain the particular integral Jp ~4e* + Dsin x — 08 x ‘Thus the general solution of the differential equation under consideration is eye + cge* Je + Bsn x — cos x > example 4.37 4+ Dy = 2x2 4 et 4 axe" + de ‘The corresponding homogeneous equation is wy rs and the complementary functidh i dem ce + eye ‘The nonhomogeneous term isthe linear combination 2x? bet + Duet + Ae of the four UC functions given by x}, e xe", and e%,148 exrcit mer#00s OF SOWING HIGHER ORDER LINEAR DIFFERENTIAL EQUATIONS. 1. Form the UC set foreach ofthese functions. We have Sy= Ax 1h Site) Sy eee), Sam te) 2 We note that 5; is completely included in S,, 50 5, is omittod from further consideration, leaving the three sets Sate 1) y= feeve}, Sy = fe). 3. We now observe that S; = {xe,e*) includes e*, which is included in the complementary function and 0 isa solution of the corresponding homogeneous Example 4.39 An Initial-Value Problem ‘Weeclose this section by applying our results tothe solution of the initial-value problem @y 44 - Fe 2h sy ret 108ia (4.46) WO)= 2, (447) yO=s (aay By Theorem 4. this problem hasa unique solution, defined forall x, —20 Example 4.40 Consider the differential equat a, grt ystane. (4.63) The complementary function is defined by ys) = eysin x + €,c08 x We assume 4b} = oy (x)sin x + v9(2)008 x, (450) ‘where the functions», and v, will be determined such that thisisa particular integral of the differential equation (463). Then pla) = v,(2)e08 x ~ valx)sin x + v5 (a)sin x + v4(x)c08 x We impose the condition vi Ge)sin x + vy(a)eos x = 0, (465) leaving pls) = oy (xJ008 x ~ 04 (sin x From this Vole) oy x)sin x — valapeos x + x {x)e0s x ~ eafxJsinx. (4.66) Substituting (4.64) and (4.66) into (4.63) we obtain ¥4(x)e08 x ~ v3 (x)sin x = tam (aon[RPUCIT METHODS OF SOLVING HIGHER-ORDER LINEAR DIFFERENTIAL EQUATIONS Thus we have the two equations (4.65) and (4.67) from which to determine o(x), v2(8): 4 (x)sin x + v5 (x)eos bCx}e08 x — v3(x)sin Solving we find sinxtan x _—sin* =I ‘cose Integrating we find: 20) = Substituting (4.68) into (4.64) we have Yolo) = (—co8 x + es)sin x + (sin x —In see x + tan x] + ¢4)c08 x cox+ ey, vax) =sin x—In|seox + tam x] +cy. (468) sin x cos x + eysin x + sin x cos x In see x + tan x| (cos x) + c4oos x = egsin x + cye0s x ~ (Gos x}in [see x + tan x{h Since a particular integral isa solution free of arbitrary constants, we may assign any particular values A and Bto.c, and c, respectively, and the result will be the particular integral Asin x + Boos x — (cos xin jsee x + tan xi). Thus y = y, + yp becomes y= cq Sinx +, €08.x + Asin x + Bos x ~ (cos xMn see x + tan x!), which we may write as y= Cy sin x + C, c0s x ~ (cos xMin [see x + tan x1), where Cy = 0) +A, Cz =e + B. ‘Thus we see that we might as well have chosen the constants c, and c, both equal to {in 4.68), for essentially the same result, y= ey Sin x + cy €08 x — (cos x}In [see x + tan x), would have been obtained. This is the general solution of the differential equation (463. ‘The method of variation of parameters extends to higher-order linear equations. We ‘now illustrate the extension to a third-order equation in Example 4.41, although we44 vantarion oF patamerers 159° hhasten to point out that the equation of this example can be solved more readily by the method of undetermined coefficients. D> Example 4.41 ‘Consider the differential equation @y fy ye we Saget ug e (4.69) ‘The complementary function is la) exe? + exe + ese assume as a particular integral Ypla) = oy (ale + nylahe™ + og(ede™ 470) ‘ince we have three functions 1, v2, 05 at our disposal in this case, we can apply three ‘cofditions. We have: Yols) = oy(adeF + Doane it Bosae™ + oh adet + vA + ob (ade Proceeding in # manner analogous to that ofthe second-order case, we impose the condition vi Gadet + vax) + vy (xe am leaving Vals) = o (xe + Dog(a}e™ + Fos(a)e™, an Then Yglx) = oGxdet + Aoabs}eP* + Duslx}e™ +o (xlet + Doptsle™ + Bu5(ale™. ‘We now impose the condition yQ)et + 2h (ade™ + Buy(ade™ = 0, 473) reaving pla) = oylalet + doglade™* + 90500). 474) From this, ps) = 8 (she + Boy(ae™* + 2Tog(sle™ + vi (ser + dogla)e™ + 904 xe. 475) We substitute (4,70), (4:72), (4.74), and (4.75) into the differential equation (4.69), obtaining: ylalet + Boa(a}e™ + 2Tos(xde™ + vi(xe™ + dos(ale™ +-904(xe* ~ 60, (x}e* ~ 2bvg(a)e™ — Séog(x}e™ + Hoy a)et + 22oa(nhe® + 33os(x)6* Eo, (x}e — Go g(xe™ — 6os(x}e* = e* or Wilade? + dude? + Suy(ade™ = et (4276)160 exticir MetH00s OF SOLVING HIGHER-ORDER LINEAR DIFFERENTIAL EQUATIONS Thus we have the three equations (4.71), (4.73), (4.76) from which to determine ¥4 03), 220) wala: viet + vyta)e™ + of(x)e™* = 0, Bi (oer + Ina (aJe™ + Bog(a)e™ = 0, Ui(a)eF + 4oy(a)e4* + 4 lx)e™ = eF Solving, we find aera 0 2 ae} tt ty(ey alee fete ee 23] a Wm | td ye je 2e2* 363" 12 3 je ae gem }> [1 4 9) pee ee ee eae eee 0) = er | ex 22 30%) ese gest giael ee 0 ete Of ll je 4 et |_| 1460) = SS = ae er ett 30% Jet det Ge We now integrate, choosing all the constants of integration to be 2er0(as the previous ‘example showed was possible) We find: fx, vale) ay(x) = — he" Thus Axet 4 eel — heme = fet + Bet, ‘Thus the general solution of Equation (4.53) is Yet Ip spe + coe + eye + hxc! + det Yr cjetcge™ + eset dre, where e} <6 +3 {In Examples 4.40 and 4.41 the coefficients in the diferential equation were constants. ‘The general discussion at the beginning of this section shows that the method applies equally well to linear differential equations with variable coeficients, once. theSee ee ae eee ee ee eee eee eee “44 vananion oF patameras 161 complementary function y, is known. We now illustrate its application to such an equation in Example 4.42. D> Example 4.42 Consider the diferental equation a4 Pref 1p 4 EE a6 B25 = 6 +0) am {In Example 4.16 we solved the corresponding homogencous equation t+ nth 284 270 OF + DEF 2 aya. From the results ofthat example, we se thatthe complementary function of equation «Mis yds) = 04% + exe = 1) To find a particular integral of Equation (4.77), we thercfore let Yolo) = L(x)x + Pala}? — 0 (478) Thea Yl) = v0) 1 vga) 2e + HL) + O(N? — We impose the condition BiG) + 0,(x)0e? — 1) = 0, amy leaving ylX) = 0400) + wale): 2s. (4380) From this, we find Vela) = v(x) + 204(2) + of(2) 2. (aan ‘Substituting 4.78), (480), and (481) into (4.77) we obtain (£2 + 19D0, (3) + 2op(x) + 2e0(3)] ~ Defoe) + 230(8)] + Df bx) + eolalix? — 1) = 6G + 0? (2 + oy Ca) + Derglal] = 662 + DE (asa) ‘Thus we have the two equations (4.79) and (4.82) from which to determine v(x) and 1, G4}; that is, v(x) and v(x) satisly the system vi Gx + 03 G)Le? 1 8,08) + 02G0[2x) = 6(x? + 1.162 cxruicrr merHo0s OF SOLVING HIGHER-ORDER LINEAR DIFFERENTIAL EQUATIONS Solving this system, we find 6x? + 107" 1) et 66x? = 1), _&G2 +) ar 7S wy) = 2x + 6x, a(e) = 3x3, (483) ‘where we have chosen both constants of integration to be zero, Substituting (4.83) into (4.78), we have ple) = (—2e! + 62)x + 3x2? ~ 1D) xt 38 “Therefore the general solution of Equation (4,77) may be expressed in the form Teed yy : a eye + e(x? = xt + 3 Exercises Find the general solotion of each of the differential equations in Evercises 118. e : 1 Ye yncas 2 Phe yates ® & % i 2 He yewex 4 Seapets 5 5. S24 ay = sect 2x, aS 6 Oe yatanxsecx. ks #y ate 1 sada yee scx #y dy BOR 9. Phases + Sy =e* tan 2x epee. 10 2-28 aya retinxtero,18. 19. 20, 21 2 44 vamarion or parawerets 163 ‘ yarns (50, Find the general solution of wy 6 23x + 6x, Fo 6B + 10y 2388 + 60% given that y= x and y= x3 are linearly independent solutions of the corresponding homogeneous equation. Find the general solution of 2 ay HP EI Mee Zar = 1, given that y =x + 1 and y = (x + 1)? are linearly independent solutions of the corresponding homogeneous equation. Find he eel sokton of e+ ry 20+ 04 2ym0c42, given that y= x+1 and y =x? are linearly independent solutions of the ‘corresponding homogeneous equat Find the general solution of 2d dy testes aes ay given that y =x and y = xe* are linearly indepen-tzut solutions of the corre- sponding homogeneous equation.164 exrvcrt memos oF SotvINe MCHERORDER UNEAR DIFFERENTIAL EQUATIONS 23, Find the general solution of, #y pein = axe 2Pet abe = 959 be? = 2) 2h = Dy = 37 — Pes, given that y =e and y = x? are linearly independent solutions of the corre sponding homogeneous equation 24, Find the general solution of Py 4 ext n+ WE + a weet 0 siven that y = x and y=(x + 1)" are linearly independent solutions of the corresponding homogencous equation, 25, Find the general solution of 242% 2 (Gin? 2) G2 ~ 2 sin x 003 x in? x, feos? Get ost x + Dy given that y = sin x and y = x sin x are linearly independent solutions of the corresponding homogeneous equation, 26, Find the general solution by two methods: @y ty dy ae Pat ae 4.5 THE CAUCHY-EULER EQUATION ‘A. The Equation and the Method of Solution In the preceding sections we have seen how to obtain the general solution of the nth- ‘order linear differential equation with constant coefficients. We have seen that in such cases the form of the complementary function may be readily determined. The general nth-ordet linear equation with variable coefficients is quite a different matter, however. and only in certain special cases can the complementary function be obtained explicity in closed farm. One special case of considerable practical importance for which i is fortunate that this can be done is the so-called Cauchy-Buler equation (or equl= dimensional equation). This is an equation of the forn WEF gyn POY 43 bay = Fe) (4 Age Tee PON Ter tt aT bay Fla), (484) where oy Ase.-osite-av Me Ae constants. Note the characteristic feature of this equation: each term in the let m form ber is constant multiple of an expression of the45 THe caucny-cutee equation 165, How should one procsed to solve such an equation? About the only hopeful thought that comes to mind at this stage of our study is to atterapt a transformation. But what transformation should we attempt and where will it lead us? While it is certainly worthwhile to stop for a moment and consider what sort of transformation we might use in solving a “new” type of equiation when we first encounter it itis certainly not ‘worthwhile (0 spend a great deal of time looking for clever devices which mathema- ticians have known about for many years. The facts are stated inthe following theorem. THEOREM 4.14 The transformation x = e! reduces the equation By gett aor gr tae S 10.a linear differential equation with constant coeficlents petaceayero aH This is what we need! We shall prove this theorem for the case of the second-order ‘Cauchy-Euler differentia! equation 2 PY ax + ayy = FO) ag? SE tax 2 + any = Fo (485) ‘The proof in the general nth-order case proceeds ina similar fashion. Letting x = e, assuming x > 0, we have f = In x. Then boa iy ae” at dx x dt and ay), yd (1) (Gt) aaa) 1 (yl) bat (@y_ a x\aP x) ar ae Tus ayy 9 ona 2 Substituting into Equation (4.85) we obtain Py ay\ ay My a) a Os aay = Feel (3 ta; Ft ay = Fe) oi a Area, GW= Fle166 exructr meTHODs OF SOLVING HICHER-ORDER LINEAR DIFFERENTIAL EQUATIONS. This sa second-order linear differential equation with constant coefficients, which was ‘what we wished to show. Remarks. 1. Note that the leading coefficient ag x" in Equation (4:84) is zero for x = 0, Thus the basic interval a < x < 6, referred to in the general theorems of Sec- tion 4.1, does not include x = 0. 2 Observe that in the above proof we assumed that x > 0.1 x < 0, the substitution x= —e' is actually the correct one. Unless the contrary is explicitly stated, we shal! ‘assume x >0 when finding the general solution of a Cauchy-Euler differential equation. B. Examples D Example 4.43, 487) ‘Thus Equation (4.87) becomes fy ae ay a (4388) ‘The complementary function of thisequation is y, = cye' + cxe% Werfind particular integral by the method of undetermined cosficients. We assume. yp = Ae, Then p= 3Ae™, yg = 9de™, and substituting into Equation (4.88) we obt tet =e ‘Thus A = J and we have y, = fe The general solution of Equation (88) is then pace tere + fet But we are not yet finished! We must return to the original independent variable x Since e' = x, we find yout ee the ‘This is the general solution of Equation (4.87). Romarks. 1. Note carefully that under the transformation x = ¢' the right member of (4.87), x°, transforms into e™. The student should be careful to transform both sides45. tie caucny-cuter equation 167, ‘of the equation if he intends to obtain a particular integral ofthe given equation by finding a particular integral ofthe transformed equation, as we have done bere. 2. We hasten to point out that the following alternative procedure may be used. After finding the complementary function of the transformed equation one can in mediately write the complementary function of the original given equation and then proceed to obtain a particular integral ofthe original equation by variation of param- cers. In Example 443, upon finding the complementary function cye' + ce of Equation (4.8), onecan immediately write the complementary function ¢,x + ¢,x? of Equation (487), then assume the particular integral y,(x) = vy(x)x + ty(e)s", and from here proceed by the method of variation of parameters. However, when the ‘nonhomogeneous function F transforms into @ linear combination of UC functions, as it does inthis example, the procedure ilustrated is generally simpler. > example 4.44 22Y a ga ly pie pe y . By = din x. (489) Assuming x > O, we etx = e4 Then = In x and ‘Thus, substituting into Equation (889), we obtain ey 481.8) 4(€ G@ 3 at) 4(Ge- ~ ty ate or ey 48 gt i by a4 4.90) Sha 1F + ayaa 490) ‘The complementary function of the transformed equation (4.90) is yom yet + ce + cnet168 cxrucir merno0s OF SOLVING HIGHER-ORDER LINEAR DIFFERENTIAL EQUATIONS | We procnd to obtain «partlar integral of Equation (4.90) by the method of un determined coins, We assume j, = A+ B.Then p= A, 9, = Jp = Sub tung into Equation (0) we nd 144 — Bat 8B = 4t. | -84=4, 44-88 =0, |. Thus the general solution of Equation (4.90) is yaad tee" tee Hak and so the general solution of Equation (489) is year text text bine} Remarks, In solving the Cauchy-Euler equations of the preceding examples, we observe that the transformation x = ef reduces ay, @y of 0 Gr ayy Py yey ee Poe ‘We now show (without proof) how to find the expression into which the general term ay ae dy +29. | where mis an arbitrary positive integer, reduces under the transformation x = ¢. We | present this as the following formal four-step procedure. 1. For the given positive integer n, determine Hr — Mr —2)-f — tn DD 2 Expand the preceding as a polynomial of degree m in r. | 3. Replace by PF. for each k= 1.2.3... ty ast | For example, when n = 3, we have the following illustration. 4, Equate x" to the result in Step 3 1. Since n= 3,0 ~1 = 2.and we determine r(r — 1)r — 2). 2. Expanding the preceding, we obtain r? — 3r2 + 2r Py yf yt a ae ae45 Tie caucny-suten equavion 169 4 Bau? 2 to this, we have the relation dy yy dy Rae ae a ‘Note that this is preisly the elation we found in Example 444 and stated above Exercises Find the general solution of each ofthe differential equations in Exercises 1-19.Ineach ‘ease assume x > 0. xX ayna, 2 tap ay #4 1y=0. ay “BF ty 4 eB aS saan ay canoe 6 x #y 6 spy = 1B. £3 eB + 18700. ad@y ay Et 4c ® syn aes 20y 5 Ws gy odes is BET ate ity dy 16 Th aes ye dine ay, a nL May atin 1, Eh eB say ares170 expucit merHo0s OF SOLVING MIGHER-ORDER LINEAR DIFFERENTIAL EQUATIONS. 28 4 Ms ym asinin 8 Dax Beye asinin a@y dy 7 wv. PE Eh ae ayer Solve the initial-value problem in each of Exercises 20-27. In each case assume x > 0. 0 eS ae tym, n= Re y=. #y : a. vo Zee = vans Maye Ofer vv Phase Zt ay w= yt we fhyem ned” ys ayy dy 26 i 2 Eade aye dst 6, 24 Y= HL 2Py ody 2 y= 25, 8 haga tor, tp y= ~6. 26 EP eye, =a yO x. efh—oy=inn =k y= 28. Solve: cere SY y+ 92-30 0. 29. Solve = 9 EY 6a — 3) 4 try 0 (x ~ 39 FF 60x ~ 92 + 12y = 0. 4.6 STATEMENTS AND PROOFS OF THEOREMS ON THE SECOND-ORDER HOMOGENEOUS LINEAR EQUATION Having considered the most fundamental methods of solving higher-order linear differential equations, we now return briefly to the theoretical side of the subject and present detailed statements and proofs of the basic theorems concerning the second ‘order homogencous equation. The corresponding results for both the general nth- order equation and the special second-order equation were introduced in Section 4.18 and employed frequently thereafter. By restricting attention here to the second-order case we shall be able to present proofs which are completely explicit in every detail However, we point out that each of these proofs may be extended ina straightforward“46 TOROS ON THE SECOND-ORDER HOMOGENEOUS UNEAR EQUATION 171 ‘manner to provide a proof ofthe corresponding theorem for the general nth-order case. For general proofs, we again refer to Chapter I We thus consider the second-order homogencous linear differential equation ay ay nt 2960) GE + 41 (0) E+ onlay = 0 (49) Where a5, 4,, and a, are continuous real functions on a real interval a < x Oisealled the damping constant. When the massis moving downward, Facts inthe upward direction (opposite to that of the motion) and so Fy < 0. Also, since mis moving downward, xis increasing and dat is postive. Thus, assuming Equation (5.4) 10 hold, when the mass is moving downward, the damping force is given by (a>o, 6s)182. srrucations oF stcOND-ORDER LINEAR DIFFERENTIAL EQUATIONS 52 ‘This also gives the damping force when the mass is moving upward, as one may see by replacing each italicized word in the three preceding sentences by its opposite. 4. Fe, any external impressed forces that act upon the mass."Let us denote the resultant of all such external forces at time ¢ simply by F(t) and write A= Fi). 66) We now apply Newton's second law, F = ma, where F = Fy + Py + Fy + Fe. Using (6.1), (5.3), (5.5), and (56), we fin x dy mG = mg — kx — mg a + FU) or ax dx : mir ath + kx = Fl 59) This we take as the diferential equation for the motion of the mass on the spring. Observe that itis a nonhomogeneous second-order linear differential equation with constant coeficients, If ¢=0 the motion is called undamped, otherwise it is called ‘damped 1F there are no external impressed forces, F() = O forall and the motion is called fre: otherwise is called forced. In the following sections we consider the solution of 5.7) in each of these cases FREE, UNDAMPED MOTION. We now consider the special case of free, undamped motion, that is, the case in which both a = 0 and F(t) = 0 forall r. The differential equation (5.7) then reduces to x m5 + hem 0, 68) where m(>0) is the mass and k(>0) is the spring constant. Dividing through by m and Jetting k/m = 2, we write (58) in the form where c, and e, are arbitrary constants. Let us now assume" that the mass was initially displaced a distance x from its equilibrium position and released from that point with initial velocity vg. Then, in Addition to the differential equation (5.8) for (59)]. we have the initial conditions x0) = Xo. on $10) = t (512)32. Fue, unpawren Morin 183, Differentiating (5.10) with respect to 1, we have & ee B oc roost ey sin it 613) Applying conditions (S11) and (5.1210 Equations (S10) and (8.13) spectively, wesee at once that ett, ‘Substituting the values of c and cy so determined into Equation (5.10) gives the particular solution ofthe diflerential equation (5.8) satisfying the conditions (5.11) and (6:12) in the form x= "8 sin dt + x9 £08 2 2 We put this in an alternative form by fist writing it as afi sinat 720s} 619) where : wy ‘Then, eting 516 Equation 514 reduces at once to x= coost+ #), on where cis given by Equation (5.15) and ¢ is determined by Equations (5.16). Since 1. /kim, we now write the solution (5.17) in the form a. sm ‘This, then, gives the displacement x of the mass from the equilibrium position O as a function of the time t(t > 0). We see at once that the free, undamped motion of the mass 's a simple harmonic motion. The constant is called the amplitude of the motion and ives the maximum (positive) displacement of the mass from its equilibrium position, ‘The motion isa periodic motion, and the mass oscillates back and forth between x = ¢ and x = =e, We have x = cif and only if184 APPLICATIONS OF SECOND-ORDER LINEAR DUTERENTIAL EQUATIONS. n= 0,1,2,3,...j¢ > 0. Thus themaximum (positive) displacement occursif and only it to fier —a>0 (619) where = 0,1,2,3, “The time interval between two successive maxima is ale the period ofthe motion Using (19) nese tha ti ven by ae Jen “The reciprocal of the period, which gives the number of oseitations pec second, is called the natural frequency (or simply frequency) of the motion. The number gis called ‘the phase constant (or phase angle). The graph of this motion is shown in Figure 5.2 (520) Figure 5.2 D Example 5.1 ‘An 8-4b weight s placed upon the lower end ofa col spring suspended from the ceiling ‘The weight comes to retin its equilibrium position, thereby stretching the spring 6 in, ‘The weight is then pulled down 3 i. below its equilibrium position and released at 1 = Owith an initial velocity of 1 f/sec, directed downward. Neglecting the resistance fof the medium and assuming that no external forces are present, determine the am plitude, period and frequency of the resulting motion. Formulation. ‘This is clearly an example of fre, undamped motion and hence Equation (58) applies. Since the 8lb weigh stretches the spring 6 in, ~ $t, Hooke’s Jaw P= ks gives 8 k()) and so k= 16 bf. Also, m = w/g = $s (lugs), and 30 Equation (58) gives eae nae +16 or fF ona wan2 fret, uNpaween moTION 185, Since the weight was released with a downward initial velocity of 1 ft/sec from a point 3in.(= 4) below ite equilibrium position, we also have the initial conditions x0=4, XO=1 (522) 1. The auaitiary equation corresponding to Equation (521) isr? + 64 = 0, £81, Thus the general solution of the differential equation (5.21) may be e; sin 8+ €3 608 8, (523) Where cy and ¢, ae arbitrary constants. Applying the fist of conditions (522) to this, we find cy = 4 Dileretiating (5.23), we have ax Fe = Bey 0s Br — Beg sin Applying the second of conditions (5.22) to this, we have 8c, = 1 and henoe cy = ‘Thus the solution of the differential equation (521) satisfying the conditions (5.22) is xa dsin Bt + $008 8 (524) ‘Let us put this in the form (5.18). We find ‘and thus write Thus letting sin 6 5 Wa (52) aa we write the solution (5.24) in the form J xa conlte +o), (526) where i determined! by Equations (5:25). From these equations we id that = 6 radians Taking /5 = 2236 the solution (526) thus given approximately by (0280 cos(8+ — 0.46) The amplitude of the motion /5/8 = 0280. By formula (520), the petiod is 2/8 = 4/4 ee), and the Trequcne is 4/ oxillations/ee. The graph is shown in Figure 53 Before leaving this problem, fetus be certain that we cams up ital conditions conety. Let us replace the thd sentence inthe statement ofthe problem by the following: “The weight i then pushed up +n, above its equlibrium postion and186 Arnuicarions oF seconp-onDee LINEAR DIFFERENTIAL EQUATIONS Figure 5.3 released at ¢ = 0, with an initial velocity of 2 ft/sec, directed upward The initial ‘conditions (5.22) would then have been replaced by x)= 5, x@) = ~2 “The minus sign appears before the } because the initial position is 4 in, = } foot above ‘the equilibrium position and hence is negative. The minus sign before the 2 is due to the fact that the initial velocity is directed upward, that is, in the negative direction, Exercises Noto: In Exercises 1-7 neglect the resistance of the medium and assume that no external forces are present |. A 12-10 weight is placed upon the lower end of a coil spring suspended from the ceiling, The weight comes to rest in its equilibrium position, thereby stretching the spring 1.5 in. The weight is then pulled down 2 in, below its equilibrium position and released from rest att = 0, Find the displacement of the weight as function of, the time; determine the amplitude, period, and frequency of the resulting motion; and graph the displacement as a function of the tre, 2. A 16-lb weight is placed upon the lower end of a col spring suspended vertically from a fixed support. The weight comes to rest in its equilibrium position, thereby stretching the spring 6 in, Determine the resulting displacement as a function of time in each of the following cases (a) If the woight is then pulled down 4 in. below its equilibrium position and released at ¢ = O-with an initial velocity of 2 ft/sec, directed downward. (b) IF the weight is then pulled down 4 in, below its equilibrium position and released at ¢ = 0 with an initial velocity of 2 f/see, directed upward. (6) If the weight is then pushed up 4in. above its equilibrium position and released ut 1 = 0 with an initial velocity of 2 fysee, directed downward,| \ | | | | | 52 free, uNDAwPeO MOTION 187 3. A 4b weight is attached 10 the lower end of a coil spring suspended from the ceiling. The weight comes to rest in its equilibrium position, thereby stretching the spring in, Attime = Othe weight is then struck so as to setit into motion with an initial velocity of 2 ft/sec, directed downward, (@) Determine the resulting displacement and velocity of the weight as functions of the time, (©) Find the amplitude, period, and frequency of the motion, (©) Determine the times at which the weight is 1.5 in. below its equilibrium Position and moving downward (4) Determine the times at which itis 1.5 in, below its equi ‘moving upward. jom position and 4. A 641b weight is placed uposi the lower end of & col spring suspended from a cigié beam. The weight comes to res in its equilibrium position, thereby stretching the spring 2ft. The weight's then pulled down 1 ft below its equilibrium position and released from rest at = (@)_ Whats the position ofthe weight at moving atthe time? (&) At what time is the weight 6 in. above its equilibrium post downward? What is its velocity at such time? i /12? How fast and which way iit and moving 5. A coll spring is such that a 25.1b weight would stretch it 6 in. The spring is suspended from the ceiling, a 16-b weight is attached to the end of it, and the weight then comes to rest in its equilibrium position. Its then pulled down 4 in below its equilibrium position and released at ¢ = 0 with an initial velocity of 2 fifse, directed upward. (2) Determine the resulting displacement of the weight as a function ofthe time, (6) Find the amplitude, period, and frequency of the resulting motion. (0) Atithat time does the weight fest pass through its equilibrium position and ‘what is its velocity at this instant? 6. An8-1o weight is attached to the end of a coil spring suspended from a beam and comet (ret ints equlfvium postion, The wight ten pul down A el below te equlbrivm potion and teesed at f= wth an inal velo of 5 jue dreceddownvard Detrine the apringconstant kan th constant i the ample of he esting motions andthe peso 2 7. An Selb weight is placed atthe end of a col spring suspended from the ceiling. After ‘coming to retin its equilibrium position, the weight is set into vertical motion and the period of the resulting mation is sec. After atime this motion is stopped, and the 8b weight is replaced by another weight, After this other weight has come t0 rest in its equilibrium position, itisset into vertical motion, Ifthe period ofthis new motion is 6 sec, how heavy is the second weight? 8. Assimple pendulum is cognposed of a mass m (the bob) atthe end ofa steaight wire of negligible mass and length | It is suspended from a fixed point $ fits point of support) and is fee to vibrate in a vertical plane (see Figure 5). Let SP denote the straight wire; and let @ denote the anglc that SP makes with the vertical SP, at time "postive when measured counterclockwise, We neglect air resistance and assume that only to forces act on the mass m: F, the tension in the wire; and Fy, the force188 arpucarions oF seconD.ORDER LINEAR DIFFERENTIAL EQUATIONS 3 Figure 5.4 due to gravity, which acts vertically downward and is of magnitude mg. We write F, = fy + Fy, where Fri the component of F; along the tangent to the path of m and Fy s the component of F; normal to Fr. Then fy = —F, and F; = —mg sin ‘and so the net force acting on mis F, + Fz = F, + Fy + Fy = —mg sin 8,alongthe are P,P. Letting sdenote the ength of the arc P,P, the acceleration along this arcis, ¥s/de® Hence applying Newton's second law, we have md#s/dt? = —mg sin 8. But since s = 19, this reduces tothe differential equation a6 oo 9 miSa= —mgsind ot So4 Fsin (a) The equation 0 bin = Gat ising =0 is a nonlinear second-order differential equation. Now recall that sind =0— Hence if 0 is sufficiently small, we may replace sin @ by 8 and consider the ‘approximate linear equation Ti+ fona Assume that = @ and d9/dt = 0 when # = 0. Obtain the solution of this approximate equation that satisfies these initial conditions and find the amplitude and period of the resulting solution. Observe that this period is independent of the initial displacement (b) Now return to the nonlinear equation #0 99 Set fsino=o.53. Fu, pare Morton 189 Multiply through by 2 d6/dr, integrate, and apply the initial condition 6 = 8, 446jdt = 0. Then separate variables in the resulting equation to obtain a From this equation determine the angular velocity d/dt as a function of 6. [Note that the left member cannot be integrated in terms of elementary functions to obtain the exact solution fi) of the nonlinear differential equation. i 5.3 FREE, DAMPED MOTION We now consider the effect of the resistance of the medium upon the mass on the spring, Still assuming that no external forces are present, this is then the case of free, damped motion. Hence with the damping coefficient a> Oand F(e) = Ofor allt, the basic differential equation (5,7) reduces to ax dx mor tag tk 2? and a/m = 2b for convenience) we have (527) Dividing through by m and putting the diferential equation (5.27) in the form yw Bex 2 cree (528) ‘Observe that since a i positive, bis also positive, The auxiliary equation is Pa 2b $2220. (629) Using the quadratic formula we find that the roots of (529) are he Ne PE 530) 7 J 5 ‘Taree distinct cases occur, depending upon the nature of these roots, which in turn depends upon the sign of 8? = 22. Case 1. Damped, Oscillatory Motion. Were we consider the casein which b < 2, Which implies that_b* — 73 <0, Then the roots (5.30) ate the conjugate comples numbers ~8 + JF and the general solution of Equation (5.28) i thus sin Zr +5 cos VBE 0 (san where ¢, and ¢, are arbitrary constants. We may write this in the alternative form xm ce Meco FB Oh (532) where c= Sef Fel > O and @ is determined by the equations sin 9, 4| 190 APPLICATIONS OF SECONO-ORDER LINEAR DIFERENTIAL EQUATIONS ‘The right member of Equation (5.32) consists of two factors, co and cot Ji? Fr + 9) ‘The factor ce“ is called the damping factor, or time-oar ying amplitude, Since ¢ > 0, it {s positive; and since b> 0, it tends to 2e¢0 monotonically as t+ 0. In other Words, as time goes on this positive factor becomes smaller and smaller and eventually becomes negligible. The remaining factor, cosi./2 — Bt + 4), is, of course, of & periodic, cosellatory character; indeed it represents a simple harmonic motion. The product of these two factors, which is precisely the right member of Equation (5.32), therefore represents an oscillatory motion in which the oscillations become suocessively smaller and smaller. The oscillations are sid to be “damped out," and the motion is described as damped, oscillatory motion, Of course, the motion is no longer periodic, but the time interval between two successive (positive) maximum displacements stil referred toas the period. Tiss given by 2 yee ‘The graph of such a motion is shown in Figure SS, in which the damping factor ce"* and its negative are indicated by dashed curves. ‘The ratio of the amplitude at any time T to that at time Te cone period before T is the constant 2xb °(-zaH} ‘Thus the quantity 2x /22 — b?isthe decreas in the logarithm ofthe amplitude ce-* ‘over atime interval of one period. It scaled the logarithmic decrement Figure 5.552. Fee, onwpen motion 191 If we now return to the original notation of the diferentiat equation (5.27), we see from Equation (5.32) that in terms of the original constants ma, and &, the general solution of (5.27) is sawrem onl EE) os Since b< 2s equivalent to a/2n < /fm, we can say thatthe general soltion of (627) gen by (633) and tet amped, oscllatory motion oewars when @ <2, Kn. “The frequency of the osilations oo( E8) ws If damping were not present, a would equal zero and the natural Irequency of an ‘undamped system would be (1/2n)//m. Thus the frequency of the oscillations (5.34) in the damped oscillatory motion (5.33) is less than the natural frequency of the corresponding undamped system, Case 2, Critical Damping. ‘This is the case in which 6 = A, which implies that 6? — 22 = 0. The roots (530) are thus both equal tothe rel negative number ~6, and the general solution of Equation (528) s thus eae teane™ (6535) ‘The motion is no longer oscillatory; the damping is just great enough to prevent ‘oscillations. Any slight decrease in the amount of damping, however, will change the situation back to that of Case 1 and damped oscillatory motion will then occur. Case 2 then isa borderline case; the motion is sad to be critically damped. From Equation (5.35) we see that ate lima 2 = tim ‘Hence the mass tends to its equilibrium position as + oo. Depending upon the initial conditions present, the following possibilities can occur in this motion: 1. Themass neither passes through its equilibrium position nor atlains anextremum (maximum or minimum) displacement from equilibrium for ¢> 0. It simply approaches its equilibrium position monotonically as t > co. (See Figure 5.62} 2. The mass docs not pass through its equilibrium position for > 0, but its displacement from equilibrium attains a single extremum for = T, > 0. After this extreme displacement occurs, the mass tends to its equilibrium position monotonically 5 £00, (See Figure 5.6b,) 3. The mass passes through its equilibrium position once at ¢ = T; > O and then attains an extreme displacement at = T, > T,, following which it tends to its equilibrium position monotonically as ¢—+ oo (See Figure 5.6)192 APRCATIONS OF ECOND-ODER LINEAR DIFFERENTIAL EQUATIONS @ ies ee oo) at hoot © Figure 5.6 Case 3. Overeritical Damping. Finally, we consider here the case in which b > 2 which implies that b? — 2? > 0. Here the roots (5.30) are the distinct, real negative numbers and. reba ‘The general solution of (5.28) in this ease is se + 60 (6539 TThe damping is now so great that no oscillations can oceur. Further, we can no longer say that every decrease in the amount of damping will result in oscillations, as we could in Case 2, The motion here is said to be overcritically damped (or simply overdamped). Equation (5.36) shows us at once that the displacement x approaches 2er0 as ><. Asin Case 2 this approach tozero is monotonic for ¢ sufficiently large. Indecd, the three possible motions in Cases 2 and 3 are qualitatively the same, Thus the three motions illustrated in Figure 5 6can also serve to illustrate the three types of motion possible in Case 3 D Example 5.2 A 32-bit is atached to the fower end ofa col spring iapended from the ceiling The weight comes to cet in its equlbrim positon thereby stetching the spring 233 ree, oaMrED Motion 193 ‘The weight is then pulled down 6 in, below its equilibrium position and released at t= 0. No external forces are present; but the resistance of the medium in pounds is ‘numerically equal to 4(dx/di), where dxdt i the instantaneous velocity in feet per second. Determine the resulting motion of the weight on the spring. Formulation. This is a free, damped motion and Equation (5.27) applies, Since the 32-1b weight stretches the spring 2, Hooke’s law, F = ks, gives 32 = k(2} and 50 k= 16]bjft. Thus, since m= w/g = 32= 1 (slug), and the damping constant a= 4, Equation (527) becomes wx | dx Sb 16x = 7 arth g tomo (537) ‘The initial conditions are x(0}= 4, x) =0 638 Solution. The ausiliary equation of Equation (5.37)is P44r4 16-0, Usrootsare the conugatecomplex numbers ~2 & 2/3. Thus the general soltion of (S31) may be writen xe eM sin 274, 0082/7, 63) where e, and ¢, ate arbitrary constants. Dierentiatng (5.39) with respect to 1 we obtain Boe 2t\-2e ~ 2fseasin 2/314 BV ey — 2eoos 2/304 40) Applying the intial conditions (5.38) to Equations (5.39) and (5.40), we obtain 1 fe, — 20, =0. ruse = V)heg fad the solution em V3 sin far bcos 2/3 oe G ana irs Lena J #9 Lotus put this in the alternative form (5.32). We have BianrJirr beara [5 snare cos2 it] = Peor(2vir-F) Thus the solution (541) may be written *cos( 2/5 - 3) (52)194 APPUCATIONS OF SECOND-ORDER NEAR DIFFERENTIAL EQUATIONS. erpretation. Thisis a damped oscillatory motion (Case 1). The damping factor is (/3/3)e", the “period” is 2x/2./3 = V/3n/3, and’ the logarithmic decrement is /An/3. The graph ofthe solution{5.42}is shown in Figure 5.7, where the curves x = i/3/3)e"2 ate drawn dashed. D example 5.3, Determine the motion of the weight on the spring described in Example $2 if the resistance of the medium in pounds is numerically equal to 8{dx/dt} instead of 4(dx/at) {as stated there, all other circumstances being the same as stated in Example 52 Formulation. Onoe again Equation (5.27) applies, and exactly asin Example S2 we find that m = I (Slug) and k = 16 lb/ft, But now the damping has increased, and wo have a = 8. Thus Equation (5.27) now becomes @x | dx Ge at xno, 5.43) a ta see ‘The initial conditions x(0) = 3 x10 =0, snged from Example 52. (544) are, of course, unc! Solution, The auxiliary equation is now Ps 84 1600 and has the equal roots r —4. The general solution of Equation (543) is thus. 549) xm ley beste” Figure 5.7pawn motion 195, where ¢, and care arbitrary constants. Differentiating (5.45) with respect tof, we have ar a Applying the initial conditions (5.44) to Equations 5.45) and (5.46) we obtain (= ey — dayne (5.46) Thus e, 2 and the soltuion is wate % (647) Interpretation. The motion is critically damped. Using (5.47), we see that x = Oif, and only if = ~4. Thus x #0 for ¢ > 0 and the weight does not pass through its equilibrium position, Differentiating (5.47) one finds at once that ds/dt < 0 for all > 0. Thus the displacement of the weight from its equilibrium position isa decreas- ing function ofr for all: > 0. In other words, the weight starts to move back toward its equilibrium position at once and x+0 monotonically as 1+ co, The motion is therefore an example of possibility 1 described in the general discussion of Case 2 above. The graph of the solution (5.47) is shown as the solid curve in Figure 58, Figure 5.8 > Example 5.4 Determine the motion of the weight on the spring described in Example 5.2 if there sistance of the medium in pounds is numerically equal to }0(dx/dr) instead of 4(dx/d) (as stated there), all other circumstances being the same as stated in Example 52. Formulation. The only difference between this and the two previous examples i in the damping constant. In Example 52, a = 4:in Example 53,4 = 8; and now we have even greater damping, for here a = 10. As before m= 1 (sig) and k= 16 1b/f. The differential equation (527) thus becomes #x ode Sat 10S + 16x ~0. (S48) The initial conditions ($.44) or (5.38) stil hol.196 APPLICATIONS OF SECOND ORDER NEAR DIFFERENTIAL EQUATIONS Solution. ‘The auxiliary equation is now +104 16=0 and its roots are r = —2, 8, Thus the general solution of Equation (5.48) is xmge eee where ¢, and c, are arbitrary constants, Difleentiating this with respect to 10 obtain a a and applying the initial conditions (5.44), we find the following equations for the determination of ¢, and ey Deven ~ See"* ate ‘The solution of this system is ¢, —t5 thus the solution of the problem is xade hate (549) Interpretation, Clearly the motion deseribed by Equation (5.49) is an example of the overdamped case (Case 3). Qualitatively the motion is the same as that of the solution (5.47) of Example $3, Here, however, duet the increased damping, the weight returns to its equilibrium position at a slower rate. The graph of (5.49) is shown as the dashed curve in Figure 58. Note that in each of Examples 52, 5.3, and S4, all circumstances (the weight, the spring, andthe initial conditions) were exactly the same, except for the damping. In Example 52, the damping constant a = 4 and the resulting motion was the damped oscillatory metion shown in Figure §.7. In Example 5.3 the damping was increased to such an extent (a = 8) that oscillations no longer occurred, the motion being shown by the solid curve of Figure 5.8. Finally in Example $44 the amping was further increased (a = 10) and the resulting motion, indicated by the dashed curve of Figure 58, was similar to but slower than that of Example 53. |. An 8.1b weight is attached to the lower end of a coil spring suspended from the ceiling and comes to rest in its equilibrium position, thereby stretching the spring. (0.4 ft The weight is then pulled down 6 in. below its equilibrium position and released at = 0. The resistance of the medium in pounds is numerically equal to 2ds ld), where dx/de is the instantaneous velocity in feet per second, (a) Set up the differential equation for the motion and list the inital conditions (b) Solve the initil-value problem set up in part (a) to determine the displacement of the weight as a function of the time. (6) _Expressthe solution found instep (b)in the alternative form (5,32) of the text. (4) What i the so-called “period” of the motion? (@) Graph the displacement asa function of the time.sa. ener, pawrep worion 197 A 16-tb weight is placed upon the lower end of a coil spring suspended from the ceiling and comes to rest in its equilibrium position, thereby stretching the spring, Sin. At time t = O the weight is then struck so as to set it into motion with an initial velocity of 2 fy/sec, directed downward. The medium offers a resistance in pounds numerically equal to 6(dx/dt), where dx/dt is the instantaneous velocity in feet per second. Determine the resulting displacement of the weight as a func~ ton of time and graph this displacement. ‘An 8:1b weightis attached to the lower end ofa col spring suspended from a fixed ‘support. The weight comes to rest in its equilibrium position, thereby stretching the spring6 in. The weight is then pulled down9 in, below itsequilibrium position and released at: = 0, The medium offersa resistance in pounds numerically equal to Aféx/de, where de/dr is the instantaneous velocity in feet per second. Determine the displacement of the weight asa function ofthe time and graph this, displacement A 16410 weight is attached to the lower end of col spring suspended from a fixed support, The weight comes to rest in its equilibrium position, thereby stretching thespring 6 in. The weightis then pulled down 3 in below its equilibrium position and released at¢ = 0. The medium offersa resistancein pounds numerically equal to 10(dx/dr), where dx/dt is the instantaneous velocity in feet per second. Determine the displacement of the weight as a function of the time and graph this displacement. AA spring is such that a force of 201 would stretch it 6 in. The spring hangs vertically and a 4-15 weight is attached to theend ofit. After this 4-1b weight comes to rest in its equilibrium position itis pulled dowa 8 in. below this position and then released at ¢ = 0. The medium offers a resistance in pounds numerically equal to 2(dxjdr), where dx/de isthe instantaneous velocity in feet per second, (a) Determine the displacement of the weight as a function of the time and express this displacement in the alternative form (5.32} of the text. {b) Find the so-called “period” and determine the logarithmic decrement (©) At what time does the weight first pass through its equilibriver position? ‘Adstb weight is hung upon the lower endo a coilpring hanging vertically rom a fred support. The wsigh comes to retin Hs equim postion, they Steching he spring in. Thewsight is then pulled own acerain distance below thisequirium postion and released at ~ 0 The mediimoflers a eistanes i pounds numerical equal todd where a> O and duds the instantaneous Nelo nfet per second. Show that themotio wosditoryi'a< V3 cally damped it a= 9/3, and overdamped it a> 3. ‘A 4b weight is attached go the lower end of a coil spring that hangs vertically from a fixed support, The weight comes to rest nits equilibrium position, thereby stretching the spring 6in. The weight is then pulled down 3 in. below this equilibrium position and feleased at 1 ~ 0, The medium offers a resistance in ‘pounds numerically equal to a(dx/dt), where a > and dx/dt isthe instantaneous velocity in feet per second, (@) Determine the value of a such that the resulting motion would be critically Oand dxaeis the instantaneous velocity in feet per second {a) Determine the smallest value of the damping coefficient a for which the motion is not oscillatory (b) Using the value ofa found in part (a) find theisplacement of the weight asa function of the time (©) Show that the weight attains a single extreme displacement from its equilibrium position at time ¢ = ab, determine this extreme displacement, and show that the weight then tends monotonically co its equilibrium position as 0» cc (4) Graph the displacement found in step (by ‘A 32-1p weight is attached to the lower end of a col spring suspended from the cling. After the weight comes to rest in its equilibrium position, its then pulled ‘down a certain distance below this position and released at ¢ = 0. Ifthe medium offered no resistance, the natural frequency of the resulting undamped motion would he 4/x cycles per second, However, the medium does offer a resistance in pounds numerically equal to a(dx/d), where @ > Oand d/dtis the instantaneous velocity in feet per second; and the frequency ofthe resulting damped oscillatory motion is only half as great as the natural frequency. {2}, Determine the spring constant k (b) Find the value of the damping coefficient a ‘The differential equation for the vertical motion of a mass m on a col spring of spring constant & in a medium in which the damping is proportional to the instantaneous velocity is given by Equation (5.27). In the case of damped ‘oscillatory motion the solution of this equation is piven by {5.33}, Show that the displacement x so defined attains an extremum (maximum or minimum) at the times (0 = 0, 1,2... given by [ne tan| (-s5,) a +} where 8 I i ‘The dferemal equation or the vere motion of uit mss on ei cai sprog in erin mam &, aesa rorceo wonon 199 where b > 0. The inital displacement of the massis A feet and its initial velocity B fet per second, (@) Show that the motion is critically damped and that the displacement is piven by x= (A+ Bet Abe. (b) Wf Aand Bare such that eee ea “asa ™ Era are both negative, show that the mass approaches its equilibrium position monotonically as ¢-+ co without either passing through this equilibrium position or attaining an extreme displacement from it for t > 0. (© If A and B are such that —4/(B + 4b) is negative but B/o(B + Ab) is postive, show that the mass does not pass through its equilibrium position, for ¢ > 0, that its displacement from this position attains a single extremom at c= B/b(B + Ab), and that thereafler the mass tends to its equilibrium position monotonically as ¢~+ co (4) I Aand Bare such that — 4/(B + Ab) is positive, show that the mass passes through its equilibrium position at ¢= —A/(B + Ab), attains an extreme displacement at = B/b(B + Ab), and thereafter tends to its equilibrium position monotonically as ¢ > co 5.4 FORCED MOTION ‘We now consider an important special case of forced motion. That is, we not only consider the effect of damping upon the mass on the spring but also the effect upon it of a periodic external impressed force F defined by F(t) = F, cos.at for allt > O,where Fy and w are constants. Then the basi differential equation (5.7) becomes we alk hea Renee 59 Dividing through by m and letting this takes the more convenient form #2 4% ae SF mE 4 Px £, cos ot sn) ‘We shall assume thatthe positive damping constant a is small enough so that the damping i less than critical. Tn other words we assume that b< 2. Hence by Equation (532) the complementary function of Equation (5.81) can be writen -* cos Pa DK + oh. (552) x (VP200 arrucanions oF secovn-o8D tn LINEAR DIFFERENTIAL EQUATIONS - We shall now find a particular integral of (551) by the method of undetermined coefficients, We let , Acos at + Bsin at (553) Then Fe = ~OA sin at + @B cos ot, & 7 7 og ‘A cos cot ~ 0B sin ot Substituting into Equation 5.51), we have [-2hod + (2 - 0)BJsin o + [2 ~ 0) + 2boBJoos ot = E, 605 wt Thus, we have the following two equations from which to determine A and B: ~2bed +02 — 0)B=0, (2 = 0)A + 2beB = By. Solving these, we obtain (5.54) Substituting these values into Equation (5.53), we obtain a particular integral in the form Ey ae ea agra HU Peas ot + 2ho sino) ‘We now put this in the alternative “phase angle” form; we write G2 = @}¢08 of + 2be sin ot - Hae? | cos at 7 a JG? oa + B70 i Te = Ve GAP ¥ AB7H [o0s ot cos 0 + sin cot sin 6] JER OF FBS costur ~ 0), where Be Se VR oF aot ba eeeete : sin = a : MB OP AF 0? cos = (555)sa rorceo moron 201 ‘Thus the particular integral appears in the form costo — 8), (556) where @is determined from Equations (5.55). Thus, using (5.52) and (5.56) the general solution of Equation (5.51) is cet apse Meo FP =P + d+ Ey spelen ~ (557) ‘Observe that this solution is the sum of two terms. The first. term, ce cost /7* — Br + ¢), represents the damped oscillation that would be the entire ‘motion of the system ifthe external force F cos xt were not present. The second term, & ee Vero? Ve ony + 4) Which results from the presence of the external force, represents a simple harmonic motion of period 2n/a. Because of the damping factor ce™™ the contribution of the first tetm will become smaller and smaller as time goes on and will eventually become negligible. The first term is thus called the transient term. The second term, however, being a cosine term of constant amplitude, continues to contribute to the motion in a periodic, oscillatory manner. Eventually, the transient ferm having become relatively small, the entre motion will consist essentially of that given by this second term. This second term is thus called the steady-state term. D> Example 5.5 ‘A 16-15 weight is attached to the lower end of a coil spring suspended from the ceiling, the spring constant of the spring being 10 Ib/f. The weight comes to rest in its equilibrium position, Beginning at t = @ an external force given by F(t) = 5cos 2s applied to the system. Determine the resulting motion if the damping force in pounds is ‘numerically equal to 2(dx/dr), where dxdt isthe instantaneous velocity in feet per second. Formulation, The basic differential equation for the motion is Px dx mG ta + kx = Flo. 6.58) Herem = wig =$8= 4 (loga=2.k becomes 10,and Fe) = 5 cos 2 Thus Equation (5.58) Lax dx P5448 roc tes «59202. sucations oF seconD-onneR uNE«® DM ‘The initial conditions are x40) <0) (5:60) Solution. ‘The auxiliary equation of the homogeneous equation corresponding 10 (5.59}isr? + 4r +20 = 0; 118 roots are —? + 4 Thus the complementary function of Equation (5.59) is xe = Me, sin 4+ €3 608 40), where ¢, and c, arearbitrary constants, Using the method of undetermined coefficients {o obtain a particular integral, we let xp = Acos 2r + B sin 2 Upon differentiating and substituting into (5.59), we find the following equations for the determination of 4 and B. 8A +168 =0, Solving these, we find ‘Thus a particular integral is p= feos 21+ fin 2¢ and the general solution of (5.59) "(e,sin 4 +; cos At) +Yeos 2+ 4sin2 (5.61) ing (5.61) with respect tof, we obtain (= 2ey — dea)sin de + (—2ey + Aeq)oos 4e] — sin-2¢ + Y eos 20. (5.62) Applying the initial conditions (5.60) to Equations (5.61) and (5.62), we see that et hed, de, 2, + $20, From these equations we find that “ ast Hence the solution i x= 27 "(—] sin 41 — $ cos dt) + $ cos 2r + 4 sin 2. (5.63) Let ws write his inthe “phase angle form, We have ist 3sin 4+ 4608 dt = SU sin +4 0084) = 5 coslde~ 9, where cos sin d= 2 (564)EEE eee eee eee eee 4 toxcen motion 203 Also, we have a \ : 20s 21+ sin = /5(-2ecos2e +t. sin 21) = /Beosat— 0), goon asin tt) = Bont where 1 sino=t, (865) B Thus we may write the solution (5.6) as se ve = cont 4) + Beast, (566) where ¢ and @ are determined by Equations (5.64) and (5.65) respectively. We find that 064 (rad) and @ ~ 0.46 (rad). Thus the solution (5.66) i given approximately by X= = 0.636" cos(dt ~ 0.64) + 0.56 cos(2r ~ 0.46 Interpretation. The term HES cost 6) 5 0630s 069 isthe transient term, representing a damped oscillatory motion. It becomes negligiblein aa short time; for example, for t > 3, its numerical value is less than 0.002. Its graph is shown in Figure 59a, The term /s SF cos(2r ~ 6) = 0.86 cose ~ 0.46) is the steady-state term, representing a simple harmonic motion of amplitude fs WP 3056 and period x. Its graph appears in Figure 595. The graph in Figure 59cis that of the ‘complete solution (5.66). It is clear from this thatthe effect of the transient term soon becomes negligible, and that after a short time the contribution of the steady-state term is essentially all that remains. Exercises L.A 61b weight is attached to the lower end of a coil spring suspended from the ceiling, the spring constant of the spring being 27 lbyfl. The weight comes to rest inits equilibrium position, and beginning at 1 = O an external force given by F(t) = 12 cos 20¢ is applied to the system. Determine the resulting displacement 425. a function of the time, assuming damping is negligible. 2. A16-1b weight is attached 10 the lower end of a col spring suspended fram the ceiling. The weight comes to rest in its equilibrium position, thereby stretching the spring0.4 @. Then, beginningatr = 0,anexternal force given by Fl) = 40.cos 167204. sreucarions oF sccoND-oRDUR LINEAR DIFTERENTIAL EQUATIONS Figure 5.9 is applied to the system. The medium offers a resistance in pounds purnerically equal to 4(dx/dt), where dxdt is the instantaneous velocity in feet per second. (@)_ Find the displacement of the weight asa function of the time E (6) Graph separately the transient and steady-state terms ofthe motion found ih step(a)and then use the curves so obtained to graph the entire displacement itselt, 3. A 10-Ib weight is hung on the lower end of a coil spring suspended from the ceiling, the spring constant of the spring being 20 Ibyft. The weight comes to rest in its equilibrium position, and beginning at r = 0 an external force given by F(Q)= 100s &t is applied to the system. The medium offers a resistance in4 torcep moron 205, pounds numerically equal to S(ds/d), where dx/dtis the instantaneous velocity in feet per second, Find the displagement of the weight as a function of the time. ‘A.4.1b weight is bung on the lower end of a col spring suspended from a beam. ‘The weight comes to rest in its equilibrium position, thereby stretching the spring 3 in, The weightis then pulled down 6 in. below this position and released at = 0, AC this instant an external force given by F(c} = 13 sin 4cis applied to the system, The resistance of the medium in pounds is numerically equal to twice the instantaneous velocity, measured in fet per second. (a) Find the displacement of the weight asa function of the time. {b) Observe that the displacement is the sum of a transient term and a steady. state terra, and find the amplitude of the steady-state term. ‘A 6-1b weight is hung on the lower end of a coil spring suspended from the ceiling. ‘The weight comes to rest in its equilibrium position, thereby stretching the spring 4in, Then beginning at =O an external force given by F(t) = 27 sin 4t — 3 c0s 4r is applied to the system. If the medium offers a resistance in pounds numerically equal to three times the instantaneous velocity, measured in eet per second, find the displacement as a function of the time. Aceriain col spring having spring constant 10 Ibis suspended from the ceiling. A 32-6 weight i atached to the lower end of the spring and comes to rest in its equilibrium position: Beginning at c= 0 an external force given by FQ) = sin +4 sin 2c + sin 3 is applied to the system. The medium offers resist ance in pounds numerically equal o twice the instantaneous velocity, measbred in fet per second. Find the displacement of the weight as a function of the time, using Chapter 4, Theorem $.10 to obtain the steady-state term, ‘coil spring having spring constant 20 Ibyftis suspended from the ceiling. A 32ib ‘weight is attached to the lower end of the spring and comes to rest in its equilibrium position. Beginning at ¢=0 an external force given by Fl) = 40 cos 2 is applied to the system. This force then remains in effect until ¢ = n, at which instant it ceases to be applied. For > x, no external forces are present ‘The medium offers a resistance in pounds numerically equal to 4(dx/dt}, where dx/deis the instantaneous velocity in feet per second. Find the displacement of the weight asa function of the time for all: > 0 Consider the basiccifereatia uation (5.7fo the tion ofa mass m vibrating up and down on coll spring Suspended vericaly from afted suppor: and Suppose the external impressed force F isthe periodic function éened by F()= F sn for als > 0, where Fy and care constants Show tet nth ae the seady stat er inthe solution of Equation (31h may be writen & Pers en where b= a/2m, 2? = k/m, E, = F,/m, and @ is determined from Equations (555). ‘A 32-1b weight is attached to the lower end of a coil spring suspended vertically from a fixed support and comes to rest in its equilibrium position, thereby streiching the spring 6 in. Beginning at r = 0 an external force given by Ful) = 1S cos 71 is applied to the system. Assume that the damping is negligible.206 serucamioNs oF sccOND-oRDGR LINEAR DIFERESTIAL EQUATIONS {@)_ Find the displacement ofthe weight asa function of the time (b) Show that ehis displacement may be expressed as x = AUt)sin(151/2), where -A() = 2 sin 4. The function () may be regarded asthe “slowly varying” amplitude of the more rapid oscillation sin({5t/2), Wien a phenomenon involving such fluctuations in maximum amplitude takes place in seoustical applications beats are sai to occur. (¢) Carefully graph the slowly varying amplitude A(¢) = 2 sif(c/2) and its negative —A(?) and then use these “bounding curves” 10 graph the displacement x = A(Qsin( 1502) 10. A 16-1b weight is attached to the lower end of a coil spring that is suspended vertically from a support and for which the spring constant & is 10 1b/ft. The weight comes to rest in its equilibrium position and is then pulled down 6 in. below this position and released att = 0. At this instant the support of the spring ‘begins a vertical oscillation such that its distance from its initial postion is given bby sin 2 for 2 0, The resistance of the medium in pounds is numerically equal to 2(dx/dr), where dxdt is the instantaneous velocity ofthe moving weight in feet per second. (a) Show that the differential equation for the displacement of the weight from its equilibrium position is bade dy GE a Me 25, where y= $sin24, and hence that this differential equation may be written ax, yds Sts 4 20x = 10sin 2 (b) Solve the differential equation obtained in step fa), apply the relevant intiat ‘conditions, and thus obtain the displacement x as a function of time. 5.5 RESONANCE PHENOMENA We now consider the amplitude of the steady-state vibration that results from the periodic external force defined for all s by F(t) =F, cos x, where we assume that F, > 0. For fixed b, 2, and E, we see from Equation (5.56) that this isthe function of w defined by } E {08 aaa ae (967) If w= 0, the force F(t) is the constant F and the amplitude fl) has the value E,(22 > 0. Also, a8 @ + 20, we see from (5.67) that f(a} +0. Let us consider the function f for 0 < @ < co, Caleulating the derivative J (io) we find that this deriva- __ tive eauals zero only if 4o(2b — U2 — 0] and hence only if « = 0 oF w= fF = 25% If 2 < 26%, /7— 2? is a complex number. Hence in this case f has no extremum for 0 << 25% Then the function f has a relative maximum at eo, = /#2— 265, and this maximum values given by Ey E, VERT RED BIE F Slo)= ‘When the frequency of the forcing function F, cos ar is such that « = a, then the forcing function is said to be in resonance with the system, In other words, the forcing fanetion defined by F, cos ct isin resonance with the system when c assumes the value (a, at which fw) is # maximum, The value o,/2z i called the resonance frequency of the system. Note carefully that resonance ean occur only ifs? > 262. Since then 22 > 6% the damping must be less than critical in such a case We now return to the orignal notation of Equation 5.50. In terms of the quantities, im. ak, and F, of that equation, the function f is given by 668) In this original notation the resonance lrequency is L fk 4 5.69) Bem 2 ee Since the frequency ofthe corresponding free, damped osilation is 1 koa ’ ian ae we see that the resonance frequency is less than that of the corresponding free, damped oscillation, ‘The graph of f(o)isealled the resonance curve of the system. For a given system with fixed m, , and F,, there is a resonance curve corresponding to each value of the damping coefficient a > 0, Let us choose m = k = F, = 1, for example, and graph the resgnance curves corresponding to certain selected values of a. In this case we have Seo) Vines or land the resonance frequency is given by (1/2n)/T—a%/2. The graphs appear in Figure $10, . _ ‘Observe that resonance occursin thiseasconly ita < V3. As adecreasesfeorn /Zto 0, the value «, at which resonance occurs increases from 0 to 1 and the corresponding, maximum value of f(«) becomes larger and larger. In the limiting case a= 0, the maximors fas disappeared and an infinite discontinuity occurs at w = J, In this case ‘our solution actually breaks down, for then 1 Ja-0F 1 Slo)208 —reruications oF secoND-ORDEK LINEAR DIFFERENTIAL EQUATIONS Figure 5.10 and (1) is undefined. This limiting case is an example of undamped resonance, phenomenon that we shall now investigate. ‘Undamped resonance occurs whien there is no damping and the frequency of the Jmpressed fore is qual to the natural frequency of the system. Since inthis case a = 0 and the frequency a/2n equals the natural frequency (1/2n),/k/m, the differential ‘equation (5.0) reduces 10 ax fk Gee bem foo [Ee fey heon on fis 6% where E, = Fi/m. Since the complementary function of Equation (5:70) is is swersin Ars epeos Ee om ace va ‘we cannot assume a particular integral of the form55 eesonanct pHenomens 209 i Differentiating this twice and substituting into Equation (6.70), we find that E, fm 1-8 FE me a0 ‘Thos the particular integral of Equation (5.70) resulting from the forcing function ooh Fw fE Expressing the complementary function (5.71) in the equivalent “phase-angle" form, we see that the general solution of Equation (5.70) is given by m(frns)o Sion ‘The motion defined by (5.72) is thus the sum of a periodic term and an oscillatory term ‘whose amplitude (E,/2)/m/ke inereases with t The graph of the function defined by this latter term, Es I sin fe Basin [Ee appears in Figure 5.1. As r increases, this term clearly dominates the entire motion. ‘One might argue that Equation (5.72) informs us that as ¢—+ oo the oscillations will become infinite, However, common sense intervenes and convinces us that before this ‘exciting phenomenon ean occur the system wil break down and then Equation (5.72) will no longer apply! Rather we must assume 67210 APPLICATIONS OF SECOND-ORDER LINEAR DIFFERENTIAL EQUATIONS > Example 5.6 ‘A 64-1b weight is attached to the lower end of acoil spring suspended from the ceiling, the spring constant being 18 lb/ft The weight comes to rest in is equilibrium position {tis then pulled down 6 in. below its equilibrium position and released at ¢ = 0. At this instant an external force given by F(t) = 3 c0s wt is applied to the system, 1. Assuming the damping force in pounds is numerically equal to 4(ds/di), where dx/dt is the instantaneous velocity in feet per second, determine the resonance Frequency of the resulting motion, 2 Assuming there is no damping, determine the value of w that gives tise to undamped resonance, Solution. Since m = w/a = $$ = 2(slugs), ferential equation is 18, and F(r) = 3 cos or, the dif ax dx 255 + ah + 18x = 308 0%, where ais the damping coefficient. In Part 1, a = 4 and so in this case the differential equation reduces to Ox | dx Gat AG 4 Ie = 300s 0% Here we are not asked to solve the differential equation but merely to determine the resonance frequency. Using formula (5.69) we find that this is » -4 ala) "as BaV 22a) wag VT XO ores Thus resonance oceurs when w = /7 = 265, In Part 2, = 0/and so the differential equation reduces 10 ax Girt 9x = feos ot 1573) Undamped resonance occurs when the frequency «o/2: of the impressed force is equal to the natural frequency, The complementary function of Fquation (5.73) is sin 3t + €3 00s 3, and from this we see that the natural frequency is 3/2. Thus a = 3 gives rise to uundamped resonance and the differential equation (5.73) in this case becomes SF 4 9 toon 64) ‘The initial conditions are x(0) = 4, x'(0) = 0. The reader should show that the solution of Equation (5.74) satislying these conditions is cos 3¢ + $e sin 346 succtmc circum exomtens 211 Exercises 1, A 12-1 weight is attached to the lower end of a coil spring suspended from the ceiling, The weight comes to retin its equilibrium position thereby stretching the spring6 in, Beginning att = Oan external force given by F(e) = 200s «is applied to the system. @)_ I che damping forceiin poundsis numerically equal to 3dx/de), where dx/deis, the instantaneous velocity in feet per second, determine the resonance frequency of the resulting motion and find the displacement as a function of the time when the forcing function isin resonance with the system, (©) Assuming there is no! damping, determine the value of «w that gives rise 10 ‘undamped resonance and find the displacement as a function of the time in this case ‘A 20-1 weight is attached to the lower end of a coil spring suspended from the ceiling, The weight comes to resin its equilibrium position, thereby stretching the Spring 6 in. Various external frees ofthe form F(0) = e08 ot are applied to the system and itisfound that the resonance frequency is05eycles/se. Assuming that the reitance of the medium in pounds is numerially equal to o(@x/d), where dxfde is the instantaneous velocity in fet per second, determine the damping cotfcient a 3. The diferential equation forthe motion of a unit mass on 8 certain coil spring under the ation ofan externa fore ofthe frm Fd) = 30605 wt is @x dx #5 4% 4 rae = 3008, Oa pan where a 2 O's the damping coeficent (a) Graph the resonance curves of the system for a= 0, 2,4, 6 and 4/3. (b) a= 4, find the reconance frequency and determine the amplitude of the steady-siate vibration when the foreing function is in resonance with the (©) Proceed as in part (b)if @ 5.6 ELECTRIC CIRCUIT PROBLEMS In this section we consider the application of differential equations to series circuits containing (1) an electromotive foree, and (2) resistors, inductors, and capacitors. We assume thatthe reader is somewhat familiar with these items and s0 we shall avoid an extensive discussion, Let us simply recall that the electromotive force (for example, a battery or generator) produces flow of current in aclosed circuit and that this current produces a so-called voltage drop acroxs each resistor, inductor, and capacitor. Further, the following three laws concerning the voltage drops across these various elementsare known to hold 1. The voltage drop across a resistor is given by Ex = Ri, (575) where R isa constant of proportionality called the resistance, and iis the current. | i212 APPLICATIONS OF SECOND-ORDER LINEAR DIFFEENTIAL EQUATIONS 2. The voltage drop across an inductor is given by a RnLd. (5:76) Where L is a constant of proportionality called the inductance, and Vagain denotes the ccorrent 3, The voltage drop across a capacitor is given by 1 Bem Go (77) ‘where C is a constant of proportionality called the capacitance and q is the instantaneous charge on the capacitor. Since i = dg/dt, this i often written as Beng [ia ‘The units in common use are listed in Table 5.1 TABLE 5.1 Quantity and symbol Unit ‘emforvolageE volt(V) current | ampere charge g coulomb resistance R ‘obm (9) inductance L henry (H) capacitance C farad ‘The fundamental law in the study of electric circuits is the following: Kirchhoff’s Voltage Law (Form 1). Thealgebraic sum of the instantaneous voltage drops around a close circuit ina specific direction is zero, Since voltage drops across resistors, inductors, and capacitors have the opposite sign from voltages arising Irom electromotive forces, we may state this lew in the following alternative form: Kirchhoft’s Voltage Law (Form 2), The sum of the voltage drops across resistors, inductors, and capacitors is equal fo the total electromotive force in a closed circuit. ‘We now consider the circuit shown in Figure 5.12, R| | | | | 56 ttcrnc circurt rrowums 213 Here and in ater diagrams the following conventional symbols are employed —O— £ Elestromotve fore (battery or generator) Awe & Resistor osama Inductor Ec Copssitor Let us apply Kirchhof’s law to the circuit of Figure 5.12. Letting E denote the Example 5.7 A circuit has in series an electromotive force given by E = 100 sin 401 V,a resistor of, 10.Q and an inductor of 0.5 H. Ifthe initial current is 0, find the current at time ¢ > 0. Formulation. The circuit diagram is shown in Figure 5.13. Let idenote the current in amperes at times. The total electromotive forces 100 sin 40¢ V. Using the laws | and 2, we find that the voliage drops are as follows: 1, Across the resistor: By = Ri = 10% didi 2, Across the inductor: = Le = 55 Applying Kirchhoff’s law, have the differential equation bai 39 710i = 100 sin 40, ai $+ 208 = 200 sin or (582) Since the initial current is 0, the initial condition is HO) 20. 683) Solution. Equation (5.82) isa first-order linear equation, An integrating factor is R{ t } | i \ ' / | | | } | } i | } \ } { \ s6 exrcmic cucu rons 215, Multiplying (5.82) by this, we obtain & (e2%) = 200e™ sin 404 Integrating and simplifying. we find f= 2jsin 408 — 2.608 401) + Cem, Applying the condition (5.83), i= 0 when ¢ = 0, we find C = sin 40¢ — 2 €08 408) + de-2*, the trigonometric terms in a “phase-angle” form, we have ‘Thus the solution is = 2/5(Fysinsor— 5, cos sr) 4 2 or . es i= 2S sint4oe + 4) + 46*, 584 where ¢ is determined by the equation We find ¢ ~ ~ 1.11 rad, and hence the current i given approximately by 47 sin(40e ~ 1.11) + 40" 2 Interpretation, The curent clearly expresed asthe stm ofa sinusoidal erm and an exponential. The exponential becomes so very small ina short ie that it eect x soon practically neligibe isthe tramsen term, Thus after a short time, essentially all that remains ste sinusoidal term itis the steady-stare current. Observe that its period r/20 i the same as that ofthe electromotive force. However, the phase angie = —L1 indicates thatthe electromotive fore leads the steady-state curent by approximately dy (11) The graph of the curren asa function of time appears in Figure 516216 srpucations oF secono.oRDUR LNcAR DUFERENTIAL EQUATIONS D> Example 5.8 AA circuit has in series an electromotive force given by E = 100 sin 60: V.a resistor of, 2.0, an inductor of 0.1 H, and a capacitor of zby farads. (See Figure 5.15) If the initial ‘current and the initial charge on the capacitor are both zero, find the charge on the capacitor at any time t > 0. Formulation 1, by dicectly applying Kirchhof’s law: Let i denote the current and q the charge on the capacitor at time ¢ The total electromotive force is 100 sin 60¢ (volts). Using the voltage drop laws 1, 2, and 3 we find that the voltage drops. are as follows: 1. Across the resistor: Ey = Ri = 2i aid a i0d 2. Across the inductor: Ey, 3. Across the capacitor: 260. ot Now applying Kirchhofl’s law we have at once: Lai Thi + 21+ 2604 ~ 100 sin 6 Since = da/a, this reduces to aq oar 2424 200;- sna oss Formulation 2, applying Equation (5.80) for the charge: Wehave l= vy C= xhs, £ = 100 sin 60¢. Substituting these values directly into Equation (S80) we ‘obtain Equation (585) at once. “Multiplying Equation (585) through by 10, we consider the differential equation in the form #e £4 208 209 «9 R= 2ohms G bcm gtarte Figure 5.158 eutcraic cmcurt proses 217° Since the charge q is initially zero, we have as a first initial condition 40) =0. (587) Since the current is also initially zero and i= dg/dt, we take the second initial condition in the form 4) =0. (588) Solution. The homogeneous equation corresponding to (S86) has the auniliary ‘equation 74 20r +2600 =0. ‘The roots of this equation are —10 + 50i and so the complementary function of Equation (5:86) is 04, sin $0 + C,c0s $0, Employing the method of undetermined coeffcien’s to find a particular integral of (686), we write 4, = A sin 608 + B 60s 604. Differentiating twice and substituting into Equation (5.86) we find that and B= —3, and so the general solution of Equation (5,86) q =e !%(C; sin 50r + C, cos 50¢) — #f sin 60¢ — 3 cos 60r. (5.89) Differentiating (5:89), we obtain 4 a '[(—10C, — SOC, )sin $0r + (SOC, ~ LOC, )eos SOF] — 4892 cos 60r + 4422 sin 60%. (5.90) Applying condition (587) to Equation (5.89) and condition (5.88) to Equation (5.90), we have — C,-B=0 and 506, ~ 10, — 1892 =0. From these equations, we find that C<2f and G=2 ‘Thus the solution of the problem is bert (sin 0+ 5 50 — 3 in r+ 620060 6/5 oF 1% cot50e— 6)— 515 eostsor— 0, a2 6218 armucanons oF secono-onoen LINEAR DIFFERENTIAL EQUATIONS where cos $= 5//61, sind = 6/,/61 and cos @ = 6/./61, sin 0 = 5//61. From these equations we determine @ © 0388 (radians) and 6'~ 0.69 (radians). Thus our solutions given approximately by q = OTe" cos(S0t ~ 0.88) ~ 0164 cox(60e — 0.69). Interpretation. The fist term in the above solution clearly becomes negligible after 1 relatively short time; it isthe transient term, After a sufficient time essentially all that remains is the periodic second term; this is the sceady-state term. The graphs of these wo components and that of their sum (the complete solution) are shown in Figure 5.16. % 0.50 a Ost om f\ - os -00 Figure 5.16| | s& eurctec circum rrowuns 219 Brercises 1. circuit has in series constant electromotive force of 40 V, a resistor of 10 9 and aan inductor of 0.2 H. Ifthe initial current is 0, ind the current at time 2 > 0 2. Solve Exertise 1 if the electromotive force is given by E(t) = 150 cos 200r V instead of the constant electromotive force given in that problem, 3. A circuit has in series a constant electromotive force of 100, a resistor of 10.2, and a capacitor of 2 x 10~* farads, The switch is closed at time ¢ = 0, and the chargé on the capacitor at this instant is zero. Find the charge and the curent at time > 0 4. A-circuithasin series an electromotive orce given by E(e} = $ sin 100¢ V, a resistor of 109, an inductor of 0.05 H, and a capacitor of 2 x 10°* farads. Ifthe initial ccurrent and the inital charge on the capacitor are both zero, find the charge on the ‘capacitor at any time t > 0 5. A cireuit has in series an electromotive force given by E(t) = 100 sin 2001 V, 2 resistor of 40.0, an inductor of 0.25 H, and a capacitor of ¢ x 10 farads. If the initial current is zero, and the initial charge on the capacitor is 0.01 coulombs, find the current at any time t > 0. 6 Acircuithasin series anclectromotive force given by Et) = 200€" 1 a resistor of 800, an inductor of 021, and a capacitor of 5 x 10-* farads. If the inital current and the initial charge onthe capacitor are ero,find thecurrent at any time 150. 17. A circuit has in series resistor RQ, and inductor of LH, and a capacitor of C farads. The initial current is zero and the initial charge on the capacitor is Op coulomts. (@)_ Show thatthe charge and the current are damped oscillatory functions of time if and only i R< 2V/L/C, and find the expressions for the charge andthe curren in this case. (b) HER > 2/E/C, discuss the nature ofthe charge andthe current as functions of time 8 Acircuit has in series an electromotive force given by E(t) of RQ, an inductor of LH, and a capacitor of C farads. (@) Show that the steady-state current is osin wt V,a resistor 1-8 (Gino where X= Lo ~ 1/Cw and Z = /X¥ RE. The quantity X is called the reactance of the circuit and Z is called the impedance. (©) Using the result of part) show that the steady-state current may be written sin(on — 6), where ¢ is determined by the equations cosge8, sing =%220 APRLUCATIONS OF SECOND-OROER LINEAR OMFERENTIAL EQUATIONS @ «@ “This show thatthe steady-state current stains its maximum asolte value By/Z at times + 9/, where -3{2 al (11,2300, ae the times a which the letromotive force attains is maximum absolute value £y. Show that the amplitude ofthe steady-state current is a maxim when 1 jie For this valu of electrical resonance is sid to occ IER = 20, = 4, C= 10+, and fy = 10, the vue of that gives ise to electrical resonance and determine the amplitude of the steady-state current in this case,6a Series Solutions of Linear Differential Equations {In Chapter 4 we learned that certain types of higher-order linear differential equations (for example, those with constant coefficients) have solutions that can be expressed as finite linear combinations of known elementary functions. In general, however, higher- ‘order linear equations have no solutions that can be expressed in such a simple manner “Thus we must seek other means of expression forthe solutions ofthese equations. One such means of expression is furnished by infinite series representations, and the present chapter is devoted to methods of obtaining solutions in infinite series form. POWER SERIES SOLUTIONS ABOUT AN ORDINARY POINT ‘A. Basic Concepts and Results Consider the second-order homogeneous lina diferential equation #y sa ante) $5 + 4400) 5% + an(0)9 = 0, 61) and suppose that this equation has no solution that is expressible as a finite linear combination of known elementary fonctions Le wsassome, however, that it doeshave a solution that canbe expressed inthe form of an infinite series. Specifically, we assume that it has a solution expressible the form cot ede mo) + ee a += Ye 20% (62) where cox, 22.1 €"6 constants, An expression of the form (6.2) i called a power series in x ~ xp. We have thus assumed that the differential equation (6.1) has a so called power series solution of the form (6.2). Assuming that this assumption is valid, we 221222 \HRIES SOLUTIONS OF LINEAR OMFERENTIAL EQUATIONS ‘can proceed to determine the coeficients 9, cy, 2, ..in(62)insuch amanner that the expression (6.2) does indeed satisty the Equation (6.1). But under what conditions is this assumption actually valid? That is, under what conditions can we be certain that the differential equation (6.1) actualy does have a solution of the form (6.2)? Thisis a question of considerable importance; for it would be uit absurd to actually try to find a “solution” of the form (6.2)if there were really no such solution to be found! In order to answer this important question concerning the existence ofa solution of the form (6.2), we shall ist introduce certain basi definitions. For this purpose let us write the differential equation (6.1) in the equivalent normalized form #y ay Ft RO Z + rioy 63 where (x) = 0) 9) = 2200) Fils) 200) and P(x) ol) DEFINITION A fiction fis sid tobe analytic at xg if ts Taylor series about xo, § S (x x6)", ‘exists and converges to f(x) forall x in some open interval including Xo We note that all polynomial functions are analytic everywhere; so also are the functions with values e*, sin x,and cos x. A rational function is analytic except at those values of x at which its denominator is zero. For example, the rational function defined by 1/( example 6.4 Consider the diferenil equation 7 Veetenye 4 Py eM sot tana 64) Here P(x) = x and P,(x) = x? + 2. Both of the functions P, and P, are polynomial |61 POWER SERIES SOLUTIONS ABOUT AN OnoINARY Pow 223 functions and so they are analytic everywhere. Thus all points are ordinary points of this diferential equation, D Example 6.2 Consider the differential equation @y dy 1 @- Dart +z7 0 (65) ‘Equation (6.5) has not been written in the normalized form (6.3). We must first express. (6.5) in the normalized form, thereby obtaining * dx? Here AG) ‘The function P sanalyticyexcept atx = 1,and Pyisanalytcexceptat x = Oand x = | Thus x= 0 and x=1 are singular points of the diferential equation under ‘consideration, Allother points are ordinary points. Not clearly that x = Oisasingular point, eventhough P; isanalytic atx = 0. We mention this fact to emphasize that both ', and P, must be analytic at xp in order for x to be an ordinary point. We are now ina position to state a theorem concerning the existence of power series solutions of the form (62). THEOREM 6.1 Hypothesis. The point xq is an ordinary point of the differential equation (6.1). Conclusion. The differential equation (6.1) has two nontrivial nearly independent ‘ower series solutions of the form Syate- ar 62) and these power series conoerge in someinerel |x ~ xal < R where R > 0) about xo “his theorem gives us sufiient condition for the crstence of power series solutions of the diferential equation (6.2). It states that i x san ordinary pointof equation (6.1), then this equation has two power series solutions in powers of ~ xy and that these two power srics solutions are linearly independent. Thus it x is an ordinary point of (1), we may obtain the general solution of (61) as a linear combination ofthese twolinearly independent power serie, We shall omit he proof of this important theorem224 sams sowuTi0ns oF LINEAR DIFFERENTIAL EQUATIONS > Example 6.3 {In Example 6.1 we noted that all points are ordinary points of the differential equa- tion 6.4). Thus ths differential equation has two linearly independent solutions of the form (6.2} about any point x9 > Bample 64 In Example 62 we observed that x = Q and x = 1 are the only singular points of the differential equation (6.5. Thus this differential equation has two linearly independent solutions of the form (6.2) about any point xe # Oor 1, For example, the equation has two linearly independent solutions of the form F ee 3F about the ordinary point 2. However, we are not assured that there exists any solution. of the form about the singular point 0 or any solution of the form ex = about the singular point 1 B. The Method of Solution Now that we are assured that under appropriate hypotheses Equation (6.1) actually does have power series solutions of the form (6.2), how do we proceed to find these solutions? In other words, how do we determine the coefficients ¢,¢),€25-.. im the expression faenr wy 0 that this expression actually does satisfy Equation (6.1)? We shall first give a brief outline of the procedure for finding these coefficients and shall then illustrate the procedure in detail by considering specific examples. ‘Assuming that x is an ordinary point of the differential equation (6.1), so that solutions in powers of x — xp actually do exist, we denote such a solution by yeey teem) table = Balen (6 Since the series in (66) converges on an difleentiated term by term on this interval twice in succession to obtain ay « a Femey # Boab = no) # Boyle alt too = Besar — xe 16)61 rows sents sownons asour an oxpmuany pow 225, and fy fx — xo)? E in — Neale — x0)" Seen Peat Bele — a0) + Wedel = Zn Dele — 20)" (68) respectively. We now substitute the series in the right members of (66), (6.7), and (68) for yandits fist ono derivatives, respectively in the diferental equation 61) We then simply the resulting expression so that it takes the frm K+ Kile 9) 4 Kyla)? =O, (69) where the coefficients K(i=0,1,2,...) are functions of certain coefficients cy of the solution (6.6). In order that (69) be valid for all x inthe interval of convergence be — xol < R we must set Inother words, wemust equate to zero the coefcient of each power fx ~ xp inthe left ‘member of (6). This leads toa set of conditions that must be satisfied by the various coofcients cy in the series (6.6) in order that (66) be a solution of the diferential equation (6.0. I the care chosen to satisfy the et of conditions that thus occurs, then the resulting series (6.6)sthe desired Solution of the differential equation (6.1) We shall ilustrate this procedure in detail inthe two examples which follow. > Example 6.5 Find the power series solution ofthe differential equation fy ae in powers of x (tat abit xy = 0 +x soreaya0 4) Solution. We have already observed that xy = 0 is an ordinary point of the ‘iferential equation (6.4) and that two linearly independent solutions of the desired type actually exist. Our procedure will yield both of these solutions at once. ‘We thus assume a solution ofthe form (6.6) with xp = 0. That is, we assume * yo Sax. (6.0) Differentiating term by term we obiain (ony, and Sonn dea (6.12) Substituting the series (6.10), (6.11), and (6.12) into the differentia equation (6.4), we226 somes sowwtions oF UNEAR DIFFERENTIAL EQUATIONS ota Ender tex Srawrte Faster fasta ‘Since xis independent of the index of summation n, we may rewrite this as nla Dex? + F neat + ¥ eat? +2 F gxt= 0. (613) In order to put the left member of Equation (6.13) in the form (69), we shall rewrite the first and third summations in (613)so that xin each ofthese summations will have the exponent n, Let us consider the first summation 7 Bre nee (6.14) {n 6.13). To rewrite the summation 6.14) so that x will have the desired exponent, we first replace the present exponent n ~ 2 in (6.14) by a new variable m. That is, we let ‘m= n—2in (6.14) Then m= m + 2, and since m = O for n = 2, the summation (6.14) takes the form Entiat st 9 ‘Now since the variable of surmmation is merely a“durnmy" variable, we may replace m by min (6.15) to write the frst summation in (6.13) as Ente tea 19 In ke manner, letting m 2, the third summation Scat? 7) in (13 fies fom Secu 19 Then replacing m by m in (6.18), the third summation in (6.13) may be written as four 19 Thus replacing (6.14) by its equivalent (6.16) and (6.17) by its equivalent (6.19), Equation (6.13) may be written Line aves verssatt F nests Feat 2 Fax =0. (620) Although x has the same exponent min each summation in (620), the canges of the various summations are not all he same. In the first and fourth summations n ranges from 0 to 2, in the second m ranges from | to co, and in the third the range is from 2 to ce, The common rangeis rom 20 co. We now write out individually the cermin each summation that do not belong to this common range, and we continue (o employ the(61 POWER seRIES SOLUTIONS ABOUT AN ORDINARY POINT 227 “sigma” notation to denote the remainder of each such summation, Forexaraple, ia the first summation Srv as Deas" of (6.20) we write out individually the terms corresponding to n= O and n= 1 and denote the remainder of this summation by Eo 2404 Neue We thus rewrite Err + evar in(620) as 2g + Gye + Lone In + Negeax” In ike manner, we write met ‘in (6.20) as and, ’ in (6.20) as ey dey $2 P eye ‘Thus Equation (620) is now written as ethene § reais teats Smt ‘We can now combine like powers of x and write this equation as (Peg + 26a) + (Bea + Sesbe + F(0-+ 2H Negos + (+ Dey ene ad™ (6.21) Equation (621) is in the desired form (6,9). For (621) 10 be valid for all x in the interval of convergence |x — xo] < R, the coefficient of each power of x in the left228 sens sowvmi0Ns oF UNEAR OIFERENTAL EQUATIONS member of (621) must be equated to 210. This leads immeiatey tothe conditions ep + 2ey = 0 62 Je. + 6ey = 0 623) and (42+ Deve Het Der ea HO, m2? (6% “The condition (622) enables us to express cin terms of cq. Doing so, we find that esc. (625) “The condition (6.23) enables us to express cin terms ofc. This leads to e= Hey (626) ‘The condition (6.24) is called a recurrence formula, It enables us to express each coeticient c,, ; for # 2 2 in terms of the previous coefficients ¢, and ¢,-., thus giving Wet Mentone ease z 27) ee et Dnt 2) ta ade For n = 2, formula (627) is Now using (6.25), this reduces to cxa deo, (628) which expresses cg in terms of cg. For n = 3, formula (627)is +e esp ‘Now using (6.26), this reduces to es then (629) Which expresses in terms of cy. In the same way we may express each even coeflicient in terms of ey and each odd cosficient in terms of ¢,. Substituting the values of ¢5,¢,c4, and cs, given by 6.25), (626), (6 28}, and (6.29), respectively into the assumed solution (610), we have Yseot ex cox? —feyx! + heox! + aherx? + Collecting terms in co and ¢,, we have finally yell x thst teed? tae (630) hich gives the solution of the diferential equation (64) in powers of x through terms in x*, The two series in parentheses in (6.30) ate the power series expansions of two linearly independent solutions of (64), and the constants cy and ¢, are utbtrary constants. Thus (6.30) represents the general solution of (6.4 in powers of x (through terms in x).‘6.1 POWER SeRies SOLUTIONS ABOUT AN ORDINARY POINT 229 D> Example 6.6 Find a power series solution of the initial-value problem fy nites 2 ey GP -Feae Zany 0, 630 yO) = 4, (632) ¥O)=6. (633) Solution. We first observe that all points except x = + 1 are ordinary pointsfor the differential equation (6.31). Thus we could assume solutions ofthe form (66) for any 2p # £1 However, since the initial values of y and its first derivative ae prescribed at X= 0, we shall choose x» = 0 and seek solutions in powers of x. Thus we assume ye Bae 9 Direniating xm by rm e obtain (635) and #y 2 § no Fre Sno tee (636) Substituting the series (6.34), (6.35), and (636) into the differential equation (631), We obtain Sin — Wega’ — § ntn — Dewar? 43. meget + F eget? = 0. (637) We now rewrite the second and fourth summations in (6.37) so that x in each ofthese summations has the exponent n. Doing this, Equation (6.37) takes the form ¥ ln = Negx* x20, (638) Eerie tant ss £ met £ ‘The common range ofthe four summations in (6:38) is from 2 to oo. We can write out ‘the individual terms in cach summation that do not belong to this common range and ‘hus express (6.38) inthe form Xm Megat —2ey— Geax — Fn 2+ Negoaet siqct3fmartart Seouteo ‘Combining like powers of x, this takes the form Deg + (eq + Bey ~ 6es)x (6.39) + Et ait Doves tint Des he230 sames souimions oF uta onFERENTINL EQUATIONS For (6.39) to be valid for all x in the interval of convergence [x — xol = 0; and from (6.41, ¢5 = bea + Jey. The recurrence formula (642) gives Geo OE nz2 Using this, we find successively Wey ten 1 3 BF ict pe ‘Substituting these values of c3, cy, cas €5...- into the assumed solution (6.34), we have 44 (24%) +See (24 Nery yeestans(248 n* tet yell tbe? + bet tte tbe eat thet (643) ‘The solution (6.43)is the general solution ofthe differential equation (6.31)in powers of (through terms in x*). ‘We must now apply the intial conditions (6.32) and (6.33). Applying (6.32) to (6.43), ‘we immediately find that, ena Differentiating (6.43), we have Bccdxt ixt satel tat +h? +Hat to) (648) Applying (6.33) t (644) we find that a6 Thus the solution ofthe given initial-value problem in powers of x through terms in xis aA Bes Be bo) le da + dt Yad t Gxt Ab tat et +{ } | [ i ‘61. Power semies SOLUTIONS anoUT AN ORDINARY POT 231 Remark 1. Suppose theini and (633) of Example 66 are prescribed at x initial-value problem I values of y and it first derivative in conditions (6.32) instead of x = 0, Then we have the Py at + fLiaeti ge Gh Dg t Bg t=O. (645) W=4 ¥Q)=6 Since theft values this problem are prescribed at x» 2, we would sek solutions ‘in powers of & — 2. That is, in this case we would seek solutions of the form ye Sete - 27 (6.46) ‘The simplest procedure for obtaining a solution of the form (6.46) is first to make the substitution ¢ = x — 2 This replaces the initial-value problem (6.45) by the equivalent problem ererrttsouosernen a1) W=4 yO=6 in which isthe independent variable and the inal values are prescribed al = 0.One then seks a solution ofthe problem (647) in powers of y= Zar, (648) Dilferentiating (648) and subaituing into the dierential equation in (6.47, one determines the c, as in Examples 6.5 and 6.6. The initial conditions in (6.47) are then applied. Replacing t by x ~ 2 in the resulting solution (6.48), one obtains the desired solution (646) ofthe original problem (645). Remark 2. In Examples 65 and 66 we obtained power Series solutions of the ‘differential equations under consideration but made no attempt to discuss the ‘convergence ofthese solutions. According to Theorem 6.1 if xis an ordinary point of the differential equation @ 4 aols) G3 + a0) + ayls)y = 0, on then the power series solutions ofthe form 6.2) convergein some interval] x ~ xl < R (where R > 0} about xo. Let us again write (6.1) in the normalized form Tes reo®sroy=o, 6 where 2,03) ate) If x9isan ordinary point of (6.1), the functions P, and P, have Taylor series expansions about xo that converge in intervals |x ~ xo| < Ry and |x— xo] < Ra, respectively, ate, on 000)" and Pils)232 sens so.unons oF WEAR OWFERENTIAL EQUATIONS about x It can be proved that the interval of convergence |x ~ x9| < R of a series solution (6.2) of (6.1)isat least as great asthe smaller of the intervals |x ~ xo| < Rand be xo < Ra Tn the differential equation (6.4) of Example 6.5, P(x) = xand P(x) = x? + 2.Thus inthis example the Taylor series expansions for P, and P, about the ordinary point Xe = O converge forall x. Hence the series solutions (6.30) of (6.4 also converge for all x In the differential equation (6.31) of Example 66, 3x Pw) = and PG) = 5. In this example the Taylor series for P, and P, about xp = 0 both converge for |x| < 1 ‘Thus the solutions (643) of (6.31) converge atleast for |x| < 1 Exercises Find power series solutions in powers of x of each of the differential equations in Exercises 1-10. fy, ay 2y gtr See Be yea 2 Pheer ayaa @y dy ‘i Ee Perper 3 Byte Zs ee + nyo 4 Bite +(e ay a0. @y dy er ay dy oi 5 i Brertayen 6 St Ba army a0. 24 Dey 1 G24 Tie x Zany. ey ” 8. &- FF -Or~- OF + 2y=0. dy 9. @ 0 re 2 syn #y ay 10H DEEH HDR ym. Find the power series solution of each ofthe initial-value problems in Exekcises 11-14 ay dy Fens Zayna y= ya=0 | &y cre | 1, raxZ-2y=0, =0 yO=l a enleBeryen yen x 4 Ot 98F Beno O=-b YO(62. SOLUTIONS ABOUT SINGULAR POINTS; THE METHOD OF FROBENUS 233 Find power series solutions in powers of x — 1 of each of the differential equations in Exercises 15 and 16. ay adty ady is ofp exPeyno 6 tee 17, Find the power series solution in powers of x ~ 1 of the initial-value problem ey as 18, The differential equation Erwan yaad yay Sein ta + mln + Dy =O, aa ‘where n is a constant, is called Legendre's differential equation. (@) Show that x = 0s an ordinary point of this differential equation, and find ‘wo linearly independent power series solutions in powers of x. (b) Show thatif nisa nonnegative integer, then one of the two solutions found in part (a)is a polynomial of degree n 6.2 SOLUTIONS ABOUT SINGULAR POINTS; THE METHOD OF FROBENIUS A Regul Sing Points We agtn consider the homogeneovs ine dlifentl equation ay dy cs Ey + 04092 + ay =0, on and we assume that x9 isa singular point of (6.1) Then Theorem 6.1 does not apply at the point x, and we are not assured of @ power series solution y= Eater 62) of (6.1)in powers of x ~ xp. Indeed an equation ofthe form (6.1) with a singular point at xp does not, in general, have a solution of the form (6.2). Clearly we must seek a different type of solution in such a case, but what type of solution can we expect? It happens that under certain conditions we are justified in assuming a solution of the form = all Fes sah 649) where ris a certain (real or complex) constant, Such a solution is clearly a power series inx ~ xp multiplied by a certain power of |x ~ xy|-In order to state conditions under Which a Solution of this form is assured, we proceed to classify singular points.234 SERIES SOLUTIONS OF UINEAR DIFFERENTIAL EQUATIONS ‘We again write the differential equation (6.1) in the equivalent normalized form ay dy a ae tO, + hely=0, (63) whee 50) gg yy Se Pie) ao) dP, (x) g(x) Derinimion Consider the differential equation (6.1), and assume that at least one of the functions P, ‘and P, in the equivalent normalized equation (6.3) 18 not analytic at xq, 80 that x» is @ singular point of (6.1. If the functions defined by the products (= Xe)P(X) and (x ~ x0)? Py(a) (6.50) are both analytic at x9, then xy is called a regular singular point of the diferential equation (6.1) If either (or both) of the functions defined by the products (6.50) isnot analytic at xp, then x, is called an irregular singular point of (6.1). > Example 6.7 ‘Consider the diferential equation 2ty Dy oye si a Fea B le Sya0 (651) Writing this in the normalized form (6.3), we have fy a Here P(x) = ~1/2and P(x} s(x — 52x? Since both P, and P,filto be analytic at x=, we conclude that x =0 is a singular point of (651), We now consider the functions defined by the products xP.) ‘of the form (6.50). Both of these product functions are analytic at x = O,and so x = Ois a regular singular point of the differential equation (6.51) > Example 6.8 ‘Consider the differential equation tug op PY, ay : we ES + De DE + Oe + ty = 0. (652)62 SOLUTIONS ABOUT SINGULAR POTS, THE METHOD OF FRosENNUS 235 In the normalized form (6.3), this is dy 2 dy xe at RD) de * Pe - 2P Here xet Fey me BO aa RO)= (Cary te singular pins of the diferent equation (652) are x = Oand x = 2. We investigate them one ate. ‘Consider x= fst and form the functions defined by the products 2 a4 > and x7P,(x)= 3e- Baa of the form (6.50). The product function defined by x*P,(x)is analytic at x = 0, but that defined by xP; (x) is not. Thus x = 0 is an irregular singular point of (6.52). ‘Now consider x = 2, Forming the products (6.50) for this point, we have GDA )= 3 and (e- 27-00-24 xP Both of the product functions thus defined are analytic at x = 2,and hence x = 2isa regular singular point of (652). Now that we can distinguish between regular and irregular singular points, we shall state a basic theorem concerning solutions of the form (6.49) about regular singular points, We shal later give a more complete theorem on this topic. THEOREM 6.2 Hypothesis. The point x isa reqular singular point ofthe differential equation (6.1) Conclusion. The differential equation (6.1) has at least one nontrivial solution of the Jorn ronet $ote sh 6s) where risa definite (real or complex) constant which may be determined, and this solution is valid in some deleted interval 0 < |x — xp| < R (where R > 0) about Xo. D> Example 6.9 In Example 6.7 we saw that x = Qisa regular singular point ofthe differential equation os236 ‘SERIES SOLUTIONS OF LINEAR DIFERENTIAL EQUATIONS By Theorem 6.2 we conclude that this equation has at least one nontrivial solution of the form tar Eat valid in some deleted interval 0 < |x| < R about x =0. D> Example 6.10 Jn Fxample68 we saw that x = 2isa regular singular point ofthe diferential equation 2¢0 — 2 4 20-2) ext ny HW BEE 4 eI P + et Ny =O. (652) ‘Thus we know that this equation has at least one nontrivial solution of the form x= 2r eer, valid in some deleted interval 0 < [x ~ 2] < R about x = 2 We also observed that x = 0 sa singular point of Equation (632). However, this singular point is iregular and so Theorem 6.2 does not apply toit, We are not assured. ‘that the differential equation (6.52) has a solution of the form wi Ea in any deleted interval about x = B. The Method of Frobenius Now that we are assured of at least one solution of the form (6.49) about a regular singular point xe ofthe differential equation 6.1), how do we proceed to determine the coefficients c, and the number r in this solution? The procedure is similar to that introduced in Section 6.1 and is commonly called the method of Frobenius. We shall briefly outline the method and then ilustrate it by applying it to the differential ‘equation (61). In this outline and the illustrative example that follows we shall seek solutions valid in some interval 0 < x — xp < R. Note that forall such x, |x ~ xol is simply x — xp. To obtain solutions valid for —R < x — x9 < 0, simply replace x ~ x9 by (a ~ xo) > Oand proceed as in-the outline. Outline of the Method of Frobenius 1. Let x9 bea regular singular point of the differential equation (6.1), seek solutions valid in some interval 0 Example 6.11 Use the method of Frobenius to find solutions of the differential equation a@y dy 2 Fh xB ee sya 0 651) in some interval 0 0. However, in Step 8 of the outline preceding Example 6.11, we indicated that this is not always the case. Thus we are led to ask the following ‘questions: 1 Under what conditions are we assured that the differential equation (6.1)has two Uinearly independent solutions ye bx —nol $e — a ofthe form (649) about a regular singular point x5? 2. Ii the differential equation (61) does nor have two linearly independent solu- tions of the form (649) about a regular singular point x, then what isthe form of a solution that is linearly independent ofthe basic solution ofthe form (6.49)? In answer to these questions we state the following theorem. THEOREM 6.3 Hypothesis. Let the point x be a regular singular point of the differential equation (6D, Let ry and r, [where Re(r,) = Relr3)] be the roots of the indicial equation ‘associated with x Conclusion 1. Suppose r,—r; 4 N, where N is a nonnegative integer (that is, Fs #0,1,2,3,-.-} Then the differential equation (6.1) has two nontrivial linearly Independent solutions y, and yp of the form (6.49) given respectively by will |x aol" Seb 0), (665) where co #0, and vats) = ix aol” Bote — 00, (6656) where o #0. Conclusion 2. Suppose ry—ry=N. where Nis a positive integer. Then the diferential equation(6.1) has to nontrivial linearly independent solutions y, and y, given respectively by ala) = [x= a" Selo (665) where cy #0, and yale) = [x = Sef — aol + Cos laln x = xy (667, where e$ O.and C is a constant which may or may not be zero.{62 SOLUTIONS ABOUT SINGULAR POINTS: THE METHOD OF Roses 241 Conclusion 3. Suppose ry — rz =0. Then the differential equation (6.1) has two nontrivial Hnearly independent solutions y, andy, given respectively by yale) Lx = x0l" Sex = x0) (665) where cg # 0, and vals) = | — xo Ce — xo" + yy Gln x ~ 0) (6.68) ‘The solutions in Conclusions 1, 2, and 3 are valid in some deleted interval 0 < el < R about Xo In the illustrative examples and exercises that follow, we shall again seek solutions valid in some interval 0 < x — xo < R. We shall therefore discuss the conclusions ‘of Theorem 6.3 for such an interval, Before doing so, we again note that if 0-< x — xy < Ry then |x ~ xol is simply x —%9. From the three conclusions of Theorem 6.3 we see that point of (6.1), and 0 Example 6.12 Use the method of Frobenius to find solutions of the differential equation 20y We at Br tx P+ (et — Hy = (669) in some interval 0 Example 6.13, Use the method of Frobenius to find solutions of the differential equation fy dy pag oh (ee ay a0 676, in some interval 0 [mt nintr—1-en—eget ~ Senet =o {rir fleae’ + fle + tr + deat + Ent nner 9-49 dene a}x""=0, (678) Equating to zero the coefficient of the lowest power of x in (6.78), we obtain the indicial equation a-d=0, The roots of this equation are nehonect Although these roots themselves are not integers, the diflerence r, — 1, between them is the positive integer 3. By Conclusion 2 of Theorem 6.3 we know that the differential ‘equation (6.76) has a solution of the assumed form (6.77) corresponding to the larger root r, = 4. We proceed to obtain this solution. Equating to zero the coefficients of the higher powers of x in (6.78), we obtain the condition (e+ Dr + )-De (6.79) and the recurrence formula [fn Hae D= Ot mde nz (680) Letting r= ry = $n (679), we obtain 4e,=0 and hence ¢, =0. Letting = r= fn (6.80), we obtain the recurrence formula mint See e202, corresponding to the larger root $. Since n > 2, we may write this in the form62. SOLUTIONS ABOUT SINGULAR POINTS; THe METHOD oF Feoweus 245, From ths we obtain successively a ‘We note that all odd coefficients are zero, The general even coefficient may be written a = O ince ¢ Tees Te ea Ft Letting r= j in (6.77) and using these values of ¢2y, sponding tothe larger root ry = 4. This solution is ve obtain the solution core- = yx (0), where nome teeta] moor} +2 ee ams 79 ~ =e] ‘We now consider the smaller root r; = —4. Theorem 6.3 does not assure us that the diferential equation (6.7) hase inary independent solution of the assumed form {6:77 coresponding to this smaller root Conclusion 2 of that theorem merely eli us that theresa inary independent solution a the form Sarees cyan (682) where C may or may not be zero. OF course fC = 0, then the linearly independent Solution (682) ofthe assumed form (6.77) and we can let r= r2 = —}in the formula {69)and the recurrence formula (680) and proceed asin the previous examples. Let us assume (hopefully, but without justification!) that this is indeed the case. ‘Thus we let r =r, — —$.in (6.79) and (680), Letting r = —Jin (6.79) we obtain 2c, = Oand hence c, = 0. Letingr = —+ in(6.80) we obtain the recurrence formula nin —Deq— e202 22, (683) corresponding to the smaller root —}. For n #3, this may be written rey nad #3 (684) For n = 2, formula (6.84) gives c, = —c9/2. For n = 3, formula (6.84) does not apply land we must use (683). For n= 3 formula (683) is 0° ¢, ~ cy = 0 or simply 0'= 0 (since ¢, = 0) Hence, for n = 3, the recurrence formula (683) is automatically satisfied with any choice of e,. Thus cis independent ofthe arbitrary constant cit isa second arbitrary constant! For n > 3, we can again use (684). Proceeding, we have 34 re es raeee eT Test We note that all even coefficients may be expressed in terms of co and that all odd246 semes sowuTions oF UNEAR DIFFERENTIAL EQUATIONS. coefficients beyond cy may be expressed in terms of ¢s. In fact, we may write oR ee aD ST mF (even coficens cand beyond), nd n22 {odd coefficients c, and beyond). Letting x = —} in (6.77) and using the values ofc, in terms of ¢, (for even n} and ¢, (for odd m beyond ), we obtain the solution ‘corresponding to the smaller root r; = —$. This solution is y = ya(x), where pabd=ox | 2 $ a DPA EHO GT ST OD] sovults Sao TES TI Gas and ¢9 and c, are arbitrary constants If we now let co = 1 in (681) we obtain the particular solution y = rae aarels & re Gas 7 corresponding to the larger root 4; and if we let cy = 1 and cs = 0 in (6385) we obtain the particular solution y = y,, (x), where ratornen[ orem eT ‘corresponding to the smaller root —}. These two particular solutions, which are both ‘of the assumed form (6.77), are linearly independent. Thus the general solution of the 1, we may write this in the form From this we find successively Letting 7 sponding to the larger root r; = confi We recognize the series in brackets inthis solution as the Maclaurin expansion for e°* Thus we may write in (689) and using these values of c,, we obtain the solution eorre- This solution is y =y, (x), where x ae Seg Md vate exe (692) and express the solution corresponding to r, in the closed form where cis an arbitrary constant, We now consider the smaller root r; = 1. Asin Example 6.13, we have no assurance that the differential equation has a linearly independent solution of the assumed form (689) corresponding to this smaller root. However, as in that example, we shall tentatively assume that such a solution actually doesexist and letr = r; = 1 in(6.91)in the hopes of finding “it” Further, we are now aware that this step by itself might even provide us with the general solution. “Thus we let r = ry = 1 in (6.91) to obtain the recurrence formula nit —2)e, + ney y= 0, Bh, (693) ‘corresponding to the smaller root 1. Form # 2, this may be written, nel ned (694) Forn ust use (693). For 1. formula (694) gives, 2.formula(6,94)does not apply and we 2 formula (693) is 0c, + 2c = 0, and hence we must have62, SoumoNs Asour sincLuaK powTS: THe METHOD OF rRosemes 249) 0. But then, since ¢, = cg, we must have cy = 0. However, co # Oin the assumed solution (6.89) This contradiction shows that there is no solution of the form (689) with co #0, corresponding to the smaller root 1 Further, we observe thatthe use of (6:94) for n = 3 will only lead us to the solution yy already obtained, For, from the condition 0- cz + 2c, = O-we see that c, is arbitrary; and using (694) for n > 3, we obtain successively ase, aa-faG. we ee es Fe Geren = ne Thus letting r = in (689) and using these values of c,, we obtain formally sore. feet | Comparing this with (692), we sce that i is essentially the sotation y = y4{2)- We now seck a solution of (6.88) thats linearly independent of thesolution y,. From, ‘Theorem 63 we now know that this solution is ofthe foren See's cen «99 where «3 # Oand C + 0. Various methods for obtaining such a solution are available; \we shall employ the method of reduction of order (Section 4.1). Welety = ftx)e, where {J{s)i8 a known solution of (6.88). Choosing for f the known solution y, defined by (692), with ¢g = 1, we thus let yawe te, (696) From this we obtain (697) ‘and # ée Ba deta brte t+ be (698) we OE 4 aches = x04) Substituting (6.96), (697), and (6:98) for y and its first wo derivatives, respectively, in the differential equation (6.88), after some simplifications we obtain & le ot aie Ga rb-0F =0. (699) Letting w = do/da, this reduces at once to the first-order diferentiat equation vit £Q- aw a0.250 SERIES SOLUTIONS OF LINEAR DIFFERENTIAL EQUATIONS particular solution of which sw Thusa particular solution of (6.99) is given iy vo frre ae and hence y = ya(x), where a(x) = ver{s Fe dx (6.100), is a particular solution of (6.88) that is lincarly independent of the solution », defined by (6.92). ‘Wenow show that the olution y, defined by (6100) is f the form (695) Introducing the Maclaurin series for e* in (6100) we have ) a noe stern Pror(Ltae se [festa ttyl yt x weer f(eteet abe ete zen jan Integrating term by term, we obtain wpe te[ a tatinee tea bate sab =x ryt ent get gee Now introducing the Maciaurin series for e°*, we may write pote atete de vats (Pox . sacitper ges ) a 4detinx Finally, multiplying the two series involved, we have dxdt dbx t tbe tin, hich is ofthe form (695), where (x) = xe"*. The general solution of the differential ‘equation (688) may thus be written C6) Fava where Cy and Cy are arbitrary constants. Tn this example it was fortunate that we were able to express the first solution y, in closed form, for this simplifed the computations involved in finding the second solution y- Of course the method of reduetion of order may be applied to find the second solution, even if we cannot express the frst solution in closed form. In sch ceases the various steps of the method must be carried out in terms of the series expression for y,. The computations that result are generally quite complicated Examples 6.12, 6.13, and 6.44 illustrate all of the possibilities listed in the conclusions of Theorem 6.3 except the case in which r, ~ r, = O(thatis, the casein which the roots ‘of the indicial equation are equal). In this case it is obvious that for 0 0, which we have already introduced at the beginning of Section 6.3A, and seek solutions valid for O-< x < R This isthe equation Pg Og at pt Fx Zewe- Wy = 0, (10H) igure 6142 arssevs eQuarion AND aes. FUNCTIONS 257 ‘whore we now assume that p is real and positive. We see at once that x = 0 isa regular singular point of Equation (6.101); and since we seek solutions valid for 0 < x < R, we ‘may assume a solution Sew, (6112) where ¢y # 0, Diflerentiating (6.112), substituting into (6.101), and simplifying asin our previous examples, we obtain (P= pox! + [Ue + UF pei +E tot pile tea} = 0 6113) Equating to zero the coefficient of each power of x in (6.113), we obtain uy) Cr+ 1? — p*Je, = 0, (6.115) and [mtn plate = 0, 222 (6116) Equation (6.114) isthe ndicial equation ofthe differential equation 6.101). Its roots are r= p>Oandr, = ~p.Ifr, —r; = 2p > Oisunequal toa positive integer, then from “Theorem 6.3 we know that the differential equation (6.101) has two linearly inde- pendent solutions of the form (112) However, fr, 3 = 2p is equal toa postive integer, we ae only certs of a solution ofthis form corresponding t the larger root rp, We shall nw proced to obtain this one solution the existence of which ahve assured Letting r = r, = p in (6.115), we obtain (2p + Ie, = 0. Thus, since p > 0, we must have e: = 0. Letting in 6.116}, we obtain the recurrence formula nin + Ie + e-2=0 022, ate, ne (at) “ain + 2p)" corresponding tothe larger root p. From this one finds that all odd cocficents are zo" (ince = 0) and that the general even coefficient is given by TR QM IER + 2p + 2p) Gee + 2p} Yeo "FTC e+ + BI nel Hence the solution of the differential equation (6.101) corresponding to the larger root is given by y= yx}, where (6.18)258 SERIES SOLUTIONS OF LINEAR DIFTEREVTIAL EQUATIONS If pis a positive integer, we may write this in the form _ _iy fxyeer vals = cotta So 8 (113) It pisunequat toa positive integer, we need a generalization of the factorial funtion in order to express J() in a form analogous to that given by (6.119). Such a gen- eralization is provided by the so-caled gamma function, which we now introduce. For N > O the gamma function is defined by the convergent improper integral tape [Pen 1f Nisa postive integer, it can be shown that NI= TW +1), (121) If Nis positive but not an integer, we use (6.121) to define N?, The gamma function has been studied extensively. It can be shown that F() satisfies the recurrence formula TW += NIN) (> 9} (6122) Values of F(N) have been tabulated and are usually given for the range 1 < N <2. Using the tabulated values of F(N)for 1 < N < 2,onecan evaluate F(N forall. > 0 by repeated application of formula (6.122). Suppose, for example, that We wish to ‘evaluate (3). From the definition (6.121),we have(3)! = (2). Then from (6.122) we find that 1) = 31) From tables one finds that F(f) © 0.8862, and thus @! = €{3) = 13383 For N < 0 the integral (6.120) diverges, and this FN) isnot defined by (6.120) for negative values of N. We extend the definition of F{N) to values of N-<0 by demanding that the recurtence formula (6.122) hold for negative (as well as positive) values of N. Repeated use of this formula thus defines fN) for every nonintegral negative value of N. Thus F(N) is defined for all N #0, — 1, ~2, ~5,.... The graph of this function is shown in Figure 62. We now define NI for all Nw ~1, ~2, —3,... by the formula (6121, ‘Wenow return to the solution y, defined by (6.118), for the ease in which pis unequal 10.2 positive integer. Applying the recurrence formula (6.122) successively with NV npn p—ln+p—2,...,p-+ 1, we obtain Pint p+ (nt ole tp— Do + Wp + DD ‘Thus for p unequal toa positive mteger we may write the solution defined by 6.118) in the form ae nbo=eFe oF sa orien Sat worn ‘Now using (6.121) with NV = pand N = 7 + p,wesee that (6.123) takes the form (6.119), ‘Thus the solution of the differential equation (6.101) corresponding to the larger63 BESSEL'S EQUATION AND sesstt FUNCTIONS 259) Figure 6.2. Graph of I (N) root p> 0 is given by (6.119), where pl and (n+ p)! are defined by Tp +1) and Tn p + li respectively, if p is nota positive integer Ir weset the arbitary constant cy in (6.119) equal to the reciprocal of 2p, we obtain ‘an important particular solution of (6.101). This particular solution defines 2 function denoted by J and called the Bestel function of the first kind of order p. That is, the fonction J, is the particular solution of Equation (6.101) defined by ty ga oe Z ee Dy=0, (6129 ‘which is Ressel’s equation of order one. Letting p = 1 in (6.124) we obtain a solution of [Equation (6.125) that is called the Besse! function of the first kind of order one and is denoted by J. That is, the function J, is defined by 1o- Lame) 9260 sens so.mions oF uNeaR DIFFERENTIAL EQUATIONS. ‘The graphs ofthe functions Jp, defined by (6.107), and J,, defined by (6.126), are shown. in Figure 63. Several interesting properties of the Bessel functions of the first kind are suggested by these graphs. For one thing, they suggest that Jy and J, each havea damped oscilla- tory behavior and that the postive zr0s of J and J, separate each other. Thisis indeed the case. in fact, it may be shown that for every p 2 0 the function J, has a damped oscillatory behavior as x + oo and the positive zeros of Jpand J, eparateeach other. ‘We now know that for every p 2 0 one solution of Bessel’s equation of order p (6.101) is given by (6.124). We now consider briefly the problem of finding a linearly independent solution of (6.101). We have already found such a solution for the case in which p ~ 0; itis given by (6.110) For p > 0, we have observed that if 2pis unequal toa positive integer, then the differential equation (6.101) has a linearly independent solution of the form (6.112) corresponding to the smaller root r; = ~p. We now ‘proceed to work with this smaller root. Letting r = r3 = ~pin (6.115), we obtain (2p + He, =0. (6127) Letting r= ry = —p in (6.116) we obtain the recurrence formula ME 2Phee ey a =O, mB? (6.28) n22, n#2p. (6129) Using (6.127) and (6.128) ot (6.129) one finds solutions y = y3(x), corresponding to the smaller root ~ p. Three distinct cases occur, leading to solutions of the following forms: 1. IF 2p # a positive integer, v2) =eor(t +3 ‘where cg is an arbitrary constant and the a3,(% 2. If 2p = an odd positive integer, 1100)» (1 +3 fox) tare(1 +B yee \, (6.3) = au") (6130) 4,2...) are definite constants, Figure 6.3463 Esser’s eQUATION AND Besse ruNcTions 261 where ¢o and cap are arbitrary constants and fz_ and ya = 1,2...) are definite constants, 3. If 2p = an even positive imeger, I20) = Cay (is fae} wn 3...) are definite constants, where cz, is an arbitrary constant and the 83.(0 In Case | the solution defined by (6.130) s linearly independent of J,.In Cass 2 the solution defined by (6.131) with cap = Os also linearly independent of J. However, in ‘Case 3 the solution defined by (6.132) is merely a constant multiple of Jj(x), and hence this solution isnot linearly independent of J. Thus if 2p is unequal to an even positive integer, there exists @ linearly independent solution of the form (6.112) corresponding. to the smaller root —p. In other words, if p is unequal to a positive integer, the dliffezential equation (6.10%) has a solution of the form y = ys(x), where vat0)= F eans™% (6.133) and this solution ys linearly independent of J, Itis easy to determine the coeficients¢ in (6.133), We observe thatthe recurrence formula (6.129) corresponding to the sinaller root pis obtained from the recu formula (6.117} corresponding to the larger root p simply by replacing p in (6.117) by =p. Hence a solution of the form (6.133) may be obtained from (6.124) simply by replacing pin (6.124) by —p. This leads at once to the solution denoted by J-, and defined by where (n — pt is defined by Fn ~ p+ 1}. ‘Thusif p > Ois unequal toa postive integer, two linearly independent solutions of the diferental equation (6.10) are J, defined by (6.124), and, defined by (6.134. Hence, fp > 0is unequal toa positive integer, the general olation of Bessesequation ‘of order pis given by Y= Cr.) + Crd), where J, and J., are defined by (6.124) and (6.134), respectively, and C, and C, are arbitrary constants. If pis positive integer, the corresponding solution defined by (6.123) isnot linearly independent of J, as we have already noted. Hence in this case a solution that is linearly independent of J, must be given by y = y,(x), where pooner Rene schoatns where C #0, Such a linearly independent solution y, may be found by the method of, ‘eduction of order. Then the general solution ofthe differential equation (6.101) may be written asa general linear combination of J, and y,.. However, asin the case of Besse!'s ‘equation of order ero, itis customary to choose acertain special linear combination of Jp and y, and take this special combination as the “second” solution of Equation262 series sovumions oF NEAR OMFERENTIAL EQUATIONS. (6.101) This special combination is denoted by ¥Y, and defined by yee) yoo? {(in§ + ae) hor (rk) laraG) ]} a ‘where 7 is Euler's constant. The solution ¥, is called the Bessel function ofthe second kindof order p (Weber's form. Thus f pis postive integer, two linearly independent solutions ofthe éiferential ‘equation (6.101) are J, defined by 6.124), and ¥,, defined by 6.135). Hence if ps & positive integer, the general solution of Besse’ equation of order pis given by PH CLe)+ CYC), where J, and ¥, are defined by (6.124) and (6.135), respectively, and C, and C; are arbitrary constants. Exercises 1. Show that Jo(kx), where kis a constant, satisfies the differential equation dy dy xit + ea =0 2. Show that the transformation reduces the Bessel equation of otder p, Equation (6.101), to the form f3[ie(t-#) fea Use the result of Exercise to obtain a solution of the Bessel equation of order 4 4. Using the series definition (6.124) for J, show that 4 ergs) = kx" alk) Ceryflen] = koxp ah) and (0, 4 erp ixy} = Fle rye) = be where kis a constant 'S._ Use the results of Exercise 4 to show that Ae = bys) —2 40, A cyte = —t.s0ho +2 ten63 BUSSES EQUATION AND aessét uNCTions 263, Hence show that a k By Wad) = 5 Lh) — J lad] (ex) = Jp (he) + Jp fh) ory 6 Using the results of Exercise 5 (a) express Ja) and 2, (0 in terms of Ja) and Ja (0s (©) express Jp a(x) im terms of J,yya(x) and Jy aC= CHAPTER SEVE ‘Systems of Linear Differential Equations In the previous chapters we have been concerned with one differential eq unknown function. Now we shall consider systems of two differential equ: unknown functions, and more generally, systems of n differential equations in n unknown functions. We shall restrict our attention to linear systems only, and we shall begin by considering various types of these system. Alter this, we shall proceed to introduce differential operators, present an operator method of solving linear systems, and then consider some basic applications of this method, We shall then turn toa study of the fundamental theory and basic method of solution for a standard type of linear system in the special case of two equations in two unknown functions. Following ths, ‘ve shall outline the most basic material about matrices and vectors. We shall then present the basic method for the standard type of linear system by means of matrices and vectors, first in the special case of two equations in two unknowa functions and ‘then in the more general case of » equations in n unknown functions. The fundamental theory fr this general case wll be presented in Chapter 11 7:1 DIFFERENTIAL OPERATORS AND AN OPERATOR METHOD. 264 ‘A. Types of Linear Systems ‘We start by introducing the various types of linea systems that we shall consider. The general linear aystem of two firstorder differential equations in two unknown Fnetions x and y's of the form (+ 0,0 + agit + aultdy = Fl. aul) Ge + all) + anl0x + auly = Fuh oy & yet BOF + b.(0 3 + balOx + bal = FO.74 DIFFRENTAL OPERATORS AND AN OreATOR MEIOD 265, We shall be concerned with systems of this type that have constant coefiicients. An example of such a system is of system (7.1) san ordered pair of real functions (f, 9) such that x = (0 y = g(¢)simultaneously satisty both equations of thesystem (7.)on some real interval a St b. ‘The general linear system of three first-order differential equationsin three unknown functions x, y, and zis of the form 20 F + ano Z + esl + ln + aly + 41 = Fl 50 + (0D + bE + bane + Boldy + BANE HO — 22) Yee es ext E+ ext 2 + est FE + cate tes(y + eal? BO ‘Asin the case of systems ofthe form 7.1), s0 als inthis case we shall be concerned with systems that have constant coefficients. An example of such a system is dx dy_ yd i Gitta ayee dx_ dy, yd By 3B 4x say—sesing, 2 Gra G eet ay Se =sing de ay ae Ft iG aE — dx 2y- 2m cont We shall say that a solution of system (72) an ordered triple of real functions (fg, fi) such that x = f(9, » = g(®), 2 = h(®) simultaneously satisfy all three equations of the system (7-2) on some real interval a <1 < b. Systems of the form (7-1) and (7.2) contained only first derivatives, and we now ‘consider the basic linear system involving higher derivatives. Ths is the general linear system of two second-order diferential equations in two unknown fucutions x and y, and isa system of the form Pe yD & aa? a0 SF + 0 EF + agi + 040% + asl + a4(y = Fe 2, 2, 03) Pe Ta mn% nin” y= bio Te + bE + bul E+ bal) + bald + duly = FO. We shall be concerned with systems having constant coeficients in this case also, and ‘anexample is provded by’ #y 58 59, ye SerBasBeyerss, tx ty yay i 305 428} 28 at ymo.266 SYSTEMS OF UNEAR DIFERENTIAL EQUATIONS For given fixed positive integers m and n, we could proceed, in like manner, exhibit fother gencral linear systems of m mih-order differential equations in n unknown functions and give examples of each such type of system. Instead we proceed to introduce the standard type of linear system referred to inthe introductory paragraph at the start of the chapter, and of which we shall make a more systematic study Tater. We introduce this standard type as a special case of the system (71) of two first-order differential equations in two unknowns fanctions x and y ‘We consider the special type of linear system (7.1), wbich is of the form Fm asl + ays(0y + Filth i a i a) Bw asst # arly + Fld This isthe so-called normal form in the case of two linear differential equations in two tunknowa functions. The characteristic feature of such a system is agparent fcom the ‘manner in which the derivatives appear init. An example of such a system with variable coefficients is dx . Fen Oe Hee Dy ee, while one with constant coeficients is ax a dy a e+ Tye, By + ‘The normal form in the case of a linear system of three differential equations in three unknown functions x,y, and 2is a Fm austne taaldy + ays + FO, Bo anite + anttly + ant? + Fld, a G7 NOX + any + analOe + Fld. ‘An example of such a system is the constant coefficient system bape Barw-ay45—0, ty 3et 24124. DIFFERENTIAL OFGRATORS AND AN OFsRATOR METHOD 267) ‘The normal form in the general case ofa linear system of » differential equations in n unknown functions 1, X35.-1.%i8 = aya (hx, + aialthea bo + aaaltDSs + Full as (ther + anal +70 + aaa + Oe os) FE aa lOny + aga lOXs H+ al, + Fld An important fundamental property of a normal linear system (7.S)isits relationship toa single nth-order linear differential equation in one unknown function. Specifically, consider the so-called normalized (meaning, the coeficient of the highest derivative is one) nth-order linear differential equation as as ae Get al eet ta Se tals = FO) 76) inthe one unknown function x. Let _dx 2 oma a) x ats ats aye Grohe = From (7.7), we have dx dey Bx dey ah deg a _dp na z wea ae a a ‘Then using both (7.7) and (78), the single mth-order equation (7.6) can be transformed into eae & fiers, as) = a8, = a 4(0K2 ~ 1° — 6484 + FIO, which is @ special case of the normal linear system (7.5) of n equations in n unknown functions. Thus we see that a single nth-order linear differential equation of form (7.6) is ‘one unknown function is indeed intimately related to a normal linear system (7.5) of x first-order differential equation in n unknown functions.268 SYSTEMS OF UNEAR DIFERENTIAL EQUATIONS B. Differential Operators In this section we shall present a symbolic operator method for solving linea systems ‘with constant coefficients. This method depends upon the use of so-called diferential operators, which we now introduce. Let x be an n-times differentiable function of the independent variable t. We denote the operation of diflerentistion with respect to ¢ by the symbol D and call Da differential operator. In terms of this differential operator the derivative dx/dt is denoted by Dx. That is, Dx = dxjat Tolike manner, we denote the second derivative af x with espect tt by Dx. Extending this, we denote the nth derivative of x with respect tor by D'x, That i, ude (12.03 wr Further extending this operator notation, we write dx (Dex todenore tex and ae te oth pete (op" + bD")x todenote a Fe + dT, where a band ¢ are constants. In this notation the general linear differential expression with constant coefiicients dx oye tn GaeT IE Oa GH a is written agD* + aD! 4 ba aD bas ‘Observe carefully that the operators D*,D*~",..., Din this expression do not represent ‘quantities that are to be multiplied by the Function x, but rather they indicate operations (differentitions) that are to be carried out upon this function, The expression gD" + DI! b+ ay D+ dy by itself, where a dyson ‘with constant coeficients 4, are constants, is called a linear differential operator D> Example 7.1 Consider the linear differential operator 3D? + 5D~273 DIFFERENTIAL OPERATORS AND AN OPERATOR weTHIOD 269 If xis twice diferentiable function oft, then 4 @x | de (08 +5D~2}x denowes 35 + 5H — 2x For example, if x= ¢, we have 2 sas ast : (80° + 5D — 2) = 3-0) 45-4 (0) — He) = 18+ 15 = 20, ‘We shall now discuss certain useful properties ofthe linear diferental operator with constant coefficients, In order to facilitate our discussion, we shal let L denote this ‘operator. That is, 7 1 iD? + aD" where do, ayy.+-« 4-1 dq are constants. Now suppose that f, and J are both n-times differentiable functions of ¢ and c, and ¢3 are constants. Then it can be shown that Leif + eal = Lf + elf For example if the operator L = 3D? + SD — 2isapplied to 31? + 2 sin, then U3 + 2 sin f] = 32] + 20fsin 2) HE aD + Oy (BD? + SD ~ 2)31* + 2 sin 1) = 3(3D* + 5D ~ Je? + 213? + 5D — Asin Now let Ly = 09D" + aD"! 4-44, D+ aq and Ly BoD + DUE to by yD +d, bbe two linear differential operators with constant coefficients, 4... mts @av nd Boy Byyoons Buss bys respectively, Let Lip) agp” + a St Ogi oy and Lal) = bgt + By! boo bby art By 'e the two polynomials in the quantity r obtained from the operators Ly and Ly, respectively, by formally replacing D by r, D? by r?,..., D! by r® Let us denote the product ofthe polynomials £,() and £3() by L(t thats, Un = Lantatn “Then, if isa function possessing m + m derivatives it can be shown that Libaf = balif = Lh (7.10) where Lis the operator obtained from the “product polynomial” £4) by formally replacing r by D, 1? by D?,...,*** by D™'*. Equation (7.10) indicates two important Properties of linear differential operators with constant coefficients, Firs it states the elfectof first operating on f by 1. and then operating on the resulting function by L, is the same as that which results from first operating on f by L, and then operating on270 SYSTEMS OF UNEAR OFFERENTIAL EQUATIONS this resulting function by Ly. Second, Equation (7.10) states that the effect of first ‘operating on J by either L, oF 1 and then operating on the resulting function by the others the ame as that which results from operatingon f by the “produet operator” L Weillustrate these important properties inthe following example. > Example 7.2 Let by =D? + 1,1; 23D +2. f()= 7. Then Likaf = (2? + GD +2}0 =D? +190 + 20°) = 910+ Ne +20 + HE 9241) 26614) = 2 +917 + 128+ 18 and Lal f = BD + 21D? + 1)P = BD 4 246+) 60D + 2) + 6D 42) 63-421) + O12 + 20) = 2 4908 4 124 I Finally, L = 3D° + 20? + 30 + 2and Lf = GD" + 20" + 3D + 20? = 366) +2160) + 37) 4 20° 32? 49 + 2+ 18 Now let = 0g" = aD! 4-0 4 ay Dot dy, Where a, dyyony Bynys AFC constants; and let Up) = dor" + ayr"! 4" + dy” yr + a, be the polynomial in ‘obtained from L by formally replacing D by r,D? by r®,..., D™ bY. Let, Fayones Fa BE the roots of the polynomial equation Lp) = 0. Then Lip) may be written in the factored form Uh = ag(h = rir = ryote =) Nov formally replacing r by Din the right member of this identity, we may express the ‘operator = dgD" + a,D""! +--+ a,_4D + a, in the factored form, L = ay(D ~ ry (D = r3)---(D =r) ‘We thus observe that linear differential operators with constant coefficients can be formally multiplied and factored exactly as if they were polynomials inthe algebraic quantity D. C. An Operator Method for Linear Systems with Constant Coefficients We now proceed to explain a symbolic operator method for solving linear systems with constant coeflicients, We shall outline the procedure ofthis method on a strictly formal basis and shall make no attempt to justify it. We consider a linear system of the form Ext Lay flO Lyx + Lay = 00. where L,, Ly, Ly, and Ly are linear differential operators with constant coefficients oan74. IFFERENTIAL OPERATORS AND AN OpsaAroR METHOD 27 That is, Ly Ls, Ly, and Ly are operators ofthe forms Ly 8 aD" + aD"! $0 Og 1D + gy Ly 5 byD* + 5D $+ 6D by, Ly gD" + aDP $0 $95 D + ayy Lag = BoD8 + BDI + By D+ Bs where the a's, b's,0°S and p's are constants. ‘Asimple example ofa system which may be expressed in the form (7.11) is provided by ay = 35, ay 22a ava sy, aoa : Introducing operator notation this system takes the form (@D ~3)x ~ 20) (20 +3)x + (2D +8) is clearly of the form (7.11), where Ly = 2D ~ 3, L; = ~2D, Ly = 2D + 3,and Returning now to the general system (7.11), we apply the operator L to the fist equation of (7.11) and the operator Ly tothe second equation of (7.11), obtaining Lalit + balay = bah Lyla + Laley = Lah We now subtract the second of these equations from the first. Since LyL,) = LaLy we obtain Lalyx a Ualsx = Lafi~ babs (abe = Labbe = bah = bah (712) ‘The expression Lyla — Labs in the left member of this equation is itself a linear differemtial operator with constant coeficionts. We assume that its neither zero nor 8 nonzero constant and denote it by Ly. If we further assume thatthe functions f; and f, are such that the right member Lf, ~ La of (7.12) exists, then this member is some Fanetion, say g,, oft. Then Equation (7.12) may be written Lex =o oy Equation (7.13) is linear differential equation with constant coeficients in the single dependent variable x. We thus observe that our procedure has eliminated the other ‘dependent variable j. We now solve the diferential equation (7.13) for x using the rethods developed in Chapter 4 Suppose Equation (7.13) is of order N. Then the general solution of (7.13) is ofthe form Xm eyth Heaty tooo cay > (n14)272 SYSTEMS OF UNEAR DIFFERENTIAL EQUATIONS where 1). us, ty af€ NV linearly independent solutions of the homogeneous linear equation Lex = 0,¢,,€3..-.,ey are arbitrary constants, and U, isa particular solution of Lyx = ay ‘We again return to the system (7.11) and this time apply the operators Ly and Ly to the fist and second equations, respectively, of the system, We obtain Lglix + Lylay = Lif Iybyx + Libay = Life Subtracting the frst of these from the second, we obtain (Lyla = kala)y = Lif Lah. Assuming that f, and f; are such that the right member L, fs ~ LsJ of this equation exists, we may express it as some function, say g,, of t. Then this equation may be written Lyon (745) where Ls denotes the operator LyL4— LaLs. Equation (7.15) isa linear differential equation with constant coefficients in the single dependent variable y. This time we have eliminated the dependent variable x, Solving the differential equation (7.15)for y, wwe obtain its general solution in the form 3 Ratt, + Rag bot katy + Uy 16 where uy, tg,-1..4y are the N linearly independent solutions of Lsy = O(or Lyx = 0) that already appear in (7.14), kj, kayo» ky are arbitrary constants, and Us is a particular solution of Lay = gs We thus see that if «and y satisfy the linear system (7.11), then x satisfies the single linear differential equation (7.13) and y satisfies the single linear differential equation (7.15)-Thusif x-and y satisfy the system (7.11), then xis ofthe form (7.14) and ys of the form (7.16). However, the pairs of functions given by (7.14) and (7.16) do not satsty the given system (7.11) for all choices of the constants ¢y,¢3y.--5¢y» ky, Kay---s ky. Thai, these pairs (7.14) and (7.16) do not simultaneously satisfy both equations of the given system (7.11) for arbitrary choices of the 2N constants ,,...,ys kis Kase Ry: I ‘other words, in order for x given by (7.14) and y given by (7.16) to satisfy the given system (7.11), the 2N constants c,,€25-.-ys kis Ka.--y ky eanmot all be independent Dut rather certain of them must be dependent on the others. It can be shown that the ‘number of independent constants inthe so-called general solution of the linear system (7.11)isequal tothe order of the operator LL, ~ LL, obtained from the determinant lt, La les Lal of the operator “coeficients”of x and yin (7.11), provided that this determinant is not zero, We have assumed that this operator is of order N. Thusin order for the pair (7.14) and (7.16) to satisty the system (7.11) only N of the 2N constants in this pair can be independent. The remaining N constants must depend upon the N that ace independent. In order to determine which of these 2V constants may be chosen as independent and how the remaining N then relate to the N so chosen, we must substitute x as given by (7.14) and y as given by (7.16) into the system (7.11) This de~ termines the relations that must exist among the constants cy, C2y---s€ys Kasay ky24 DIFFERENTIAL OPERATORS AND AN OPERATOR METHIOD 273 in order that the pair (7.14) and (7.16) constitute the so-called general solution of (7.11), Once this has been done, appropriate substitutions based on these relations are ‘made in (7.14) andor (7.16) and then the resulting paic (7.14) and (7.16) contain the required number N of arbitrary constants and so does indeed constitute the so-called general solution of system (7.11). ‘We now illustrate the above procedure with an example, > Example 7.3 Solve the system 4 oR axa a ax ae au aS arbaesy Mino opto tation and tether (QD -3x-2Dy=1, (2D + 3)x + (2D + Bly = 2. a8) ‘We apply the operator (2D + §) to the ist equation of (7.18) and the operator 2D to the second equation of (7.18), obtaining (2D + 81D ~ 3)x ~ QD + 8)2Dy = 2D + 8}, 2DQD + 3x + DED + 8)y = QD)2. Adding these wo equations, we obtain [0D + 8120 — 3) + 20D + 3)}x = (20 + e+ (20/2 (8D? + 16D — 24) =24 8140 of finally (42D - Semis (2.19) ‘The general solution of the difeential equation (7.19) is yet te HH (720) ‘Wenow return othe system(7.18)and apply the operator (2D + 3).othefistequation of (7.18) and the operator (2D ~ 3)to the second equation of 7.18. We obtain + QD4312D—3)x- QD + 3)2Dy = 2D +3}, (2D — 3)2D + 3)x + QD ~ 3)2D + 8)y = QD ~ 392, Subtracting the frst of these equations from the second, we have [0D ~ 3120 + 8) + 20 + 920]y = @D—3)2- D+ 3)" or (8D? + 16D — 24y =0- 6-2-3274 SYSTEMS OF LINEAR DIFFERENTIAL EQUATIONS or, finally, (0? +2D~sy=-he ray) ‘The general solution of the differential equation (7.21) is y tht 2) Thusifxand y sais the system (7.17), then xmust be of the form (7.20}and y must be of the form 7.22) for some choiee of the constants c,,¢3, ks. ks-The determinant of the operator “coefficients” of x and y in (7.18) is 10+ hae! 20 +3 20 +8| Since this is of order 2, the number of independent constants in the general solution of the system (7.17) must also be two. Thus in order forthe paic (7.20) and (7.22) to satisty the system (7.17) only two of the four constants cy,c2,,and k can beindependent. In order to determine the necessary relations that must exist among these constants, we substitute x as given by (120) and y as given by (7.22) into the system (7.17). Substituting into the fist equation of (7.17), we have [Reet 6eye"™ = §] — [Dkje! ~ 6kye"" +2] - [Beet + Senet HJ =e (rer = hide + (92 + 6hJer™ = 0 “Thus in order that the pair (7,20) and (7.22 satisfy the frst equation ofthe system{(7.17) we must have =e, - 2k =O, 9c, + by = 0. Substitution of x and y into the second equation of the system (7.17) will lead to relations equivalent to (7.23), Hence in order for the pair (7.20) and (7.22) to satisly the system (7.17), the relations (7.23) must be satisfied. Two of the four constants in (7.23) ‘must be chosen as independent. If we choose ¢, and cy as independent, then we have 723) ter and hy = dey ‘Using these values fork, and k, in (7.22), the resulting pair (720) and (7.22) constitute the general solution of the system (7.17). That is, the general solution of (7-17) is given by tere! t dee + at + te where c; and care arbitrary constants, If we had chosen k, and ky as the independent constants in (7.23), then the general solution of the system (7.17) would have been written wee Dye + Pye — He HB, yeketkhe +h+a| | | | i | i | \ | [ ! } 74. DIFERENTIAL OPERATORS AND AN OreRATOR METHOD 275 ‘An Alternative Procedure. Here we present an alternative procedure for selving a linear system of the form Lyx + Lay= filth Lyx + Lay = fil where L,, La, Ly, and L, ate linear differential operators with constant coeficints This alternative procedure begins in exactly the same way asthe procedure already described, Thats, we fst apply the operator 1, othe frst equation of (7.11) and the ‘operator Ls to the second equation of (711), obtaining Labyx + balay = Lafs Iglax + Lilay = Lah ‘We next subtract the second from the first, obtaining a) (Ly La ~ Lag) = Laf — Lahey a) Which, under the same assumptions as we previously made at this point, may be written Lex =a (143) ‘Then we solve this single linear differential equation with constant coefficients in the single dependent variable x. Assuming its order is, we obtain its general solution in the form meu team to bent + ay Where w,, Us,-.., uy are N linearly independent solutions of the homogeneous linear equation Ly¥ = O.e,¢2,..0.¢y ate N arbitrary constants, and U, is a particular solution of Lgx = gs, Up to this point, we have indeed proceeded just exactly as before. However, we now return to system (7.11) and attempt to eliminate from it all terms that involve the derivatives of the other dependent variable y. In other words, we attempt to obtain from system (7.11) 2 relation R that involves the still unknown y but none of the derivatives of y. This relation R will involve x and/or certain ofthe derivatives of x;but xis given by 7.14) and its derivatives can readily be found from (7.14). Finding these derivatives of x and substituting them and the known x itself into the relation R, we see that the result is mereiy a single linear algebraic equation in the one unknown . Solving it, we thus determine y without the need to find 7.15) and (7.16) or to relate the arbitrary constants. ‘As we shall ee, this alternative procedure always applies in an easy straightforward manner if the operators 11, Ly, Ls, and Ly are all of the first order. However, for systems involving one or more higher-orde operators, it is generally difficult 10 eliminate all the derivatives of y. We now give an explicit presentation of the procedure for finding y when Ly,L3,Lyy and L, are all first-order operators, Specifically, suppose Ly = aD +04, Ly = oD + bn Ly2%D +a, Le = BaD + Bs276 SYSTEMS OF LINEAR DIFERESTIN. EQUATIONS ‘Then (711) is (aD + a,)x + (BoD + By)y = fil, (oD + a,)x + oD + Bi) = LO. Multiplying the first equation of (7.24) by fy and the second by ~by and adding, we obtain L(eoBo ~ bota)P + (4B ~ by 2s) + (PaBo ~ BoBaby = Bolt) ~ Bolt) [Note that this involves y but none of the derivatives of y, From this, we at once obtain y= ota = aafe)Dx + (bos: — a Babs + Baht) — bo fat) (724 (7.23) assuming bo ~ bop # 0. Now xis given by (7.14) and Dx may be found from (7.14) by straightforward dilferentiation. Then substituting these known expressions for x and Dx into (7.25, we at once obtain y without the need of obtaining (7.15) and (7.16) and hence without having to determine any relations between constants c, and k= 1,2....N), asin the original procedure ‘We illustrate the alternative procedure by applying it tothe system of Example 7. > Example 7.4 Solve the system a Fig amas, 7) ax dy 25 +2 ant By of Example 7.3 by the alternative procedure that we have just described. Following this alternative procedure, we introduce operator notation and write the system (7.17) in the form (2-3) ~2Dy=1, (0D +3)x+2D+8y=2, 0.18) [Now we eliminate y, obtain the differential equation (0 +2D-3x—c4d 19) for x, and find its general solution xace qe td, 620) exactly as in Example 7.3 ‘We now proceed using the alternative method. We first obtain from (7.18) relation that involves the unknown y but not the derivative Dy. The system (7.18) of this exampleisso very simple that we do so by merely adding the equations (7.18), Doingso, we at once obtain 4De 4 By = 042,7. DIFFERENTIAL OPERATORS AND AN OreRATOR METHOD 277 which dos indeed involve y but no the derivative Dy, as desired, From this, we at once find ya e+ 2 40x, 026) ‘From (7.20), we find Dx = eye! —3eye 4 Substituting into (7.26), we get ale +2 dee! + Reve +9) mate tie bet ‘Thus the general solution ofthe system may be written eee tee! ahd, ye toe tie + beta, where ¢, and ¢, ate arbitrary constants, Exercises Use the operator method describe this section to find the general solution of ech of the following linear systems a gy “ a .a ax a z a ae ay a a a ax a dx 6 1% 1 a a ae a ax a ay ae Bay = sins, wo & dx ra278 sustems oF vent DIFFERENTIAL EQUATIONS de | dy ay u. 2842 tetsy Riga sty Wd. 1. 4 De + Dy 2207-2, a ad I, 2 a2 aye aes 16, dx dy Gitte ey ds Pad a & 9. S342 xty 24 Hh 2 ax dy Bay ata TY In each of Exercises 23-26, transform the single linear dferential equation of the form (76) into a system of first-order differential equations of the form (7.9). x jde @s dix dx ra a. Taye 4 42GB awe Be x al ge ictal dae ee 2, ater E—stao 16 SFT gtx cos 7.2 APPLICATIONS A. Applications to Mechanics Systems of linear differential equations originate in the mathematical formulation of ‘numerous problems in mechanics. We consider one such problem in the following example, Another mechanics problem leading to a linear system in given in the exercises at the end of this sectioneee aE EEE 7a xemucanons 279 > cxampte 7.5 On a smooth horizontal plane BC (for example, a smooth tabie top) an object A, is connected 10. fixed point P by a massless spring S, of natural length L,. An object 4, isthen connected to 4, by a massless spring S, of natural length L in such a way that the fixed point P and the centers of gravily A, and A, all lie ina straight line (see Figure 7.1, A A 1s & PYOOOOOOTIO Exrovvvctivoncel 2 : i © ! a i a 1 o, Os Figure 7.1 ‘The object 4, is then dispiaced a distance a, to the right or Zeft of its equilibrium position O,, the object 4, is displaced a distance a; to the right or let of its equilibrium position 0; ,and at time * = 0 the wo objects are released (sce Figure 7.2). What are the positions of the two objects at any time 1 > O? fener i Ss Sa a a Teer 7 c On Oo ee Formulation. We assume first that the plane BC is so smooth that fictional forces may be neglected, We also assume that no external forces act upon the system. Suppose ‘object 4; has mass m, and object 4; has mass m,. Further suppose spring S, has spring constant k, and spring S; has spring constant k, Let x, denote the displacement of 4, {rom its equilibrium position O, at timer > 0 ‘and assume that x, is positive when A, isto the right of O. [n like manner, let x) denote the displacement of 4, from its equilibrium position O, at time ¢ > 0 and ‘assume that x; is positive when 4280 SYSTEMS OF UNEAR DIFFERENTIAL EQUATIONS Consider the forces acting on 4, at time ¢ > 0. There are two such forces, F, and Fy where F, is exerted by spring 5, and F; is exerted by spring S,. By Hooke’s law (Section 5.1) the force F; is of magnitude k, |x| Since this force is exerted toward the [eft when A, is to the right of O, and towatd the right when A is tothe left of O,, we hhave F, = ~ ky x. Again using Hooke's la, the force Fis of magnitude k,s, where sis the elongation of Sat time r, Since s = {xy — x|, we See that the magnitude of F) is alsa — xl, Further, since this force is exerted toward the left when x; ~ x, < Oand. toward the right when xp ~ x, > O, we see that F Now applying Newton's second law (Section 3.2) to the abject 4y, we obtain the differential equation Px moe We now turn to the object 4; and consider the forces that act upon it at time 1 > 0, ‘There is one such force, F,, and this is exerted by spring S,. Applying Hoeke’s law once again, we observe that this force is also of magnitude kzs = ks|xy ~ x,|. Since Fy 8 exerted toward the let when x, ~ x; > Oand toward the right when x, — x, <0, we see that F, = —klx2 —,). Applying Newton's second law to the object 4, we obtain the differential equation yxy + kala — 4) oy (728) In addition to the differential equations (7.27) and (7.28), we see ftom the statement of the problem that the initial conditions are given by s1O= 4, 4 0)=0, [Oa 10) (729) ‘The mathematical formulation of the problem thus cotssts of the diferential equations (7.27) and (7.28) and the initial conditions (7.29). Weiting the differential equations in the form @x, im SE Oy + aden — hae ee th had 2, 030) ax mm Ga Rem thx ‘we set that they form & system of homogeneous linear differential equations with constant coefficients Solution of a Specitic Case. Rather than solve the general problem consisting of the system (7.20) and conditions (7.29), we shall carry through the solution in a particular case that was chosen to facilicate the work. Suppose the two objects A, and A, areeach of unit mass,so that m, = m, = |. Further, suppose that the springs S, and 5; have spring constants k = 3 and &, = 2, respectively. Also, we shall take a, = ~ 1 and Then the system (7.30) reduces 10 an Sey 2x =O, 2 31) O82 ax, + a ( \72 ammucanoss 281 and the initial conditions (7.29) become x= X4=% OM=2 O=0 (732) ‘Writing the system (731) in operator notation, we have (08 + 5)x,— 20, 2x, + (0? + 2)xp = 0. We apply the operator (D? +2) to the first equation of (7.3), multiply the second equation of (7.33) by 2, and add the two equations to obtain [0D +210" +5) 4x, = 0 (733) (Dt + 10? +x, = 0. (734) ‘The auxiliary equation corresponding to the fourth-order: mt + Im? +£6=0 or (m? + 6Km? + 1)=0. ‘Thus the general solution of the differential equation (7.34) is rential equation (7.4) is y= cain t+ cg 0081 +05 sin /Bt + cx con ft. 735) ‘We now multiply the frst equation of (7.33) by 2, apply the operator (D? + 5) to the second equation of (7.33), and add to obtain the differential equation (04 + 1D? + 6x, = 036) {or xz. The goneral solution of (7:36) is clearly ya hy sin + ky 6051+ ky sin fr + hy cos. / 37) ‘The determinant ofthe operator “coeicients” in the system (7.33) is +S -2 | ae ane ay pan? +146. Since this isa fourth-order operator, the general solution of (7.31) must contain four independent constants. We must substitute x, given by (7,35) and x, given by(7.37}into the equations ofthe system (7.31) to determine the relations that must exist among the constants <4,3:¢s, ca, ki, ks» ks, and ky in order that the pair (735) and (7.37) represent the general solution of (731). Substituting, we find that ka2y, b=20, betes, kee te. ‘Thus the general solution of the system (7.31) is given by pean e+ 6,008 865 Sin 6-4 0008/6, y= esi 4 2c, 008 — Jey sin fat ~ Jey cos ft ‘We now apply the initial conditions (7.32). Applying the conditions x, = dx {dt = Oat ¢ = 010 the first of the pair (7.38), we find Hlme ten ome, + Vb (738) (739)282 SYSTEMS OF UNEAR DIFFERENTIAL EQUATIONS Applying the conditions x; = 2, dx,/dt = Oat ¢ = Oto the second of the pair (7.38), we obtain (740) From Equations (7.38) and (740), we find that hoon gent ‘Thus the particular solution ofthe specific problem consisting ofthe system (731) and the conditions (732) is cos — feos y= feos + feos ft B. Applications to Electric Circuits In Section 5.6 we considered the application of differential equations toelectric circuits consisting ofa single closed path. A closed path in an electrical network i called a loop, We shall now consider electrical networks that consist of several loops. For example, consider the network shown in Figure 74 ‘This network consists of the three loops ABMNA, BIKMB, and ABJKMNA. Points such as Band M at which two or more circuits join are called junction points oF branch points. The direction of current flow has been arbitrarily assigned and indicated by arrows 1m oder to solve problems involving multiple loop networks we shall need two fundamental laws of circuit theory. One of these is Kirchhof’'s voltage law, which we hhave already stated and applied in Section 5.6, The other basic law that we shall employ isthe following: Kirchholt’s Current Law. Tn an electrical network the total current flowing into a junction point is equal to the total current flowing away from the junction point.7a arucanons 283 ‘As an application of these laws we consider the following problem dealing with the cireuit of Figure 7.4 > Example 7.6 Determine the currents in the electrical network of Figure 74 if Eis an electromotive force of 30 Y, Ry isa resistor of 10.0, Ra is a resistor of 209 Ly is an inductor of 0102 H, L isan inductor of 0.04 H, and the currents are initially zero, Formulation. ‘The current flowing in the branch MAB is denoted by i, that flowing on the branch BM by i,, and that flowing on the branch BJKM by is. We now apply Kirchhot’s voltage law (Section 5.6) to each of the three loops ABMNA, BIKMB, and ABJKMNA For the loop ABMNNA the voltage drops ate as follows: 1, Across the resistor Ry 10k 2 aos entry 002 ‘Thus applying the voltage law to the loop ABMINA, we have the equation ai, ont +10 2 oan forte bop KNB, lage dp tos: 1 Acs te ss 2 2 Ass itor 0042 3. Across the inductor Ly: ~002 4 ‘The minus sign entersinto 3 since we traverse the branch MB in the direction opposite to that of the current i, a5 we complete the loop BJKMB. Since the loop B/KMB ‘contains no electromotive foree, upon applying the voltage law to this loop we obtain the equation ai, a, 0.02 4 004% + 201, = 7. 0.02 Ft + 0045? + 20), = 0. (742) For the loop ABJK MIVA, the voltage drops are as follows: 1, Across the resistor Ry: 10 2. Across the resistor Re: 20%, ay 3, Across the inductor Ly: 0.04 52284 systeus oF unean DurreReNTIAL EQUATIONS Applying the voltage law to this loop, we obtain the equation sooo 4 a0 (4s We observe that she three equations (7.4), (742) and (7.43) are not all independent For example, we note that (7.42) may be obtained by subtracting (7.41) from (7.43). ‘Thus we need to retain only the two equations (741) and (7.43). ‘We now apply Kirchhofs current law to the junction point B. From this we seat once that imi ti. 44 In accordance with this we replace / by f, + fy in (7-61) and (743) and thus obtain the linear systern| 024 4 10%, + 10 = 30, 7 (745) 10i, +004? + 30), = 30, at Since the currents are initially zero, we have the initial conditions i) =0 and 4O=0 (746) Solution. We introduce operator notation and write the system (7.45) in the form: (02D 4 10)i, + 10%, = 30, 10;, + (0040 + 30}, = 30, We apply the operator (0.04D + 30) to the first equation of (7.47), multiply the second by 10, and subtract to obtain [10.040 + 30V(0.020 + 10) — 1004, a7) (0.04D + 30)30 — 300 (0.0008? + D + 200), = 600 or, finaly, (D* + 12500 + 280,000), = 750,000. (748) ‘We now solve the differential equation (7.48) fori. The auxiliary equation is rm? 4 3250m + 250,000 = 0 (m+ 250m + 1000) = 0. ‘Thus the complementary funetion of Equation (7.48) is 3. Hence the general solution of the ‘and a particular integral is obviously differential equation (7.48) is. by eye 9H ego 100 4 3, (749)| i | 72 amucaions 285 ‘Now returning to the system (7.47), we multiply the first equation of the system by 10, apply the operator (002 + 10) to the second equation, and subtract the first from the second. After simplifications we obtain the differential equation (D? + 12800 + 250000); = 0 for ip. The general solution ofthis differential equation is clearly ip = bye 284 hye 100 050) Since the determinant of the operator “coefficients” in the system (7.47) i a second- order operator, the general solution of the system (745) must contain two independent constants. We must substitute, given by (749) and i, given by 7.50) into the equations of the system (7.45) to determine the relations that must exist among the constants xs Cay kr» ky i rder that the pair (7.49) and (7.50) represent the general solution of (745), Substituting, we find that k= hey cs asi) ‘Thus the general solution of the system (7.45) is given by e775 + ce 1000+ 3, 006, 132) Hee? + exe Now applying the inital conditions (7.46), we find thatc, + ca + 3 = Oand —4ey + x= Oand hence c; = —2.and cy = —1. Thus the solution ofthe linear system (7.45) that satisfies the conditions (7.46) is iy = = 247280" — @ 10004 3, fy e729 100, Finally, using (7.44) we find that a e180 29100014 3 We observe that the current is rapidly approaches zero. On the other hand, the currents, i, and i= iy + is rapidly approach the value 3 C. Applications to Mixture Problems Jn Section 3.3C, we considered mixture involving the amount of a substance $ in & ‘mixture in a single container, nto and from which there flowed mixtures containing S. Here we extend the problem to situations involving wo containers. Thats, we consider a substance $ in mixtures in two interconnected containers, into and from which there ow mixtures containing S, The mixture in each container is kept uniform by stirring; and we seek to determine the amount of substance S present in each container at time t Let x denote the amount of $ present in tank X at time , and let y denote the amount of S present in tank Y at time 1. We then apply the basic equation (3.55) of Section 3.3C in the ease of each of the unknowns x and y. Now, however, the “IN” and “OUT” terms in each of the resulting equations also depends on just how the two containers are interconnected. We illustrate with the following example,286 SYSTEMS OF LINEAR OIFERENTIAL EQUATIONS Example 7.7 ‘Two tanks X and ¥ are interconnected (see Figure 7.5). Tank X initially contains 100 liters of brine in which there is dissolved 5 kg of salt, and tank ¥ initially contains 100 liters of brine in which theres dissolved 2 kg of saltStarting at time = 0,(1) pure water flows into tank X atthe rate of 6 liters/min, (2) brine flows from tank X into tank YY at the rate of 8 iters/min, (3) brine is pumped from tank Y back into tank X at the rate of 2liters/min, and (4) brine flows out of tank Y and away from the system at the rate of 6 lters/min. The mixture in each tank is kept uniform by stirring. How much salt is in each tank at any time t > 0? 2 rs veaemin Toa Tae ¥ Figure 7. Formulation. Let x= the amount of salt in tank X at time t, and let y = the ‘amount of salt in tank Y at time 1, each measured in kilograms, Bach of these tanks initially contains 100 liters of Nui, and fluid flows both in and out of each tank at the same rate, 8 lters/min, so each tank always contains 100 liters of fluid. Thus the concentration of salt at time 1 in tank X is x/100 (kgyliter) and that in tank Y is _y/100 (kgfliters}. ‘The only salt entering tank X is in the brine that is pumped from tank Y back into tank X. Since this enters at the rate of 2liters/min and contains y/100 keyliter, the rate at which salt enters tank X is 2y/100. Similarly, the only salt leaving tank X isin the brine that flows from tank X into tank Y. Since this leaves at the rate of 8 liters/min and ‘contains x/100 ka/liter the ate at which salt leaves tank X is Sx/100, Thus we obtain the differential equation ax i (153) for the amount of salt in tank X at time ¢. In a similar way, we obtain the differentiat equation dy_& 8 oy eee er 0, adr ~ 100” 100 (58 for the amount of salt in tank Y at time Since initially there was 5 kg of salt in tank X and 2 kg in tank ¥, we have the initia} conditions x)= 5, (0) =2 of Thus we have the linear system consisting of differential equations (7.53) and (7.54) initial conditions (7.55)72 seucanons 287 Solution. We introduce operator notation and write the differential equations (753) and (7.54) in the forms 756) ‘We apply the operator (D + rfp tothe first equation of (7.56, multiply the second equation by rg, and add to obtain [(e+5)(2 +5) Which quickly reduces to [ers theraihefeeo om We now solve the homogeneous differential equation (7.57) for x. The auxiliary ccquation is m+ do met oo, 106” * (jog? = (ma) m+ a 100 /\" * 109, with eal distinct roots (—1)/25 and (~3)/25. Thus the general solution of equation Osis x= ce + ce OAM, 3) ‘Now applying the so-called alternative procedure of Section 7.1C, we obtain from system (7.56) a relation that involves the unknown y but not the derivative Dy. The system (7.56 isso especially simple that the first equation of this system is itselfsuch a relation. Solving this for y, we at once obtain y= S0Dx + 4x. 7.59) From (7.58), we find pro Se Substituting into (7.59), we get y= 2eje- OAS — 200-088 ‘Thus the general solution of the system (7.56) is wm eye OH 4 eye OSH, ya dee OPS" Doge 0%, 69)2B8 srstems oF uineaR otreReNTIAL EQUATIONS ‘We now apply the initial conditions (75). We at once obtain ates, | 2c, De = 2, from which we find ash ead ‘Thus the solution ofthe linear system (7.56) that satisfies the conditions (7.55) is w= Be UHH 4 29 01256, y= be H8" ger ‘These expressions give the amount ofa xin tak Xan the moun yin tank Y, | respetvy each measured in lograms a an tne mn) > Tht, fr example 10-25 min wend | x= eT De = 1.208 (he), y= 60° * — 40-4 2.008 (ke | Notethatas¢-> 2, both xand y ~ 0. Thusisin accordance with the fact that nosalt at all (but only pure water lows into the system from outside | | Exercises 1. Sots the problem of Example 75 forthe cas in which the object Aas mass tm 2 tobe Ay ha mess my = the ping hn sping cortan fy = | the wring 8 ha sping constant by = andthe condions ne, feo x0) sand 0) | 2. Aprowctit mas mised itch om gtd a aan mike brassy lspps be malty he poet ‘ond Regeln octyl pny an ea sane an ue ‘ah ts fe pel nee en te sey (a) Taking the origin atthe position ofthe gun, withthe «axis horizontal andthe _yaxis vertical, show thatthe diferential equations ofthe resulting motion are (©) Find the solution of the system of diflerential equations of part (a). i 3. Determine the curzefits in the electrical network of Figure 7.6 if Eis an electromotive force of 100 V, R, isa resistor of 20.0, Ry isa resistor of 40.0, Ly is i an inductor of 001 H, £. is'an inductor of 0.02 H, and the eurrents are intially | 210.5, Two tanks are interconnect 6. Two tanks X and ¥ are interconnected, Tank X initially contains 30 liters of bit 72 npeucaons 289 bs by 0000 ———. 0007 afl i Ri Rs MA on Figure 7.6 4. Set up differential equations for the currents in each of the electrical networks shown in Figure 77, Do not solve the equations. (@) For the network in Figure 7.7a assume that £ is an electromotive force of 15, Risa resistor of 200, Lis an inductor of 0.02 Hand Cisa capacitor of 10"* farads, (®) For the network in Figure 7.76 assume that E is an electromotive force of 100 sin 1301 V,R, is a resistor of 200, Ry isa resistor of 300, and Lis an inductor of 0.05 H. (©) For the network of Figure 7.7c assume that E is an electromotive force of 100 V, R, isa resistor of 200, Risa resistor of 10.0, C, isa capacitor of 10+ farads, and C; is a capacitor of 2 x 10°* farads. R & ey m BL = T Co @ o © Figure 7.7 “Tank X initially contains 90 liters of brine in which there is dissolved 3 kg of salt, and tank Y initially contains 90 liters of brine in Which there is dissolved 2 kg of salt, Starting at time ¢ = 0, (1) pure water ows into ‘tapk X at the rate of 4.5 liters/min, (2) brine lows from tank X into tank Y at the rate of 6 liters/min, (3) brine is pumped from tank ¥ back into tank X atthe ate of 1 liters/min, and (8) brine flows out of tank Y and away from the system at the rate of 4 liters/min. The mixture in each tank is kept uniform by stirring. How much salt is in each tank at any time t > 07 in which there is dissolved 30 kg of salt, and tank ¥ initially contains 30 liters of pure water. Starting at time ¢ = 0, (1) brine containing 1 ke of salt per liter flows into tank X at the rate of 2 lites/min and pure water also flows into tank X at the290 srsteus oF uneaR onrERENTIAL EQLATIONS rate of 1liter/min, (2) brine flows from tank X into tank Y at the rate of 4liters/min, 3) brine is pumped from tank Y back into tank X at the rate of 1 liter/min, and (4) brine flows out of tank Y and away from the system at the rate | of 3 iters/min, The mixture in each tank is kept uniform by siring. How much} salt is in each tank at any time 1 > 0? | 7.3 BASIC THEORY OF LINEAR SYSTEMS IN NORMAL FORM: TWO EQUATIONS IN TWO UNKNOWN FUNCTIONS A. Introduction i ‘Weshall begin by considering basic type of system of two linear diferentalequations in two unknown functions. This sytem sof the form é B max + arly + Fld, rs Ba ale + any + 00. real interval a < t < 6. 1f F, (0) and Fy(@)arezero for all, then the system (7.61) is called ' > Example 7.8 \. ‘The system | fy | qa -y | (76) i j js homogeneous; the system | 7.6) Js nonhomogeneous. DEFINITION ‘By a solution of the system (7.61) we shall mean an ordered patrof real functions Lo, tu \272, ASC THEORY OF LNeAR sesTEMS IN NORMAL FoNM 291 ‘each hoving a continuous derivative on the real interval a < t example 7.9 “The ordered pair of functions defined for all by (e*, ~ 3e*), which we denote by 3, (7166) isa solution of the system (7.62). That is, yn deh, (766) simultaneously satisfy both equations ofthe system (7,62) Let us verify this by directly substituting (7.66) into (7.62), We have 4 65 = 26) — (365 Here Wer — (305), | } Be) = He) + 6(—3e%), So = Det 4+ 30, = Ise = Je — 186292 SYSTEMS OF LINEAK DIFERENTIAL €QUATIONS ‘Hence (7.66 indeed a solution of the system (7.62). The reader should verify that the ordered pair of functions defined for all by (e™, ~e), which we denote by fs also a solution of the system (7.62) ‘We shall now proceed to survey the basic theory of linear systems. Weshall observea close analogy between this theory and that introduced in Section 4.1 for the single linear equation of higher order. Theorem 7.1 isthe basicexistence theorem dealing with the system (7.61). THEOREM 7.1 Hypothesis. Let the functions a,., ay21 Fy. dzi, zz, and Fy inthe system(7.61) all be ‘continuous on the interval a < t < b. Let tg be any point of the interoala < t Example 7.10 Let us consider the system (7.63). The continuity requirements of the bypothesis of ‘Theorem 7.! are satisfied on every closed interval a << b, Hence, given any point to and any two constants c, and c, there exists a unique solution x = f(0), y = ald)of the system (7.63) that satisfies the conditions f(Gq) = ct, olf) = ca. For example, there ‘exists one and only one solution x = /(9), » = g(t) such that /(2) = 5,g(2) = —7. 'B, Homogeneous Linear Systems We shall now assume that F (0) and F,(0)in the system (7.61) are both zero forall ¢ and ‘consider the basic theory of the resulting homogeneous Hinear system ax Fanart arr 4 aas(OX + a23(0).73 ASIC THEORY OF LINEAR SYSTEMS IN NORMAL FORM 293 ‘We shall sce that this theory is analogous o that ofthe single nth-order homogeneous linear differential equation presented in Section 4.1B. Our first result concerning the system (7.67) is the following. ‘THEOREM 7.2 Hypothesis. Let x= fl, A), and 88) y=, ym ant, be two solutions of the homogeneous linea system (7.67). Let e, and cy be two arblirary constans Conclusion. Then xae h(t eahl0h ym cial) + caaalts {salsa solution ofthe system (7.67), (789) DEFINITION ‘The solution (7.69) is called a linear combination of the solutions (7.68). This definition enables us to express Theorem 7-2 in the following alternative form THEOREM 7.2 RESTATED. Any linear combination of two solutions of the homogeneous linear system (7.67)isitself a solution of the system (7.67) > Example 7.11 We have already observed that xaet, x=e™ and eae yn ne™, are solutions of the homogeneous linear system (7.62). Theorem 7.2 tells us that xace™ + ege%, ya ee" — ene where ¢, and ¢; are arbitrary constants is also a solution of the system (7.62). For294 svsrews oF uneaR DIFFERENTIAL EQUATIONS: ‘example, if cy = 4nd cy = —2, we have the solution x= de — 20%, ym a I2e + 2e% DEFINITION Let x= fll x= flO and yr ath y= a1, be two solutions of the homogeneois linear system (7-67), These two solution are linearly Sependetontheneala $1 bY tee exst constants a, ROBOth 0 ch afl + afl) =, e910 + ¢202(0)=0, uel forall t such that a Example 7.12 ‘The solutions eae, a2, and ya Be, ym 665 (of the system (7.62) are linearly dependent on every interval a < t < b. For in thiscase74 BASIC THEORY OF LINEAR sySTEMS IN NORMAL FORM 295, eye" + 2ege" =0, Ane — 60,64 = 0, and clearly there exist constants ¢, and c,not both zero, such tha the conditions (7.72) hold ona st Example 7.14 Since oe ya 3%, are linearly independent solutions of the system (7.62, we may write the general solution of (7.62) inthe form xe cet +ee%, ye niue"—ce%, where y and cp are arbitrary constants. In order to state a useful criterion forthe linear independence of two solutions of system (7.67), we introduce the Wronskian of two solutions. Note that thsissimilas to, but different from, the now familiar Wronskian of two solutions of a single second- order linear equation, introduced in Section 41 DEFINITION Let 0 filth and Foal ym alts be two solutions of the homogeneous linear system (7.67). The determinant Ka FO late) a2) is called the Wronskian of these two solutions. We denote it by W(®). ‘We may now state the following useful eriterion for the linear independence of two solutions of system (7.67). 73)73 BASIC THEORY OF LINEAR SYSTEMS IN NORMAL foam — 297 ‘THEOREM 7.4 ‘Two solutions x= fills x= fll, and y= ail y= alt of the homogeneous linear system (7.67)are linearly independent onan interval a Example 7.15 Let us employ Theorem 7.4 to verify the linear independence of the solutions and. ya nde, pone of the system (7.62). We have wo 240 set on every closed interval a < t Example 7.16 “The student may verify that is a solution of the nonhomogencous system (7.63). The corresponding homogeneous system isthe system (7.62), and we have already seen that and 3%, pane ye are linearly independent solutions of this homogeneous system. Theorem 7.6 tells us,73 BASIC THEORY OF LINEAR SYSTEMS IN NORMAL FORM 299° for example, that y= 38 is a solution of the nonhomogeneous system (7.63). From the preceding definition we sce that the general solution of (7.63) may be written in the form 42th, xecet tee + Mth, 2 y= Beye — ee where ¢, and ¢; are arbitrary constants. Exercises 1. Consider the linear system a Rasy dy Fares. (@) Show that fe are solutions of this system, (©) Show that the two solutions of part (a) are linearly independent on every interval a < 1 < b; and write the general solution of the system. (6) Find the solution Wh y= a, ofthe system which is such that /(0)= 1 and g(@) = 2. Why is this solution ‘unique? Over what interval sit defined? 2 Consider the linear system Gaetan dy Bade ey. (@)_ Show that xn 3%, wees, and, y= 20%, y are solutions of this system,300 syste oF uncan ourcaexma eQuarions {b) Show that the two solutions of part (a) are linearly independent on every interval a <1 Example 7.17 dx Gm 6x —3y, 79) 2e+y.poeta 174 NoMOGENEOUS LINEAR systins wnt constANr cotrneceNTs 303. We assume a solution of the form (7.75): (780) ‘Substituting (7.80) into (7.79) we obtain Ale = be" — 3Be", Ble = 2Ae™ + Be, and this leads at once to the algebraic system -HA-3B=0, 2A+ 0-28 =0, tn in the unknown 2. For nontrivial solutions of this system we must have I; -3 21-47% Expanding this we obtain the characteristic equation P- 2+ 12=0. Solving this, we find the roots 2y = 3,23 = 4. Setting 2 = i, = 3 in (7.81), we obtain 34-38 =0, 2A ~ 2B =0. A simple nontrivial solution of this system is obviously A = B= 1, With these values of A, B, and A we find the nontrivial solution (782) We note that a different solution for A and B here (for instance, A = B = 2) would only lead to a solution which is linearly dependent of (and generally less simple than) solution (7.82 Now setting 2 = 2, = 4 in (7.81), we find 24-38 =0, 24-38 = A simple nontrivial solution ofthis system is 4 3, B = 2. Using these values of 4, B, and J we find the nontrivial solution (733) By Theorem 7.7 the solutions (782) and (7.83) are linearly independent (one may check304 sysrevs oF uivean DIFFERENTIAL EQUATIONS ‘this using Theorem 74) and the general solution of the system (7.79) may be writen xe ce + 3ee%, ya ee + 2eye%, where e, and cp are arbitrary constants, . Case 2. The Roots of the Characteristic Equation (7.78) are Conjugate Complex I the roots 4, and 4; of the characteristic equation (7.78) are the conjugate complex numbers a + bi and a ~ bi, then we still obtain two distinct solutions fom, em Agee, and (784) ya pterr™, = Ate”, ‘of the form (7.75), one corresponding to each of the complex roots. However, the solutions 7.84) are complex solutions. In order to obtain real solutions in this case we consider the fist ofthe two solutions (7.84) and proceed asfollows: We frst express the complex constants A? and BY in this solution in the forms Af = A, + id; and BY = B, + iB, where Ay, Az, By, and By are real. We then apply Euler's formula cf = cok 0+ sin 8 and expres the frst solution (7.84) in the form 25 (Ay + IAg)e(e0s be + Usin br), Y= (By + 1B,)e%(cos bt + isin by). Rewriting this, we have x= e*[(A, 008 br — Az sin bt) + H(A, 608 bu + A, sin Be), y= &*[(B, c0s br — B, sin br) + i{B, cos be + B, sin bn). Itcan be shown that a pair [/(0) + dh(t gs) + igalt)] of complex functions is a solution ofthe system (774) if and only if both the pai (f(), 90} consisting oftheir real parts andthe pair [/(0, ga(0)] consisting of their imaginary parts are solutions of (7.14). Thus both the real part x= eA, cos bt ~ A, sin bt, y= e*(B, cos bt — B, sin b, (785) (786) and the imaginary part x eM(Ay cos br + Ay sin br), y= eB, cos bt + By sin be), ce ofthe solution (7.85 of the system (774 are also solations of (7.74) Furthermore, the solutions (746) and (787) are linearly independent, We verify this by evaluating the ‘Wronskian determinant (7.73) for these solutions. We find Jer(ay cose — Ap sin Bt) eM(Ay cos be + Ay sin bt) €*(B, cos bt — By sin bt) e%(B, cos bt + By sin bt) O(A,By — A28)) 788) or a$$ a 7A HOMOGENEOUS LiNtAR sysrEMS wiTH Constant councents 305 Now, the constant Bf is a nonreal multiple of the constant Af. If we assume that A,B; ~ 4B, = 0, then it follows that BY isa real multiple of Af, which contradicts the result stated in the previous sentence. Thus 4,B;~ A,B, 40 and so the ‘Wronskian determinant ¥(1) in (7.8) is unequal to zero. Thus by Theorem 7.4 the solutions (7.86) and (7.87) are indeed linearly independent. Hence a linear combination of these wo real solutions provides the general solution ofthe system (7.74) in this case. ‘There is no need to consider the second of the two solutions (7.84). We summarize the above results in the following theorem: THEOREM 7.8 Hypothesis. The roots 2, and A; ofthe characteristic equation 7.78) assoclated with the system (7.74) are the conjugate complex mumbers a: bi. Conclusion, The system (7.74) has two real linearly independent solutions of the form x= eA, cos bt — A, sin bo), x= €M(Az 60s bt + A, sin Bt), and y= €%(B, cos br ~ B, sin by), y= (Bz cos be + B, sin bt), where Ay, Az, By, and By are definite real constants. The general solution of the system (7.74) may thus Be written Le (Ay 00s bt — A, sin bt) + (4; 008 bt + A, sin be)}, _¥ = e*[c\(B, cos be — B, sin bt) + ex(B, cos bt + By sin be], where’e, and c, are arbitrary constants D Bxample 7.18 & Fatt 2y, 789) a i Ba nsety. ‘We assume a solution of the form (7.73), xe det, } y= Be ) Substituting (7.90) nto (7.89) we obsain ! Ale# = 34e" + 2Be%, if Bie = —SAe" + Be", } and this leads at once to the algebraic system ; Q-a44 2820, 5 5441-28 =0, ea {306 SYSTEMS OF LOMEAR DIFERINTI. EQUATIONS in the unknown 2. For nontrivial solutions of this system we must have p-a 2 Poa 2-0 Expanding this, we obtain the characteristic equation P41 41320. ‘The roots of this equation are the conjugate complex numbers 2: 3, Setting 4 = 2 + 3iin (7:91), we obtain (1-394 428-0, 544 (-1- B= 0. AA simple nontrivial solution ofthis system is A = 2, ‘ve obtain the complex solution xe 20230, =34 3i, Using these values ye (14 329, of the given system (7:89), Using Euler's formula this takes the form x= e{(2 cos 31) + i2 sin 3}, y= €M{(—008 3t — 3 sin 34) + (3 o0s 3¢ — sin 30), ‘Since both the real and imaginary parts of this solution of system (7.89) ae themselves solutions of (7.89), we thus obtain the two real solutions x= 2e* e0s 3, yu ~e%(c0s 3+ 3 sin 31), 32) and x= 2e* sin 31, 3) = 2%(3.c08 34 ~ sin 30. Finally, since the two solutions (792) and (793} ae linearly independent we may write the general solution of the system (7.89) in the form xm 2eMe, 008 3+ ¢, sin 32), y= e[e((—00s 3¢ — 3 sin 3t) + 6303 cos 3r — sin 31)], where ¢, and ¢; are arbitrary constants 793) D. Case 3, The Roots of the Characteristic Equation (7.78) are Real and Equal Ifthe two roots ofthe characteristic equation (7.78) are teal and equal.it would appear that we could find only one solution of the form (7.75) Except inthe special subcase in which a, =, # 0,a, = b, = 0 (sce Exercise 33 at the end of ths section) thisis indeed true, In'general, how shall we then proceed to find a second, linearly independent solution? Recall the analogous situation in which the auxiliary equation corresponding\ 74 MOMOCEMOUS LINEAR sysTens wars CoNsTaNr comnCINTS 307 toa single nth-orderfinear equation has a double root. This would lead us to expect a second solution of the form x=Ate, y= Bee", However, the situation here is not quite so simple (see Exercise 34 at the end of this section), We must actually seek a second solution of the form x= (At + Ae", y= Bit + BM co ‘We shal illustrate this in Example 7.19. We first summarize Case 3 in the following theorem, ‘THEOREM 7.9 Hypothesis. The roots 4, and Jy of the characteristic equation (7.78) associated with the system|?.74) are real and equal. Let 1. denote their common value. Further assume that system (7.74) isnot such that ay = by # 0,4; = by = 0. Conclusion. The system (7.74) has two linearly independent solutions of the form xs Ae, xs (Att Ade, and ya Bee, y= (Bt BY where A, B, Ay, Az, B,, and B; are definite constants, A, and By are not both zero, and BA, = BIA. The general solution of the system (7.74) may thus be written Ac + elit + Ande, ym c Be" + (Bit + Baje*, where cx and cz are arbitrary constants xe > Example 7.19 Boas, aia 795) ‘We assume a solution ofthe form (7.75): 796) Substituting (7.96) into 7.95) we obtain Ale" Ble = Ae + 2BeH%, ide — Be,308 systems oF unean oirrenevTas EQUATIONS and this leads at once to the algebraic system : G-a4-B=0, A+(2—-A)B=0, (eau) in the unknown 3. For nontevial solutions ofthis system we must have Expanding this we obtain the characteristic equation P-G+9—0 798) G~ 3? =0. ‘Thus the characteristic equation (7.98) has the real and equal roots 3,3. Setting 4 = 3 in (797), we obtain A-B=0, A~B=0. AA simple nontrivial solution of this system being A = B = 1, we obtain the nontrivial solution (7.99) ‘of the given system (7.95), Since the roots of the characteristic equation are both equal to 3, we must seek a second solution of the form (7.94), with 4 = 3. That is, we must determine Ay, A2, By, and B, (with 4, and B, not both zero) such that XetA Ade”, ym (Be Be™, is a solution of the system (7.95). Substituting (7.100) into (7.95), we obtain (7.100) BA + 34g + Ade = AAC+ Ase” ~ (Bet BE, (BBE + 3B, + Ble = (Ayr + Ape" + Bt + Bye. ‘These equations reduce at once to (Ay ~ Bye + (Aa ~ Ay ~ Ba) = 0, o (ay = Bye (Ay = By — By) In order for these equations to be identities, we must have AWB, Ay~B, ‘Thus in order for (7.100) to be a solution of the system (7.95), the constants Ay. Ay, A= Ar By 0, A,— By ~ By =0. (7.108)7A HOMOGENEOUS LINEAR SYSTEMS Win CONSTANT CormriCients 309 B,, and By must be chosen to satisfy the equations (7.101). From the equations A, ~ B, =0, we see that A, = By. The other two equations of (7.101) show that 4; and B, must satisfy Ay By= Ay = By (7.102) ‘Wemay choose any convenient nonzero values or A; and B,. Wechoose A, = By = 1 ‘Then (7.102) reduces to A; ~ By = 1,and we can choose any convenient values for A, and B, that will satisfy this equation. We choose A, = 1, B, = 0. We are thus led tothe solution 7.103) By Thcorem 79 the solutions (7.99) and (7.103) are linearly independent. We may thus “write the general solution of the system (7:98) inthe form re + ext + De®, yecse™ exte™ where ¢, and c, are arbitrary constants, We note that a diflerent choice of nonzero values for Ay and B, in (7.102) andjor a 0 is sufficient, But not necessary, for the system to have two real linearly independent solutions of the form x= Ae 33, Suppose thatthe roots of the characteristic equation (7.78) of the system (7.74) afe real and equal; and let Adenote their common value, Also assume that the system (7.74) is such that a, = by ¥ 0 and as = b, = 0. Show that in this special sub- ‘case there exist two linearly independent solutions of the form (7.75). 34, Suppose thatthe roots of the characteristic equation (7.78) ofthe system (7.74) are real and equal; and let Adenote their common value, Also assume that the system (7.74)is not such that ay = b # Oand a, = b, = 0. Then show that there exists no ‘nontrivial solution of the form. xa Atel, y= Bet, Which is linearly independent of the “basic” solution of the form (7.75) 35. Referring to the conclusion of Theorem 7.9, show that B,/A, = B/A in the case under consideration, 7.5. MATRICES AND VECTORS A. The Most Basic Concepts ‘The study of matrices and vectors is a large and important subject, and an entire chapter the sizeof this one would be needed to present all of the most fundamental ‘concepts and results. Therefore, after defining a matrix, we shall introduce only those very special concepts and results that we shall need and usein this book. For the reader familiar with matrices and vectors, this section will be a very simple review of a few select topics. nthe other hand, for the reader unfamiliar with matrices and vectors, the detailed treatment here will provide just what is needed for an understanding of the ‘materia presented in this book. DEFINITIONS A matrix i defined to be a rectangular array heft 422 (On an Ba75 matmices ano vectors 313 of elements ay (i= 1, 2,....1, j= 1,2,...4), arranged in m (horizontal) cows and 1 (Certical) columns. The matrix is denoted by che boldface leter A, as indicated; and the element in its ith row and jth column, by ay, as suggested. We write A = (a), and call A fan m a matix, ‘We shall be concerned with the two following special sizes of matrices. 1A square matrix is a matrix for which the number of rows is the same as the number of columns. If the common number of rows and columns is m, we call the ‘matrix an n x m square matrix. We write aw|ts 22 o> Ga Jena 2. A vector (or column vector) isa matrix having just one column. If the vector has x rows and, of course, one column), we calli an n x 1 vector (or m x I column vector). We write ‘The elements of a matrix (and hence, in particular, of a vector) may be real numbers, real functions, real function values, or simply “variables.” We usually denote square ‘matrices by boldface Roman or Greek capita letters and vectors by boldface Roman or Greek lowercase letters. Let us give a few specific examples. ‘The matrix 13 6 i -2 0 4 ~s 7 5-3 2 4-1 3-6 A is a4 x 4 square matrix of real numbers; whereas ® defined by 7 #41 5 oo. s ) 10 m4 isa3 x 3 square matrix of real fanctions defined forall real. ‘The vector 3} 1 2 5 2)314 SYSTEMS OF NEAR DIFFERENTIAL EQUATIONS is 5 x t column vector of real numbers; the vector defined by em=| net +1 is a3 x 1 column vector of real functions defined forall real r; and the vector xy isa4 x 1 column vector in the four variables x1, x3, 3. Xe The elements of a vector are usvally called its components. Given an mx 1 column vector, the element in the ith row, for each f=, 2... m is then called its ith ‘component. For example, the third component of the column vector e illustrated above is 2, and its fourth component is 5 For any given positive integer n, them x 1 column vector with all components equal to ero is called the zero vector and is denoted by 0. Thus if n = 4, we have the 4 x 1 ero vector DEFINITION We say that two m x n matrices A = (a,) and B = (by) are equal if and only if each clement of one is equal to the corresponding elgment of the other. That i, A and B are equal f and only if a, = by fori = 1,2,...,mandj = ,2,....mIf Aand Bare equal, we write A= B, ‘Thus, for example, the two 3 x 3 square matrices 4 “(i a are equal if and only if M26 a= 5 a=) 2, and ayy —4 We then write A7s marnices ano-vectors 315, Likewise, the veotors sy 3 , 7 x=[B] aa c=] > \-6) are equal if and only if We then write x = 6 DEFINITION Addition of Matrices The sum af two m x m matrices A (byis defined to be the m x n matrix: C= (i), where cy = ay + by. fork 12,.-.4n, WenriteC = A+ B. We may describe the addition of matrices by saying that the sum of two me xm matrices is the m x n matrix obtained by adding element-by-element “Thus, for example, the sum of the two 3 x 3 square matrices jl 4 -s [-5 40 a-[s2 6} and B=[ot -1 3 \9 8 3 6 27 isthe 3 x 3 square matrix 1-5 444 —540) [-4 a -s' calor 2-1 o+3}=| 7 1 3 +6 842 347) \I5 10 10 WewriteC =A +B. Likewise, the sum of the vectors je 1) xii) =| M8) gp xtyep et hs ae vy ie seal x4] Ye ean DEFINITION. Multiplication by a Number The product of the m x matrix A matrix B = (by), where by = cay for i (a,) and the number cis defined to be the m x T,Qse.qmand j = 1,2,...ym Wewrite B= cA. We may describe the multiplication of an m x n matrix A by a number c by saying that the product so formed is the m x n matrix that results from multiplying each individual element of A by the number ‘Thus, for example, the product of the 3 x 3 square matrix 2-3 6 74 os 23116 systems oF unean DUFERENTIAL EQUATIONS by the number 3 the 3 x 3 square matrix 3-2 34-3) 3-6 6-9 18) Bals(-7 3-4 3--Hf=f-2 2 — 300 35 32 o 1 6 We write B= 3A. Likewise, the product of the vector % by the number 5 is the vector Sx, Sx 5x3 5x4 We write y = 5x DEFINITION Let X,.X2yeeosXqbemn X V vectors and let cy, €25--.s6q bem numbers. Then an element of the form EAR FED EO Cae isan x 1 vector called a linear combination of the vectors X,,Xz,.++s%m. bh Eh | a ae EG}G8-E3 4+1245-12 9 =[-248-15—15|=|-2 644-1040 0 is a linear combination of X,, x3, 3, and Xe75 mamuces ano vectoes 317 DEFINITION Let tol and x=[72 en Oa ey % be am n x n square matrix and an n x 1 vector respectively. Then the product Ax of the nx mmatrix A by the m x 1 vector x is defined tobe the nx Jvector Ok tak tt ORL axa[ G8 ta totam) 28s + aaa HF Ba [Note that Ax is a vector. If we denote it by y and write me tee In v=) shea ne be Ya Oy xy Haas Hot aay, Va = ape + ayaa to + aaa, Ya anny + aay tO + Oy that is, in general, for each i = 1,2,..., a, nem aos aan +t = Bay For example, if e 2x, — 4x + Tey y= Axa | 5x, + 3x; — 8x5 =3m + 6x, + %5, Beforeintroducing the next concept, we state and illustrate two useful results, leaving their proofs to the reader (see Exercise 7 at the end of this section).i 311B sroreus oF uneas OnsERENTIN EQUATIONS Result A. If A is anm xm square matrix and x and y aren x 1 column vectors, then A(x + y) = Ax + Ay. number, then | | Result 8. If A son nsqure mari x an mt column seco and cis ea) =. > Example 7.20 We illustrate Results A and B using the matrix | (i 4 fa “t)~ “bh | the vectors and the number ¢ = 3. Ilustrating Result A, we have | 1 a\ix ty, 3-2 5/\mtos/ sini taste) (ey + yb tls 2) +O + a) Bix, + yu) ~ Boxy + va) + Sy + 99) f 2x, + x2 + 3x5) 2y + va + 39s) | way 4a tay] ty tbe ty \Bui — Bee + Sx) By = 292 + Shy f2 UN~al (2 1 3 fy) f-1 4 iliss]+{-1 4 1][yJ=axeay. (SIEGE Nb) IMlystrating Result B, we have 2b 3) fim) $x, + 3x2 +95 | oes iiss “se 3] 3-2 s\n) | 95, ~ 64 158) Alex) dey ta +34 ++ a} (Ax, bx, = Bey + Sx,RAT 75 mammices AND vectors. 319 ‘We have seen examples of vectors whose components are numbers and also of ‘vectors whose components are real functions defined on an interval [a, 6], We now distinguish between these two types by means of the following definitions: DEFINITIONS 1. A vector all of whose components are numbers is called a constant vector. 2. A vector all of whose components are real functions defined on an interoal (0,6) is called a vector function. DEFINITIONS Let @ be the m x 1 vector fiction defined by 4.00 oo =| 4.00] forall on areal intereal (a,b). () Suppose the components y(t), dal... lt)are differentiable On [ab], Then the erivative of @ is the vector function defined by do, (2)) a do,(0) cow |G" a 4g,{0) dt for all tea, 8). @) Suppose the components 64(0, dy(th-..,d,(t)are integrable on{a, band let tard te (a, bl,where tg < &. Then the integral of 4 fromtatotis the wector function defined by . 4, [lima i oo f (0) du ‘Thus the derivative of a given vector function all of whose components are diffeemtiable is the vector function obtained from the given vector function by320 sustems oF untar oitfeRETIAL EQUATIONS liflerentiating each component ofthe given vector function. Likewise, theintegral from {fg t01 of a given vector function, all of whose components are integrable on the given ‘interval, isthe vector function obtained from the given vector function by integr each component of the given vector function from fo tof. > Example 7.21 ‘The derivative of the vector function ¢ defined for all by 4e $00 =| 27 30 ae isthe vector function defined forall by ne 400 [aes a ‘ \ 6 ‘Also, the integral of from Oto ¢ isthe vector function defined forall by if wa | fa [leena=| [aut any au| =} 30 f re du He" — 0 Exercises w aa(2 3) at a-($ 2 () A=/-1 -4 3 @ a=[-2 6 ) 2. Imeach case, find the product cA of the given matrix A and the number 1 wan 2 « 375 marmess ano vecrons 327 1-3 3) @ A={ 6 -2 0], cm ~4 3012 S12 ( A=[4 -3 -2}, c=-3 0 3-6 In each case, find the indicated linear combination of the given vectors 3 a 2 a1 2 3 4 2 ma{th oaa| sh e=[gp and =(_5 4 -2| 5) 5 @) 2x +3x,- x5, 0) 38; 28, + Axe. (©) Wy + 5x) — 245 + Ine In each case find the product Ax of the given matrix A by the given vector x wold ef 2 -3 -3 7 2 @ a=[o aif xa[ a ETaaiies| -2 10 ~3 xbox @ Aa=[ 2-5) 4), x=[x, 42m]. 30102, a8 IMlustrate Results A and B of this subsection using the matrix 3-1 2 s 4-3], -s 1 2, the vectors and the number ¢ = 4 In each case, find (i) the derivative and (i) the integral from 0 to 1 of the vector funetion 6 that is defined. se = [ote ae — se fa) (0322 _susrews oF uneas overERENTIAL EQUATIONS © 40 ra | sin 3e cos 3: tsin 3¢ 1608 34 1 Prove Results A and B ofthe text (page 318). © = 8. Matrix Multiplication and Inversion DEFINITION Multiplication of Matrices Let A = (ay) be an m x n matrix and B = (by) be an n x p matrix. The product AB is defined to be the m x p matrix C = (eq), where ey = ub + aby ++ aby = (7.104) for alli =1,2,...,mand all) = 1, 2,....p. We write © Concerning this definition, we first note that A has m rows and n columns, B has n rows and p columns, and the product C = AB has m rows (the same number as A has) and pcolumns the same number as Bhas). Next observe that the numberof columns: ‘Aisequal to thenumber of rows in B. Matrix multiplication is defined only when this is the ease, that is, only when the nummber of columns in the first matrix is the same as the ‘number of rows in the second. Then the element c, inthe ith row and jth coluran of the product matrix C = ABs the sum of the n products that are obtained by multiplying ‘each element in the ith row of A by the corresponding element inthe jth column of B. > example 7.22 a qe na[b 3) we 0-2-9 [iow (d 3 err eee ie eee meyer Meet ee onateueetat eran chomping nce diye see rd ae eee ee mot ee wl LetSeagate eee) 75. mamnices ano vectors 323 (DO +Q(-9 (2) + HB) (-0(-2) +810) H+) H)+(NG) 2) + (510), 77 =|-7 7 no ‘We now find the 2 x 2 matrix BA: 21 422 ma -( yin wa) -( (4)2) + Q=1) + (-204) a +9 + ASN 6 (F 0 (12+ K-04 OF 10+ HQ+ Os) "L-s 8 For the matrices A and B of Example 7.22, we note that although the products AB and BA are both defined, they are not of the same size and hence are necessarily not equal. Now suppose two matrices A and B are such that AB and BA are both defined and of the same size. (Note that this will always be the case when A and B themselves are both n x n square matrices of the same size) We are led to inquire whether or not ‘ABand BA are necessarily equal in such a case. Te following example shows that they are not: (io 242) + (19) 3-3. D> Example 7.23 21-1 1 3-2 A=| 43-2] and Belo -1 3 62 5, 2 1-1 We observe that A, B, AB, and BA are all 3} x 3 matrices. We find zt -Nf 3-2 ape{ 43 -alfa -1 3 -62 sflz 1-1 iD +4) + (92) HE) + N= D+ (0) Let C901) > GVA) + (= 292) (4B) + BN 1+ (—290) (- 911) + OK) + V2) (9) +EX—+ KH) (2-2) + (19) +(— 1) (4)(-2) + G9) + (2X1) (~6(—2)+ H+ OX-9) 4 4 0) e[2 073 R -15 3324 13 ~2) BAw[4 -1 3 24-4 (wr x4) + (—2K-6) (MI) + VG) + (—2K2) MR+(-D4)+ EX-9 )+(—DE+ GN) 22+ H+ (-1K- AHL + CG) + (-0@) ED + BK-2)4- 219) YD +EM-D+ BD QY+ W246, 2% 6 ~17) s[-47 3 43-9, ‘Thus, even though AB and BA are both of the same size, they are not equal. We write AB # BA in this case. The preceding example clearly illustrates the fact that matrix multiplication is not commutative. That is, we do not have AB = BA in general, even when A.and Band the two products AB and BA are all n x n square matrices of the same size. However, we point out chat matcix multiplication is associative, (ABC = ABO, and distributive, A(B+O=AB+AC, for any three matrices A, B, C for which the operations involved are defined. ‘Now consider an.» x'» square matrix A = (04). The principal diagonal of A is the diagonal of elements from the upper lft corner to the lower right corner of A; and the diagonal elementsof A are the set of elements that lie along this principal diagonal, that is, the elements 21,2, gn- NOW, for any given positive integer a, them x nsquare ‘matrix in which all the diagonal elements are one and all the other elements are zer0 is called the identity matrix and is denoted by 1, Thats, = (ay), where ay = {for alld and ay = 0 for Ff {= 1, Qoooy WJ =, Droonn th Thus / 1 1+ 0 0 \o If Ais an arbitrary m x m square matrix and 1 is the n x » identity mateix, then it follows at once from the definition of matrix multiplication that AD=IA=A78 mamices ano vectors 325, ‘We now consider the following problem involving matrix multiplication: Given an ‘nx nsquare matrix A, we seek another m x square matrix Bsuch that AB = I, where Lis the identity matria. In order to discuss the existence of such a matrix B, we need the following definition. DEFINITION Let A be an m xn square matrix, The matrix A is called nonsingular f and only if ts determinant is unequal to zero:|A| vO. Otherwise, the matrix A is called singular, D> Example 7.24 The matrix Sate ine] 23 “aleaf [4 Ef ase -62 5 7 We are now in a positon to state the foliowing basic result concerning the existence of a matrix B such that AB Result C, Let A be an n x m square matrix, and let Ibe the m x midentity matrix. ‘Thenthere exists auniquen x n square matrix B such that AB = BA = Vif and onlyif A ‘is monsingular, that i, f and only if |Al #0. DEFINITION Let A be an n x nnonsingular matrix, and lt Ibe the m x n identity matrix. The unique nx n matrix B such that AB = BA = J is called the inverse of A. We denote the unique Inverse of A by A and thus write the defining relation between a given matrix A and its incense A“ in the form AA! AAO 7.405) Now suppose that A is a nonsingular n x m matrix, so that the existence and ‘uniqueness of its inverse A~? isassured. We now consider the question of finding A» ‘Several distinct methods are known, We shall introduce, illustrate, and use a method that involves the use of determinants. Although this method is not very efficient except ‘when n = 2or7 = 3,itwill be sufficiently useful for our purposesin this text In order to describe the procedure, several preliminary concepts will be introduced and illustrated.326 srsrens oF uunean purreReNTIAL EQUATIONS DEFINITIONS Let Abe an m x n maris, and let a be the element inthe ith row and jth cokanm of A. (1) The minor of ay is defined 10 be the determinant of the (n — 1) x (a — 1) matrix ‘obtained from & by deleting the ch row and jth column of A. We denote it by My (2) The cofactor of ays defined to be the minor My of a multiplied by the mumber (1 We denote it by Cy. Thus we have G, = (~1}"IMy. {8) The matrix of cofactors of the elements of A is defined tobe the matrix obtained Irom by replacing eachelement a, of A byits cofactor C= Is2y.oruR: j= Vous We denote it by cof A. ‘Thus the minor M, of ay is formally obtained simply by crossing out the row and column in which ay appears and finding the determinant of the resulting matrix that remains. Then the cofactor C,of ay is obtained simply by multiplying the minor My by +1 or ~1, depending upon whether the sum i+ j of the row and column of ay is respectively even or odd. > Example 7.25 Consider the 3 x 3 matrix A 21-8 432 62 5 Wefind the cofactor of the element a;3 ~ —2in the second row and thitd column of A. We first find the minor Mz, of a3y. By definition, this isthe determinant of the 2 x 2 matrix obtained from A by deleting the second row and third column of A. Clearly this Now, by definition, the cofactor Cz of az given by Cay (IP PMyy = ~ 10. In like manner, we find the cofactors ofthe other elements of A. Then, replacing each clement ay of A by its cofactor Cy, we obtain the matrix of cofactors of A, pal pa 3] 2s] “|-6 03 t= 2-1 aia welt yee] i -y - ey [3-2 ~2 la 3| 10 | (3 -3 26)75 marwices aND vectors 327. DEFINITION Let A = (a,) beanm x matrix, The transpose of A isthen x mmatrix B = (b,) where by = dygi = Uy Jy f= My 2-2-y™. That is the transpose of anm x nmatrix Aisthe ‘1 x mmatrix obtained from A by interchanging the rows and columns.of A. We denote the transpose of A by AT. For example, the transpose of the 3 x 3 matrix 21-0 A=| 43-2 ~62 5 ‘of Example 1.25 is the 3 3 matrix f2 4 -6 araf 1302 “1-2 5, ‘Note, in particular, that the transpose of the n x 1 column vector is the 1 x nrowyvector XT = Oy ar DEFINITION Let A be an n x n square matrix, and let cof A be the matrix of cofactors of A. The adjoint of A is defined to be the transpose of the matrix of cofactors of A. We denote the adjoint of A by adj A, and thus write adj A = (cof AI", > Example 7.26 In Example 7.25 we found that the matrix of cofactors of the matrix 2b -1 43-2 62 5) w -3 26 ath . ¢ 10 isthe matrix 10 2,328 SYSTEMS OF UNEAR OFERENTIL EQUATIONS ‘Thus the adjoint of A, being the transpose of cof A, is given by bp o-71 adjA=(cof Ah=[-8 4 0 2% —10 2 ‘We are now in a position to state the following important result giving a formula {or the inverse of a nonsingular matrix. Result D. Let A be ann x nonsingular matrix. Then the unique inverse A given by the formula of Ais 1 med X eae 7.106) where |Al is the determinant of A and adj A is the adjoint matrix of . Looking back over the preceding definitions, we observe that to find the inverse A of a given nonsingular matrix A, we proceed as follows (1) Replace each element a of A by its cofactor C, to fnd the matrix of cofactors cofAofA: (2) Take the transpose interchange the rows and columns) of the matrix ofA found in ste (1) o find the adjoint matrix adj A of A;and (3) Divide each element of the matrix adj A found in step (2) by the determinant | A{ of A. Ths gives the inverse matrix A > example 7.27 Consider again the 3 x 3 matrix 24-1 Ae{ 43 - 62 55 already considered in Examples 7.24, 7.25, and 7.26. In Example 7.24 we found that ils and thus noted that ince] A # 0, Ais nonsingular. Thus by Result C and the definition immediately following, we know that A has a unique inverse A~', Now by Result D we ta 4 adj a) = Had Art ay lad A) = Had A). In Example 7.25 we found 19-8 26) cof A=|-7 4 —10 eee Oe75 maces ano vicross 329 and then, in Example 7.26, we obtained want adjAn(otara|-3 40 26-10 2 Thus we find wore w -7y [B -TS Av = Hadj a)= 4 84 1 ol, ssi % - of {8 -$ 5 ‘The reader should now calculate the products AA“! and A~‘A and observe that both of these products are indeed the identity matrix 1 Just asin the special case of vectors, there are matrices whose elements are numbers land matrices whose elements are teal functions defined on an interval [a, 6]. We distinguish between these two types by means of the following definitions: DEFINITIONS: () A matrix al of whose components are numbers is called a constant matrix. (2) A matrix all of whose components are real functions defined on an interval (a, Bis calfed a matrix function. DEFINITION Let A defined by al) axle) aaa ‘agn| 2219 4 2 a eau) anf) dat) bbe an mx m matrix fiction whose elements a(t) (U= 1, 2yons Rs j= Ly eons m) are ‘differentiable on an intercal (a,b). Then Ais said tobe differentiable on [a6], and the derivative of Ais the matrix function defined by ais) ays) aia) AAO) _ fo a46) atat) ai : As) eal) daalt)] Jor allt €[a,330 sisrens oF uncat oureReNTAt CQUATIONS ‘Thus the derivative of a given matrix function all of whose ckments are differentiable is the matrix function obtained from the given matrix function by i Aliflerentiating each element of the given matrix function, | > example 7.28 ‘The derivative of the matrix function A defined forall ¢ by w(t 2) is the matrix function defined forall: by aaiy (61 ae | ae” (ise 60% We close this section by stating the following useful result on the dillerentiation of a product of diferentiable matrices. Note that the order of the factors in each product ‘cannot be reversed, since matrix multiplication is not commutative. Result E. Let A and B be differentiable matrices on [a,b]. Then the product ABis differentiable on (a, 6], and Bu) a ano, CAB] = Te Be + 40) forall re (a, 6} Fxercises In each of Exercises 1-10, given the matrices A and B, find the product AB. Aiso, find the product BA in each case in which it is defined. rae(h eG) 2 ane} (2 5) 2 3) wa deGt 3 3)75 martices aND vectors 331 1203 6 4 2) wo A=] 4 -1 -2|, Beals -1 ~3]. 1 0 4 02-4) 11, Givena square matrix A, wedefine A? = AA, A’ definition, find A? and A? if ‘= AAA, and so forth. Using this 12. Given the matrix A= 2 33) 1-21 -3 -1 0) find A? + 3A 4 21, where Tis the 3 x 3 identity matrix, 13 n s-(03) is ae( 123) 43) a4 i, Aw[d 1 2 254332 SYSTEMS OF UNEAR DIFFERENTIAL EQUATIONS 32 4) A 2 0) 2 aml 4 2 mm as[37 1} 2 13 3) 10 =n i$ 12 6) 23, a=(4 ' maa( i 5 — 2 3 7-10 g 25, Verify Result E for matrices A and B defined on an arbitrary real interval [a,b] as follows: Pad Ama |2 3c}, BO PoP a 3 i) et 2 tore . Linear Independence and Dependence Before proceeding, we state without proof the following two theorems from algebra. THEOREM A A system of m homogeneous linear algebraic equations in n unknowns has a nontrivial solution if and only if the determinant of coeficients of the system is equal to zero. ‘THEOREM B A systemof n linear algebraic equations in n unknowns has awnique soltion if and only i the determinant of coefficients of the system is unequal to zero. DEFINITION, A set of m constant vectors ¥y,¥3.--..%q is ineasly dependent if there exists a set of m Iumbers€5.y---1 6m Rt al of which are zero, such that es Feats tot Gat = 0. > Example 7.29 ‘The set of three constant vectors wf Eh of75 marnices AND vectors 333 is incarly dependent, since there exists the set of three numbers 2, 3, and —1, none of which are zero, such that Iv, +30 +(— ys =O. DEFINITION A set of m constant vectors is linearly independent if and only if the set isnot linearly dependent. That is, a set of m constant vectors ,,¥3,---,%q is linearly Independent f the relation cats Hag Ht egy =O implies that 20. > Example 7.30 ‘The set of three constant vectors Pent is linearly independent, For suppose we have cit teats tea =O aon dette This is equivalent to the system that is, amano ey + Bey + 2ey = 0, 7.108) tend of three homogencous linear algebraic equations in the three unknowns ,, ¢2,¢3. The determinant of coefficients of this system is 1-10 ietaa eal roo 140, ‘Thus by Theorem A, with n= 3, the system (7.108) has only the trivial solution = ©, = 65 = 0. Thus forthe three given constant vectors, the relation (7.107) implies =e = ¢5 = 0 and so these three vectors are indeed linearly independent.334 systems oF untan DareReNTiAL EQUATIONS DEFINITION The set of m vector functions $y. 3.-.-. bn is lineétly dependent on an interval S05 bff there exists a set of m numbers ¢,. C55... eq 04 all zero, such that C1610 + eabalt) + + exalt) = 0 Sor allt (0, bY D bxample 7.31 ‘Consider the set of three vector functions), ;, and 64, defined fr all by se" He |e respectively. This set of vector functions is linearly dependent on any interval a0 6, To see this, note that fei} col hal) ‘and hence there exists the set of three numbers 3, such that |,and —2, none of which are zero, 30.00 + (—NOs10 + (26510) = 0 forall t¢ (a, 6} DEFINITION ‘set of mvector functions is linearly independent on an interval if and only f the set is ‘nt early dependent on that inceroal. That i, a set of m vector functions. 83. is linearly independent on an interval a << b if the relation 10110 + e231 4° + cadbalt)= 0 Jor alt [a, 6] implies that eee en a8 > Example 7.32 Consider the set of two vector functions 6 and é,, defined forall: by $100 = () and da(0) = (3) respectively. We shall show that &, and @ are linearly independent on any interval 151s b, To do this, we assume the contrary; that i, we assume that , and, areSEERA 75 marnces ano vectors 335 linear dependent on a, 6). Then there exist numbers c, and ¢3, not both zero, such that crit + e631 = 0, for all © (a,b). Then eet ne" =0, ere! + Irae = 0, and multiplying cach equation through by e“Y, we have ete et leet forall © (a,b). This implies that + cae" = ¢; + 2cge! and hence 1 = 2, which is an ‘obvious contradiction, Thus the assumption that ©, and @ are linearly dependent on Ta, b] is false, and so these two vector functions are linearly independent on that interval Note. If a set of m vector functions $1, ,,..-. oy is linearly dependent on an interval act' Example 7.33 Consider the 2 « 2 square matrix (7)33B syste oF uvens ovrexenriat EQUATIONS ( Q)-() = -@)-G (@ )G@)-<) “Thisisof the form Ax = x, where A isthe given 2 x 2matix, aia characteristic vale ofthe given matrix A and x characteristic vector of A ‘On te other hand, we shall now show that = 2is nora characteristic valu ofthis matrix A. Fo, if it ere then there would exist a nonzero vector x = (2) suck that 6 -3\(n\ _ fs & YE)-2) Performing the indicated multiplications on each side of this equation and then quting the covesponding components we would have ‘Observe that and so t.and x = (3). Thus (3) is a corresponding 6x, — 3x; = 2x, 2x, x2 = 2s, or simply Since the determinant of coefficients of this homogeneous linear algebraic system is ‘unequal to 2er0, by Theorem A the only solution of the system is the trivial solution X, = X= 0. That is, we must have x=0, Which is @ contradiction, We proceed to solve the problem of determining the characteristic values and vectors of an n xm square matrix A. Suppose fay, iy oe ae Ae [O22 Oe75 matmicas ano victons 339 ‘Then Equation (7.109) may be written Jax, az. aul (xi) fx fs ap ~~ eel 82] fs), and hence, mukiplying the indicated entities, aes tanks + ae 2K, Out, ants tan) _ [dy ui; + Oana t+ dake) Ae Equating corresponding components ofthese two equal vectors, we have co O51 + Ogg Fo Bank = By Aye + aay H+ daa, OX + Oka +2 dandy = Uke and rewriting this, we obtain (ayy =a, + aun tot ayx,+ (3— At + dak (710) Oxy + Baka t+ Om — Ax, =. ‘Thus wesee that 7.109} holdsif and only if(7.{10)does. Now we are seeking nonzero ‘vectors x that satisfy (7.109). Thus a nonzero vector x satisfies (7.108) if and only if its set of components x,,X2,...,% is & nontrivial solution of (7.110). By Theorem A of, Section 7.5C, the system (7.110) has nontrivial solutions if and only ifits determinant of ‘coeficients is equal to zero, that is if and only if OH 20, qn er! Itiseasy tose that (7.1 11)isa polynomial equation of the nth degrecinthe unknown 2 In matrix notation itis writen |A- a=, where I is the n x m identity matrix (see Section 7.5B). Thus Equation (7.109) has a nonzero vector solution x for acertain value of 2 and only if Z satisfies the nth-degree polynomial equation (7.111} That is, the number 2 is a characteristic value of the340 SYSTEMS OF LINEAR DIFFERENTIAL {QUATIONS matrix A if and only if it satisfies this polynomial equation. We now designate this ‘equation and also state the alternative definition of characteristic value that we have thus obtained. DEFINITION Let A = (ay) beann x n square matrix of real numbers. The characteristic equation of Ais the nthedegree polynomial equation Jaya ae a ee Jao oan may eee in the unknown 3; and che characteristic values of A are the roots ofthis equation. Since the characteristic equation (7.111) of A is a polynomial equation of the mth dearer, it has n roots. These roots may be real or complex, but of course they may or -may not al be distinct. If certain repeated root occurs m times, where | < m < then ‘we say that that root has multiplicity m. If we count each nonrepeated root once and each repeated root according to its multiplicity, then we can say that them x m matrix A thas precisely characteristic valUues, $aY 2, 235..-s Ay Corresponding to cach characteristic value 2, of A there is a characteristic vector x k=, 2..., m) Further, if % is a characteristic vector of A corcesponding t0 characteristic value Jy, then So is cx, for any nonzero number c. We shall be concerned with the linear independence of the various characteristic vectors of A. Concerning this, we state the following two results without proof, Result F. Suppose each of the m characteristic values 2, ase. by of the mm square matrix A is distinct (shat is, nonrepeated); and let Xy, Xgyo-ny%y Be a 320 0f respective corresponding characteristic vectors of A. Then the set of these m charac= teristic vectors is lnearly independent. Result G. Supposethen x nsyuare matrix A hasa characteristic value of multiplicity ‘m, where 1 < m Example 7.34 Fin th characte aes and character ears fhe mati 12 (0) Solution, The characteristic equation of A is: a2 3" 22 Ealuating the determina inthe left member, we find tha hs equation may be wen nthe for B34 (2-4 +N =0. alues of Aare 4 and A= =1,342 ‘SYSTEMS OF UNEAR DIFERENTIAL EQUATIONS ‘The characteristic vectors x corresponding to a characteristic value Zare the nonzero vectors xe (E) svenanat ax as ( a2)-«@) With 4 = 4, this is ny tle Bie, + 2 thats, Bey = 2, Bx, = 2 ik is a solution for every real k. Hence the are the vectors. We see at once that x; = 2k, xy characteristic vectors corresponding to the characteristic value 2 2k 3k)’ where k is an arbitrary nonzero number. In particular, letting particular characteristic vector 2 3 corresponding tothe characteristic value 4 ~ 4 Proceeding in ike manner, we ean find the characteristic vectors corresponding to “1. With A= —1 and x = (i) the basic equation Ax = Axis G6 a2)-7@) a is a solution for every real k, Hence the characteristic vectors corresponding to the characteristic value A = — {are the vectors () mr dt te () 4 corresponding to the characteristic value 1, we obtain the 1, we obtain the75 matmices AND vectors 343. > Example 7.35 Find the characteristic values and characteristic vectors of the matrix 7-1 6 a=[-10 4 -2 rae Solution. The characteristic equation of A is 7-2-1 6 -10 4-4 12 eat =o a Evaluating the determinant in the left member, we find that this equation may be written inthe form 2 — 1037 + 314 - 30 =0. G-2A-3HA-H=0. ‘Thus the characteristic values of A are 4=2, An3, and A=5. ‘The characteristic vectors x corresponding to a characterictic value dare the nonzero vectors x=[x;] suchthat Ax =a With A = 2, thisis T-t 6\/x:) [x =10 4 -12]/x,)=2)0) —2 1 -tf\ss} Aas Performing the indicated multiplications and equating corresponding components, we see that x, x3, %) must be a nontrivial solution ofthe system Tey =p + 6xy = 2K, 10x, + da — 12K = 2a, A2ey ty xy = 285i that is, Sey — x, + 6x, =0, 10x, + 2xq ~ 1235 = 0, 2x +x; — 3x) 20. ‘Note thatthe second of these three equations is merely a constant multiple of the frst.344 SYSTEMS OF UMAR OUFFERENTIL EQUATIONS ‘Thus we seek nonzero numbers x1, x3. x3 that satisfy the first and third of these equations. Writing these two as equations in the unknowns x, and xy, we have xa + 6 wards 2x 5x, Solving for x3 and x3, we find spe oxy and xy oxy We see at once that x; = k, xy = —ky xy = —k is solution of this for every real k Hence the characteristic vectors corresponding to the characteristic value 4 = 2are the vectors wwe obtain the where k isa atbitcary nonzero number. fn particular, letting k Particular characteristic vector i -1 a corresponding to the characteristic value 4 = 2 Proceeding in like manner, one can find the characteristic vectors corresponding to 4 =3 and those corresponding to 2 = 5. We give only a few highlights of these ‘computations and leave the details tothe reader. We find that the components x. 3. x3 of the characteristic vectors corresponding to 1 = 3 must be a nontrivial solution of the system Ax, — xy + Oxy =O, 10x, +x; = 12x, = 0, Hin +x 4x3 From these we find that ays -2y and y= —xy and hence xy =k x; = —2k, xy = —k is a solution for every real k Hence the characteristic vectors corresponding to the characteristic value 2 = 3 are the vectors k —x|, ~k where k is an arbitrary nonzero number. In particular, letting k = 1, we obtain the particular characteristic vector [; ‘corresponding to the characteristic value 2 = 37s mamices ano vecrons 345, Finally we proceed to find the characteristic vectors corresponding to 4 = 5. We find that the components x, x3, ¥3 of these vectors must be a nontrivial solution of the system 2x, 2 + 6x 10x, — x, xy = 2x +2 — 6xy = 0. ‘From these we find that xp= 2x, and 3xy = =2x, We find that x; = 3k, x2 = —6k, x; = —2k satisfies this for every real k, Hence the characteristic vectors corresponding to the characteristic value A = $ are the vectors .-(3) where k is an arbitrary nonzero number. In particular, letting k particular characteristic vector 3 -6 -2, corresponding to the characteristic value 4 = 5, wwe obtain the Exercises - In each of Exercises 1-14 find al the characterise values and vectors ofthe mat 12 3? ©G)) > 6 i) -2 7 3% 3 4 Lv,346 srsTens oF uneak OW FERENTIAL EQUATIONS -2 6 -18) 4 [12-23 66 5-0 29 7.6 THE MATRIX METHOD FOR HOMOGENEOUS LINEAR SYSTEMS WITH CONSTANT COEFFICIENTS: TWO EQUATIONS IN TWO UNKNOWN FUNCTIONS A. Introduction ‘We now return to the homogeneous linear systems of Section 7.4 and obtain solu- tions using matrix methods. In anticipation of more general systems, we change ni tion and consider the homogeneous linear system in the form xy a wank baat, qi) asim tn Fak tasks ‘where the coefficients 4,4, a135 231, d33 are real constants. ‘We shall now proceed to express this system in a compact manner using vectors and. ‘matrices, We introduce the 2 x 2 constant matrix of real numbers, an(u@), ous) -() vm ‘Then by definition of the derivative of a vector, and the vector ax, a _[al, a" ae) a ‘and by multiplication of matrix by a vector, we have ‘Comparing the components of dx/dt with the left members of (7.112) and the components of Ax with the right members of (7.112), we see that system (7.112)can be expressed as the homogeneous linear vector differential equation as Snax ats)174 THE MATRIX METHOD FOR HOMOGENEOUS uNeAR svsTens 347 ‘The rel constant matrix A that appears in (7.115) and is defined by (7.113) is called the coeficient matrix of (7.19. ‘We seek solutions of the system (7.112), that is, of the corresponding vector iffrental equation (7.115). We proceed asin Section 7.4A, but now employing vector and matrix notation. We sock nontrivial solutions ofthe form. (7.116) ‘we see that the vector form of the desired solution (7.116) is x = ae™. Thus we seek solutions of the vector differential equation (7.115) of the form, xaae", in where a is a constant vector and 2 is a number. ‘Now substituting (7.117) into (7.115) we obtain Jae = Age” which reduces at once to Ageia 7.118) ‘nd hence to (A-dle = 0, where 1is the 2 x 2 identity matrix. Written out in terms of components, this is the system of two homogeneous tinear algebraic equations (ay, ~ Alay + @2%, = 0, 5144 + (ass — Hts = 0, 19) in the two unknowns a, and 2,. By Theorem A of Section 7.5C, this system has a ‘nontrivial solution if and only if 7.129) that is, in matrix notation, taal) 21) Looking back at Section 7.5D, we recognize (7.120) asthe characteristic equation of the coeficient matrix A = (ay) of the vector differential equation (7.115). Expanding the determinant in (7.120), we express the characteristic equation (7.120) as the quadratic equation HF (ayy + a43)A + (@ixda2 ~ 13d) = 0 (zy in the unknown 4, We recall that the roots 2, and 22 of this equation are the characteristic values of A. Substituting each characteristic value Jj, ((-= 1,2) into348 SYSTEMS OF UNEAR DIFFERENTIAL EQUATIONS system (7.119), o the equivalent vector equation (7.118), we obtain the corresponding. nontrivial solution (= 1,2), of (7.119). We recognize the vector al? as the characteristic vector corre sponding to the characteristic value 2, (= 1,2). ‘We thus see that if the vector differential equation dx a (rats has a solution of the form x=ae%, in) then the number 4 must be a characteristic value , of the coefficient matrix A and the vectora must bea characteristic vector” corresponding to this characteristic value 4, B. Case of Two Distinct Characteristic Values Suppose thatthe two characteristic values A, and of the coefficient matrix A of the vector differential equation (7.115) are distinct, and let a? and a! be a pair of respective corresponding characteristic vectors of A. Then the two distinct vector functions x and x! defined, respectively, by xm aMeds, xP = aot (7.123) ‘are solutions ofthe vector differential equation (7.115) on every real interval (a,b), We show this for the solution x" as follows: From (7.118), we have jal?) = Aa and using this and the definition (7.123) of x"), we obtain BRC Ly eltghit = Aged = Axl Se ayaledt = AaiMeht = Ax’, Which states that x4%(c) satisties the vector differential equation (7.115) on [4,b} Similarly, one shows that x isa solution of (7.115). ‘The Wronskian of solutions x" and x° is, aie 026% Wee, xHK0) titan By Result F of Section 7.5D, the characteristic vectors a” and a? are linearly independent. Therefore, using Exercise & at the end of Section 7.5C, we have le 803] 29, tan an “Then since 4% 4 O forall, we have W(x", x®)() # Ofor all on [a,b]. Thus by ‘Theorem 7.4 the solutions x") and x‘ of (7.115) defined by (7.123) are linearly74 THE MATRIX METHOD FOR HOMOGENEOUS UNEAR sstEMs 349) independent on (a, b]; and so a general solution is given by ex! + c3x! where cy ‘and c, are arbitrary constants. We summarize the results obtained in the following theorem. THEOREM 7.10 Consider the homogeneous linear system ax, 1s + dats a os iy Boa take thatthe sector diferent equation . (7.15) a a-(tt 2), and 5,412,051, 43 are real constant. Suppose the two characterise alues 2, and Ay of Aare distinct, and et a" anda be 4 pair of respective corresponding characteristic vectors of. ‘Then on every real interval, the vector functions defined by ae and ae forma linearly independent set of solutions of (7.115), and x= cale! + cyeletn, where where ¢, and c, are arbitrary constants, is a general solution of (7.115) om (a,b) > Example 7.36 Consider the homogencous linear system -%, a dry frees, that he vstor diferent equation ax (6-9 x" : Ho (6 3) were x-(%), (r25 “The characteristic equation ofthe coefficient matrix A =(§.~}) 8 wa-ay= [P57350 systems oF uncax O§FERENTIAL EQUATIONS Expanding the determinant and simplifying, this takes the form 2? — 74412 =0 with roots 4, = 3, 1, = 4, These are the characteristic values of A. They are distinct (and real), and s0 Theorem 7.10 applies. We thus proceed to find respective corresponding characteristic vectors a'” and a, We use (7.118) to do this. With Am iy = 3 and a = a"? =), (7.118) becomes 6 EC): from which we at once find that a, and a, must satisfy 6a, ~ 3a; = 3a, ay 2a, +a = 3a, nam A simple nontrivial solution of this system is obviously a, = 0, =, and thus a characteristic vector corresponding to i, = 3s of? “-() Then by Theor 710, xe((Je% ati ae (c). 7126 is a solution of (7.125). ), (7.118) becomes a 8 from which we at once find that 2, and a; must satisfy 64, — 3a; = 40, Day tay = da, A simple nontrivial solution of this system is obviously a, = 3, ay = 2, and thus a characteristic vector corresponding to i; = 4s, “0 2 (J awa, 2-2) om ‘Also by Theorem 7.10 the solutions (7.126) and (7.127) of (7.125) are linearly independent, and a general solution is va(2) 9) ‘Then by Theorem 7.10,7.6 THE MATRIX METHOD FOR HOMOGENEOUS UNEAR srstems 35.1 ‘where ¢, and c2 are arbitrary constants. That is, in scalar language, a general solution ‘of the homogeneous linear system (7.124) is. xrecet + dee, se + Deze, where c, and ¢, are arbitrary constants. ‘The system (7.124) ofthis example is, aside from notation, precisely the system (7.79) ‘of Example 7.17. A comparison of these two illustrations provides useful insight. Let us return to the homogeneous linear system (7.12) and further consider the ‘result stated in Theorem 7.10. In that theorem we stated that ifthe two characteristic values 2, and dof A are distinct and if and a are two respective corresponding characteristic vectors of A then the two functions defined by ae and leit ‘constitute a par of linearly indepesident solutions of (7.112) Note that although we assume that 2, and 2, are distinc, we do not require that they be rel. Thus the case of int complex characteristic values is aso included here. Since the coefficients in system (7.112) are real, if complex characteristic values exist then they must be a conjugate-complex par. Suppose 4, = a + biand A, = a— tare. par of conjugate-complex characteristic Yalues ofthe coeficient matrix A of system (7.112). Then the corresponding linearly independent soltions ae and een are complex solutions. In such a case, we may use Euler's formula e4 = cos 0 + isin @ to replace this linearly independent complex pait of solutions by a related linearly independent rea! pair of solutions. This is accomplished exactly a explained in Section AC and is illustrated in Example 7.18. C. Case of a Double Characteristic Value ‘Suppose that the two characteristic values 4, and A, of the coefficient matrix A of the Yector differential equation (7.115) are real-and equal. Let 2 denote this common characteristic value, and let a’be a corresponding characteristic vector. Then just asin the provious subsection, itis readily shown that the vector function defined by ae isa solution of (7.115) on every Fel interval [a,b]. But now, except inthe special subeasein ‘which a11 = 32 # 0,2 = dz1 = 0, there is only one solution ofthis form. Looking back at the results of Section 74D and expressing them in vector notation, we would now seek a second solution of the form a(t Be, 125 where 7 and are to be determined so that this is indeed a solution, We thus assume a solution of differential equation (7.115) of this form (7.128), diflerentiate, and substitute into (7.115}. We at once obtai (10+ Dydel + ye = Acre + Ble, viding through by e* O and collecting terms in powers oft, we readily find y= Ane+ OB+1— AB =0.| | 352 systems OF LNeAR DIFFERENTIAL EQUATIONS ‘This holds for all ref, b] if and only if dy Ar = 0, +1 AB=0. “he fist of these gives Ay = Ay or (A — Alyy = 0. Thos we see that 7 in facta characteristic vector a of A corresponding 10 characteristic value 2 The second of (7.129) with 7 = a gives AB — AB = a, from which we have (A- inp =e (2130) ‘as the equation for the determination of B in (7.128). Thus in the assumed solution (7.128), the vector 7 is, im fact, a characteristic vector of A corresponding to characteristic valuc 2 and the vecir Bis determined from (7.130, Diret substitution of (128) with these chotees of 1 and veces that it iginded a solution, Moreover, it can be shown that the two solutions thus obtained, ac* and (at + Ble, are linearly independent; therefore, linear combination of them constitutes a general solution of (7.115). We suramarize these results in the following theorem, (7.129) ‘THEOREM 7.11 Consider the homogeneous linear system 7 xy “Ta tak aa 2 om ak hate, a that is, the vector differential equation Beaax, oats) where and ays, a2, G34, 33 are real constants whick are not such that ay 1443 = ayy = 0. Suppose the two characteristic values A, and 2 of & are real and equal, dnd let J denote their common value. Leta be a corresponding characteristic vector of A and let B be a vector satisfying the equation (A-inp=a (7.130) ‘Then on every real interval, the vector functions defined by ae and (ar + Ble" (7.31) form a linearly independent set of solutions of (7-115); and x= cae" + cat + Ble", where and c, are arbitrary constants, isa general solution of (7.115) on [B]—_———_—— 74 TMEMATRIX METHOD FOR HOMOGENEOUS UNenR svsténs 353 > example 7.37 Consider the homogeneous linear system fe, a or) an sk ee iin gon apie , (0 “Ne ene 1=() cay Techn cantina a=(t 7) & mae | 1 2-4 Expanding the determinant and simplifying, this takes the form 4? ~ 64 + 9 = 0, with the double root 2 = 3. That i the characteristic values of are real and equal and so ‘Theorem 7.11 applies. Proceeding to apply it, we frst find a characteristic vector a correspon characteristic value 2 = 3, With 2 = 3 and a = (3), (7.118) becomes ¢ DQG): from which we at once find that a, and a, must satisfy to the da a may =a, or ay +24 = 3a, a=, A simple nontrivial solution of this system is obviously 1, =; = 1, and thus a characteristic vector corresponding to 4 = 3 is =-(i) x-("Jen aaa 2-2), om) ‘Then by Theorem 7.11, is a solution of (7.133). By Theorem 7.11, linearly independent solution is of the form (at + Ble™, where ), d= 3, and P satisfies (A — 20)B = a. Thus B = (f) satisfies [G26 DIG)-(: «SYSTEMS OF UNEAR DIFFERINTIAL EQUATIONS ( )@)-0) From this we at once find that fj and fs must satisty i BiB Bi B= ‘A simple nontrivial solution of this system is 8 = 1, 8, = 0, Thus we find the desired vector which quickly reduces to ‘Then by Theorem 7.11, “TC se 022) is a solution of (7.133) Finally, by the same theorem, solutions (7.134) and (7.135) are linearly independent, onal) eal) where c, and ¢, are arbitrary constants. That is, in scalar language, a general solution of the homogeneous linear system (7.132) is nya eye healt + De xz = ce + cate”, 7.35) where ¢, and ¢, are arbitrary constants, Exercises i Find the general solution of each ofthe homogeneous linear systems in Exercises 1-18, using the vector-matrix methods of this section, where in each exercise x-(2) | 2B oe AO | 0 2s77 THE MATRIX METHOD FOR HOMOGENEOUS tNeaR systems 355, ae (2-3 wo E-(5 3) 7.7 THE MATRIX METHOD FOR HOMOGENEOUS LINEAR SYSTEMS WITH CONSTANT COEFFICIENTS: EQUATIONS IN mn UNKNOWN FUNCTIONS A. Introduction In this section we extend the methods of the previous section fo a homogeneous linear system of n first-order differential equations in n unknown functions and having real ‘constant coefficients. More specifically we consider a homogeneous linear system of the form Pe ayn, bans Ge aise aaa &, meet teks tO naXar (7. GE = aay + nat too Hon 0.36) WX Fake EOE Oakes a where the coefficients ayy (0 1.2... mf = 1,2...) ate real constants ‘We proceed to express this System in vector-matrix notation. We introduce the sn constant matsik of real numbers net aa an a Janz and the vector ie 0.138)SYSTEMS OF LINEAR DIFFERENTIAL, EQUATIONS ‘Then by definition of the derivative of a vector, dx, and by multiplication of a matrix by a vector, we have Jays yy) fi) faim + ars b+ am age {tr 47nd] Pasvn taints t+ ate an On|] oats + aks #2 + Ot ‘Comparing the components of dx/dt with the left members of (7.136) and the components of Ax with the right members of (7.136), we se that system (7.136) can be expressed as the homogeneous linear vector differential equation & Ax. 139) de ‘The real constant matrix A that appears in (7.139) and is defined by (7.137) called the coefficient matrix of (7.139), DEFINITION By asolution of the system (7.136), that is, ofthe sector diferential equation 7.139), we ‘mean an | column-vector function | o-(" \%, whose components by. Bay. each have a continuous derivative on the real interval 5 ts byand which is suc that 46.00 de SEO atl) + saat) ++ OA, = B+ B28: + + B,C 2S as dyl0) 4 Oar dsl0) ++ ddl727 THE MATRIX METHOD FoR HOMOGENEOUS UNtAR sists 357. forall t such that a < t < b. In other words, the components 4, 63» ‘hat of are such x= 60) x2 = all) x= A) simultaneously satisfy all equations of the system (7.136) identically on a cample 7.38 Consider the homogeneous linear system Tey — aa + Oxy, 10x, + 4x3 — 12s, (7.150) Sts; +p — 8s that is, the veetor differential equation i, T-l 6 xy Gen [-10 4 12 ]x, where x= [x (nasty BEEN ear ee tee xy ‘Assuming a solution of (7.151) of the form ae, that is, wwe know that 1 must be a solution of the characteristic equation of the coefficient j7-1 4 A=[-10 4 -12 “201-1177 IME MATE METHOD FOR HOMOGENEOUS NEAR svtEMs 363 ‘This characteristic equation is -a 1 6 = 4-2 -2 epee eee la Expanding the determinant and simplifying, we se that the characteristic equation is 2 — 1077 +311-30=0, the factored form of which is =A HU 5) m0. ‘We thus see that the characteristic values of A are Ay=2 A=3 and Aas. ‘These are distinct {and real), and so Theorem 7.16 applies. We thus proceed to find characteristic vectors a,2,a corresponding respectively to 2,2, 2s. We use the defining equation Aa = ia (14g) to do this. For i= 4, = 2and defining equation (7.145) becomes 7 -t 6\/a\ — /a\ -10 4 -12]/a,)=2}a; -2 1 -1}\a} las Performing the indicated multiplications and equating corresponding components of this, we find Ta, — a+ 62-20, 10a, + 4g ~ 12a = 2a, Pt ay ay Dey Simplifying, we find that, #, a must satisty Sa — a+ 615 =0, 10a + 2ay — 1245 = 0, 152) =tay + ap — 3ay=0, ‘The second of these three equations is merely a constant multiple of the frst. Thus we seek nonzero numbers #, oy that satisfy the first and third of these equations. Writing these two as equations in the unknowns a and ay, we have nay +61, = Sa, ay —3ay— 2,364 SYSTEMS OF LINER DIFFERENTIAL EQUATIONS Solving for a and as, we find a=, and ay= a, A simple nontrivial solution of thisis 2, = 1,0 Thats, 2 = 1,43 = aires = ise simple nontrivial olution ofthe ystn (7152) Thus characteristic ‘vector corresponding t0 y Y ={-1 =1, ‘Then by Theorem 7.16, t el x= {-tle% thati, (-e|, ass, a a isan 180 For = dy =3and T= 6\fa\ fay -10 4 ~12][4,)=3fe, -2 1 tle) a) Performing the indicated multiplications and equating corresponding components of this and then simplifying, we find that 2, a, 25 must satisfy 4xy ay + 625 =0, 10s, + 25 ~ 125) =0, Ha, Fan 4 =0. (7.145) becomes From these we find that and a5 =~ AAsimple nontrivial solution of this ay = t,, = 2,4) = ~1. Thusa characteristic vector corresponding to Ay = 3 is v [ } ‘Then by Theorem 7.16, ete is a solution of (7.151).17.7 THE MATRIX METHOD FOR HOMOGENEOUS NEAR systems = 365, For a= dy = Sand (7.145) becomes To=1 GY fm) fa -10 4 ~12)/2,|=5/a, -2 4 -1f\asf (ss) Performing the indicated multiplications and equating corresponding components of this, and then simplifying, we find that ay, a, 25 must satisfy Day — m+ 6a, = 0, = 102, — 2; — 124, = 0, ay +a — Gay = 0, From these we find that a= —24 and 3a = ~2x. A simple noatrivial solution of this is = 3,2 = —6,a) = —2. Thus characteristic vector vorresponding to 1) = , 3 ae = |-6]e%, thatis, |—6e% i a 7159 isa solution of (7.151). ‘Also by Theorem 7.16, the solutions (7.153), (7.154), and (7.155) are linearly inde- pendent, and a general solution is reo) where cy, ¢2, and cy are arbitrary constants. That is, in scalar language, a general solution of the homogeneous linear system (7.150) is ‘Then by Theorem 7.16, aye gett ce + Beye, xa = —eye™ — 2ege™* — ere, xy ee — ce" — Joye", where ¢,, cz, and cy are arbitrary constants.366 SYSTEMS OF LINEAR DIFFERENTIAL EQUATIONS ‘We return to the homogeneous linear system (7.136), that is, the vector differential equation ® ax 74 Goes 139) Where A isan n x real constant matt, and reconsider the result stated in Theorem 7.16. n that theorem we stated thatif each ofthe n characteristic values yay... 4yof Ais distinct and if a", a",...,2! isa set ofn respective corresponding character vectors of A, then the m functions defined by ae, gers form a fundamental set of solutions of (7.139). Note that although we assume that 4,, Aayooon iy ate distinct, we do not require that they be real. Thus distinct complex characteristic values may be present. However, since A isa real matrix, any complex characteristic values must occur in conjugate pairs. Suppose 2, = a+ bi and 2, = ‘a — bi form such a pair. Then the corresponding solutions are alerwie and alien, and these solutions are complex solutions. Thus if one or more distinct conjugate- complex pairs of characteristic values occur, the fundamental set defined by a", i= 1, 2,....m, contains complex functions. However, in such a case, this fundamental set may be replaced by another fondamental se, all of whose members afe real func- tUons. This is accomplished exactly as explained in Section 7-4C and illustrated in Ex- ample 7.18. . Case of Repeated Characteristic Values We again consider the vector differential equation ax a x, (7139) where A isann x nseal constant matrix; ut here we give an introduction to the casein which A asa repeated characteristic value. To be definite, we suppose that A has areal characteristic value 2, of multiplicity m, where 1 r= SfstdnsaiAnEESy177 THE MATRIX METHOD FOR HOMOGENEOUS UNEAR systeMs 367° > Example 7.39 ‘Consider the homogeneous linear system #80 eam (so jai — ix: 13 fa fla ‘Assuming a solution of the form that is, wwe know that J must be a solution of the characteristic equation of the coefficient matrix 31-2 Ae{i3 1 33-1) “This characteristic equation is “1 -1 =a 3 |a-al= SHH Expanding the determinant and simplifying, we see that the characteristic equation is B-S2+8-4=0, the factored form of which is Gna MA-2)= We thus see that the characteristic values of A are a= A=? and ay =2 [Note that whereas the number Li ‘repeated characteris distinct characteristic value of A, the number 2is a value. We again use Aaa 4s) to find characteristic vectors corresponding to these characteristic values.368 systems oF uneak DurreReNTIAL EQUATIONS For 4 I,and (i) (3 eG) Performing the indicated multiplications and equating corresponding components of this and then simplifying, we find that 2,, x3, a) must be a nontrivial solution of the system (7.145) becomes 2a + a 1 =0, ay + xy y= 0, 3ay + 3x —2ay = 0. ‘One readily sees that such a solution i given by ‘Thus a characteristic vector corresponding to A “4 v e x= fife, matis, (:} 7.58) 3 e) is a solution of (7.157). ae -Equating corresponding coefficients of this and simplifying, we find that 02,4, must (7.145) becomes77. THE MATRIX METHOD FOR HOMOGENEOUS LINtAR svsTeMS 369 ‘bea nontrivial solution ofthe system ay + a — ay =0, a+ am a0, 34, +32; ~ 345 =0, Note that each of these three relations is equivalent to each of the other two, and so the only relationship among a, a3, 25 i that given most simply by a tea 0 7.160) Observe that ee) and |) a =0, ay= aretwo distinet solutions of ths relation (7.160), The corresponding vectors ofthe form (7.159) are thus. 1 Y a= |—1) anda =[0}, 9 1 respectively. Since each satisfies (7.145) with 1 = 2, each is a characteristic vector corresponding to the double root 4, = 4, = 2. Furthermore, using the definition of Sinear independence of a set of constant vectors, one sees that these vectors. and a) are linearly independent. Thus the characteristic value 4 = 2 of multiplicity m = 2 has the p = 2 linearly independent characteristic vectors (i-nt aa illustration of Subcase (1) of the di preceding this example. Thus corresponding tothe twofold charactcristic value J = 2, there are two linearly independent solutions of system (7.157) ofthe form ae. These ae a2" and ae, i= that is, 161) respectively.370 SYSTEMS OF LINEAR DIFFERENTIAL FQUATIONS [H(A £6 piven by (7.158) and (7.161) ae linearly independent, and a general solution is, The three solutions “lB where ¢,, cz. and cy are arbitrary constants, That is, in scalar language, a general solution of the homogeneous linear system (7.156) is specie tia tele’ xem eet — ee ays deel bee where c,, ca, and ¢5 are arbitrary numbers. One type of vector diferential equation (7.139) which aways leads to Subcase (1) =m. in the case of a repeated characteristic value 4, is that in which the m x n Coefficient matrix A of (7138) i a fel symmetric matex. For then, by Resut J of Section 75D, there always exist linearly independent characteristic vectors of A, regardless of whether them charactriatic values of A areal distinct or not Wenow turn toa consideration of Subcaso(2),p Example 7.40 Consider the homogeneous linear system ax, Bede t3n to, ax, $e 4x, 45-20, (473) dey Soom, + det 6m, or in matex form, eo eee el a4 4 —2)x, where x={s rm 82 6 28 solution ofthe form thats, we k Ww that 1 must be a solution of the characteristic equation of the coefficient a3 4-4-2]. 8 2 6 3 1 a4 4-4-2 8 2 b6HA matrix This characteristic equation is 4 jA-alj= Expanding the determinant and simplifying, we see that the characteristic equation is 2612 + 14-8 =0, the factored form of which is(2 — 2)* = 0, We thus sce thatthe characteristic values of Aare Aya ay = dyed That is, the number 2s a triple characteristic value of A. We again use Aa = Ja (745) to find the corresponding characteristic vector(s) a. With 4 = 2 and (7.175)i } } 7.7 THE MATRIX METHOD FOR HOMOGENEOUS UNEAR systems 373 (32 -t) Performing the indicated multiplications and equating corresponding componcats of this and then simplifying, we find that a,, #2, 43 must be a nontrivial solution of the system (7.145) becomes 2a + 3a + m= 0, 4a, ~ 64, — 24 = 0, 8x, + 12a + dx = 0. Each of these three relationships is equivalent to each of the other two, and so the only relationship among 2,, 23,5 is that given most simply by Day +3; +ay = 0. 7.176) Observe that ah =O a and =O aah ay are two distinct solutions of relation (7.176). The corresponding vectors of the form (7.175) are thus t | atte! 0) and a? |. 2, 3 respectively. Since each satisties (7.148) with A = 2, cach is @ characteristic vector corresponding to the triple characteristic value 2, Furthermore itiseasy to sez that the {wo characteristic vectors a” and a ar linearly independent, whereas every st of thrée characteristic vectors corresponding to characteristic value 2 are linearly dependent. Thus the characteristic value A= 2 of multiplicity m= 3 has the p = 2 linearly independent characteristic vectors ! ol ate! o] and a=] 1 am 2 3 ‘corresponding to it. Hence this is an illustration of the situation described in the paragraph immediately preceding this example. Thus corresponding to the triple ‘characteristic value 1 = 2there are two linearly independent solutions of system (7.173) of the form ae", These are ae and ae, that is, Gr=lh374 SYSTEMS OF LINEAR DIFFERENTIAL EQUATIONS “ 0 (:,) and Us} amy 7 ‘td Sston eveyone 0 = Zit eo (ar + Be, (1179) were sates An m=0 a and B satisfies (A 20B =a (718, Since both a" anda" given by (7.177 are characteristic vectors of A corresponding 10 A= 2, they both satisfy (7.180), But, as noted in the paragraph immediately preceding this example, we need to use the more general solution a= kal + kya? ‘of (7.180) in order to obtain & nontrivial solution for Bin (7.181) Thus we et ky (6: ankastate{ a, | and pm[), = 2k; = 3k Ps ‘and then (7.181) becomes 203 Nip) k “46a i) i 8 12 alps} \-2k - 3h) Performing the indicated multiplications and equating correspor this, we obtain ing components of, 2+ Bret Bye ky 4) ~ 66a ~28y= key 7.182) 8, + 12P, +40 = ~2, — 3k Observe that the left members of these three relations are all proportional to one another. Using any two of the relations, we find that k = —2k,. A simple nontrivial solution of this Jast relation is ky = 1, ky = —2. With this choice of k, and k, we find : bh oy Ta ao7.7 THE MATRIX METHOD FOR HOMOGENEOUS LINEAR systems 375, and the relations (7.182) become 2+ 3h+ A= 4, 6f, — 2B) = ~2, BB, 4 D+ 4B, = 4 Each of these is equivalent to 2B, +38 + Bs ‘A nontriviat solution of this is Br=Bs=0, By=t: 4 vs Therefore, with a given by (7.183) and given by (7.184), the third solution (7.179) is Sip ‘The three solutions defined by (7.178) and (7.185) are linearly independent, and a teneral solution is the linear combination te = de of eo{ Salool at of these three, where ¢,, ¢, ¢ afe arbitrary constants. That is, in component form, a ‘general solution of system (7.173) is and thus we obtain that is, (785) x= e+ ete”, Kye ce! — Deste™, xy = —2eye" —Aeze™ + eglde + Des, where ¢,,¢25¢5 are arbitrary constants. Find the general solution of each of the homogeneous linear systemsin Exercises 1-24, Where in each exerciseSYSTEMS OF LINEAR DIFFERENTIAL EQUATIONS 376— CHAPTER EIGHT —— Approximate Methods of Solving First-Order Equations 1m Chapter 2 we considered certain special types of first-order diferential equations having closed-form solutions that can be obtained exactly Fora first-order differential equation that i nol of one or ano er of these special typés, it usualy isnot apparent how one should proceed in an attempt to obtain a solution exactly. Indeed, in most such cases the discovery of an exact closedform solution in terms of elementary functions would bean unexpected luxury! Therefore one considers the possbltis of ‘obtaining approximate solutions of first-order dilferental equations. In thischapter we shall introduce several approxithate methods, In the study of each method in this chapter our primary concen will be to obtain familiarity withthe procedure ite and to develop skilin applyingt. tn general we shall not be concerned here with theoretical justifcations and extended discussions of accuracy and error. We shall lave such matters important as they are, to more specialized and advanced treaties and instead shall concentrate on the formal detail ofthe various procedures. 8.1 GRAPHICAL METHODS A. Line Elements and Direction Fields In Chapter 1 we considered briefly the geometric significance of the first-order diferential equation ay ax where f is a real function of x and y. The explicit solutions of (8.1) are certain real functions, and the graphs of these solution functions are curves in the xy plane called the integral curves of (8.1). Ateach point, y)at which f(x, y)isdefined, the differential equation (8.1) defines the slope f(x,y) atthe point (x,y) of the integral curve of (81) ‘that passes through this point. Thus we may construct the tangent to an integral curve 377 ey), ey378 [APPROXIMATE METHODS OF SOLVING FIRST-ORDER EQUATIONS of (8.1)at a given point (x, ») without actually knowing the solution function of which this integral curve is the graph. We proceed to do this. Through the point (x, ) we draw a short segment of the tangent to the integral curve of (81) that passes through this point. That is, through (s,3) we construct a short segment the slope of which is f(x, y) as given by the differential equation (8.1). Such a segment is called a line element of the differential ‘equation (8.1), For example, let us consider the differential equation ay Ramey. 2 Here f(x,y} = 2x + y, and the slope of the integral curve of (8.2) that passes through the point (1,2) has at this point the value JUD=4 ‘Thus through the point, 2)weconstruct a short segment of slope dor, in other words, of angle of inclination approximately 76° See Figur 6.1), This shor segments the line element of the differential equation (8.2) at the point (1, 2 It is tangent to the integral curve of (8.2) which passes through this point. Let us now return to the general equation (8.1). A line element of (8.1) ean be constructed at every point (x, ) at which fx, ) in (8.1) is defined. Doing so for a selection of different points (leads toa configuration of selected line elements that indicates the directions of the integral curves at the various selected points. We shall refer to such a configuration a a line element configuration. For each point (x) at which fx, y is defined, the diferential equation (8.1) thus defines a line segment with slope fx, Yh, oF, im other words, a direction. Each such point, taken together with the corresponding direction so defined, constitutes the so: called direction field of the diferential equation (8.1). We say that the differential equation (81) defines this direction field, and this diccetion field is represented graphically by a line clement configuration. Clearly a more thorough and carefully constructed line element configuration gives a more accurate graphical representatio of the direction fed Figure 8.1 Line element of differential equation (8.2) at point (1, 2). | } \ \a1 caaucat mernoos 379 For a given differential equation of the form (8.1), let us assume that a “thorough and carefully constructed” line clement configuration has been drawn. That is, we assume that fine clements have been carefully constructed at a relatively large number of carefully chosen points. Then this resulting line element configuration will indicate the presence of a family of curves tangent to the various line elements constructed at the «different points. This indicated family of curves is approximately the family of integral ‘curves of the given differential equation. Actual smooth curves drawn tangent to the line elements as the configuration indicates will tus provide approximate graphs of the true integral curve’. ‘Thos the construction of the line element configuration provides a procedure for approximately obtaining the solution ofthe differential equation in graphical form. We ‘ow summarize this basic graphical procedure and ilustrate it with a simple example. Summary of Basic Graphical Procedure 1, Carefully construct a line clement configuration, proceeding until the family of “approximate integral curves” begins to appear. 2. Draw smooth curves as indicated by the configuration constructed in Step L. > Example 8.1 ‘Construct a line clement configuration for the differential equation ay Fame, @2 and use this configuration to sketch the approximate integral curves Solution. ‘The slope of the exact integral curve of (&.2)at any point (x, y)is given by fen ya dey. We evaluate this slope at a number of selected points and so determine the approximate inclination of the corresponding line element at exch ovin« selected. We then construct the line elements so determined. From the resulting configuration we sketch several of the approximate integral curves, A few typical :nclinations are listed in Table 8.1 and the completed configuration with the approximate integral curves appears in Figure 8.2. TABLE 81 dyjdx Approximate inclination x lope) ef te clement areas 4 bua 4 0 1 45° ‘eee 36 4 1 2 oo Sees i 56 1b 3 o 1 $ a 68 snus >380 srrecxamate METHODS OF SOLVING FIRST-ORDER EQUATIONS Figure 8.2 procedure outlined here is very general since it can ‘be applied 10 any first-order differential equation of the form (8.1). However, the ‘method has several obvious disadvantages, For one thing, although it provides the approximate graphs of the integral curves it does not furnish analytic expressions for ‘the solutions, either exactly or approximately. Furthermore, itis extremely tedious and ‘time-consuming. Finally the graphs obtained are only approximations to the graphs ‘of the exact intogral curves, and the accuracy of these approximate graphs is uncertain. ‘OF course, apparently better approximations can be obtained by constructing more ‘completeand careful ine element configurations. but this in turn increases the time and labor involved. We shall now consider a procedure by which the process may be speeded up considerably. This isthe so-called method of isoclines. ( \ \ | uf Hi pees ele | 2 open « / A curce along which the slope f(s, y) has a constant value ¢ is called an isocline of 1caarmicat mernons 381 differential equation (8.1). That is, the isoclins of (6.1) are the curves fx, 3) = 6, for diferent values ofthe parameter c. For example, the isoclines of the differential equation 4) Famey 62 are the straight lines 2x + y = c. These are of course the straight lines y= —2x + cof slope —2 and y-intercept c. Caution. Note carefully that th soclines of the differential equation 8.) are nrin general integral curves of (81) An isoclne is merely a curve along which al ofthe Fine clments havea single, xed inclination. This is precisely why isoclines are seul Since theline elements along a given soclne all have the same inclination, a great umber of line elements can be constructed with ease and speed once the given soci is drawa and re line element hasbeen constructed uponit. Thisisexacty the procedure that we shall now outline. Method of isoctines Procedure 1, From the diferential equation a Ze sen @) determine the family of isoctines Seuy=o 63) and carefully construct several members ofthis family 2 Consider a particular isoctine f(x, y) = ¢9 of the family (8.3). Atall points(x, Jon this isoctine the fine elements have the same slope co and hence the same inclination ‘ty = arctan cy, 0" < a < 180°. Ata series of points along this isocline construct line elements having this inclination a, 3. Repeat Step 2for each ofthe isoclines of the family (8.3) constructed in Step 1. In this way the line element configuration begins to take shape, 4, Finally, draw the smooth curves (the approximate integral curves) indicated by the line element configuration obtained in Step 3. > Example 8.2 Employ the method of isoclines to sketch the approximate integral curves of 4 Pnwety. @ Solution. We have already noted that the isoclines of the dlferential equation (8.2) are the straight lines 2x + y = €or yr 2 te. ey382 arreonware METHODS OF SOLVING FIRST-ORDER EQUATIONS Figure 8.3 In Figure 8.3 weconsteuet these lines fore =1,-4.0,4,1.3,and2.Oneach of these we then construct a number of line elements having the appropriate inclination x arctan c,0" Example s.s Employ the method of isoclines to sketch the approximate integral cures of Pavey 85) Solution. The isoeines ofthe diferetial equation (8.5) ae the concent cteles 224 y? = e,0 0. In Figure 84 the cices for which ¢ = teva fee fi de #8 ad 4 have been drawn with dashes, and several line elements having the appropriate inclination have been drawn along cach. For example, for ¢ = 4, the corresponding isocineis the circle x? + y? = 4of radius 2, and along this ice the line elements have inclination arctan 4 = 76", Several approximate integral curves are shown (deawa solidly) { | | j H384 Approximate METHODS OF SOLVING FIRST-ORDER EQUATIONS 8.2 POWER SERIES METHODS ‘A. Introduction Although the graphical methods of the preceding section are very general, they suffer {rom several serious disadvantages. Not only are they tedious and subject to possible errors of construction, but they merely provide us with the approximate graphs of the solutions and do not furnish any analytic expressions for these solutions. Although we have now passed the naive state of searching for a closed-form solution in terms of elementary functions, we might still hope for solutions that can be represented as some {ype of infinite series. [n particular, we shall seek solutions that are representable as power series, We point out that not all first-order differential equations possess solutions that can be represented as power series, and itis beyond the stope of this book to consider conditions under which a first-order differential equation does possess such solutions. In order to explain the power series methods we shall assume that power series solutions actually do exist, realizing that this is an assumption that is not always justified, Specifically, we consider the initial-value problem consisting of the differential ‘equation Pasjuy ey and the initia condition Woo)= v0 86) and assume that the differential equation (8.1) possesses a solution that is representable asa power series in powers of (x — x) Thats, we assume thatthe differential equation (8.1) has a solution of the form y ro teite mn tee ot tom Foe ah n that is valid in some interval about the point xp. We now consider methods of determining the coefficients ¢o, cy, C2... in (8.7) so that the series (8.7) actually does satisfy the differential equation (8.1. 1B. The Taylor Series Method ‘We thus assume thatthe initial-value problem consisting of the diferemtial equation (81) and the initial condition (86) has a solution of the form (8.7) that is valid in some interval about xp, Then by Taylor's theorem, foreach xin this interval the value y(x) of this solution is given by VOI = yxe) + YU — 0) + = yx) fe xeh 68)a2 rower sens memoos 385 From the inital condition (8.6), we have 50) = You and from the diflerential equation (8.1) tel, Ie) = Sl505 ‘Substituting these values of y(xo) and yea) into the series in (8.8), we obtain the frst two coefficients of the desired series solution (8.7). Now differentiating the diferential equation (3.1), we obtain 1 ya 4 Rim sel = H+ Sou) iz Ce + fl FC Yh 69) ‘where we use subscripts to denote partial dffereniations. From this we obtain Yo) = Salton Yo) + Leos Yo)f ou Yo) Substituting this value of s(x) into (88), we obtain the third coefficient inthe series solution (8.7), Proceeding in like manner, we differentiate (8) successively to obtain eras: wat From these we obtain the values Yah Yahoos Yh ee Substituting these values into (8.8), we obtain the fourth and following coefficients in the series solution (8.7). Thus the coeficients in the series solution (8.7) are successively determined, D> Example 8.4 ‘Use the Taylor series method to obtain a power series solution of the initial-value problem ® ayy, Paver, 619) y= 1, eu) in powers of x. Solution. Since we seek a solution in powers of x, we set xy = assume a solution of the form in (87) and thos : ymegter teste Fart By Taylor's theorem, we know that for each xin theinterval where this solutions valid 0 eyo) £00) = $20 ea ont ye) = 910) + YO 4 PP @.12)386 areroumare metHo0s OF SOLVING FRST-ORDER EQUATIONS ‘The initial condition (8.11) states that WO=1, 13 and from the diferential equation (&.10) we see that YO=F+P =1. (B14) Differentiating (8.10) successively, we obtain ay ao (8.17) ‘ satnig = 0.712 nt 18 hai VO) = 20) + A1)(1) = 2 8.18) sebiuing y= 1220 1.2 22m 6, weobin YO) = 2 + 21}(2) + 21P = 8. 8.19) into (8.17), we find that Finally, substituting y (0) = (2KIN8) + (69112) = 28, (620) By successive differentiation of (8.17), we could proceed to determine #y fy rae and henoe obtain YO) yO), ‘Now substituting the values given by (8.13) (8.14) (8.18).(8.19).and (8.20) into(8.12},we ~ obtain the fist five coefficients ofthe desired series solution, We thus find the solution lesedere eee al 45,7 alexectate glee 8.21 Verte apt iate (821) C. The Method of Undetermined Coefficients ‘We now consider an alternative method for obtaining the coefficients co, ¢1,¢yy--ia the assumed series solution Ym eg tex = Xo) Hebe = ag) Hom Fel — Koh en82 rower sens merions 387 of the problem consisting of the differential equation (8.1) with initial condition (8.6). ‘Wesshall refer to this alternative method as the method of undetermined coefficients. In order to apply it we assume that f(x, y)in the differential equation (8.1)is representable inthe form $089) = 090+ aol — Xo) + Boil ~ Yo) + Bs — Xa)? + ajC4 x0 y ~ Yo) + taal) — Io)? +> @2 “The coefficients a, in (822) may be found by Taylor’ form for functions of 1wo variables, although in many simple cases the use ofthis formula is unnecsvary. Using the representation (8.22) for f(x, y), the differential equation (8.1) takes the form 4 Fe Moa + aol ~ x0) + dou — Yo) + diol — x0)? “+ ayats ~ x0Xy ~ 90) + dgaly ~ Yo)? + (823 Now assuming thatthe series (87) converges in some interval |x — xl < nr > 0) about xo, we may differentiate (8.7) term by term and the resulting series wil alo converge on |x ~ xa| < and represent yx) there. Doing this we thus obtain Boe + tee 20) + Bese — Ha + (324) ‘We note that in order forthe series (8.7) 1osatisty the initial condition (6) that y = yp aU x= x, We must have co = yg and hence Y Yo eax — x0) + 26x — Xo)? Ho (8.25) Now substituting (8.7) and (8.26) into the differential equation 8.23), and making use of (8.25), we find that 64 + Beal ~ Xg) + Besta — xo}? + tog + Ayo(X — Xo) + doy Ley(¥ ~ Xo) + €4(X — Xo) + °°) Hao — Xo)? + ay (x — xo) Ley(x — Xo) + eal — Xo)? o--] + aoaLeyls ~ Xo) + cal ~ xo)? +P + (8.26) Performing the multiplications indicated in the right member of (826) and then ‘combining like powers of (x — xo), we see that (8.26) takes the form ey 4 Zeyh — x0) + Beg — x0)? + = a9 + (@ia + anges MX — 0) HF Adoye2 + ayo + ary, + aoseIVx ~ Xo) + 8.27) Jn order that (827) be satisfied for all values of x in the interval lx — xol Bxampleas Use the method of undetermined coefficients to obtain a power series solution of the initial-value problem 6.10) (ea) in powers of: Solution. Since xy = 0, the assumed solution (87) is of the form yee tartee ter + (829) {In order to satisfy the initial condition (8.11), we must hi (8.29) cakes the form (eco = tnd hence the series vel tax teat text (8:30) Differentiating (8.30), we obtain Dy eg 4 2egx + Beye? . Fee en tea Bey? + dege? + (8.31) For the differential equation (8.10) we have fx, 9) |, we must expand x? + y? in the form 24 y?, Since xp = 0 and Ewe Since eyo ey h the desired expansion is given by ste yha lt 2y- text ey 0h ‘Thus the differential equation (8.10) takes the form Bare nes suet amy82 rower series mernoos 389. [Now substituting (8.30) and (8.31) into the differential equation (8.32), we obtain ey + Dex Sogn? + deg? + BLE pert tee tpt teeta to P (833) Performing the indicated multiplications and collecting like powers of x in the right ‘member of (833), we see that it takes the form ey + Deak + Sean? ean? + 1+ Dey + (ea + Lb etx? + ey + Zea? +o. (B34) Equating the coefficients of the like powers of xin both members of (8.34), we obtain the conditions ash Yep = Ree fea tite, bial deg = 25-4 266, From the conditions (8.35), we obtain suecessvely aah (836) ey = $c, +148) = 4 = Hey + eyes) = HA = ‘Substituting the coefficiemts so determined in (8.36) into the series (8.30), we obtain the first five terms of the desired series solution. We thus find ya tere age edt ‘We note that ths is of course the same series previously obtained by the Taylor series ‘method and already given by (821). Remark. We have made but litle mention of the interval of convergence of the series involved in our diseussion, We have merely assumed that a power series solution exists and converges on some interval |x ~ xo] < r(r > 0)about the initial point xy. In a practical problem the interval of convergence is of vital concern and should be ‘determined, if possible. Another matter of great importance in practical problem is the determination ofthe number of terms which have to be found in order tobe certain ‘of a sufficient degree of accuracy. We shall not consider these matters here. Our primary purpose has been merely to explain the details of the methods, We refer the reader to more advanced treatises for discussions of the importance questions of ‘convergence and accuracy.390 armoxmare Mem00s oF SoLNG HST-O8DER EQUATIONS Exercises Exercises 1-7 by (a) the Taylor series method and (b) the method of undetermined | Obtain a power series solution in powers of x ofeach ofthe initial-value problems in i cocficients & a 4 + Q=1. 2 Fatty yo=4 » as 5 Pexssiny, =o. 6 Betansinn yO=0 tay 0) 0) = 3, dx x 1 Mectexcosy, 10-0 | dx Obtain a power series solution in powers of x ~ 1 ofeach of the nitil-value problems in Exercises 8-12-y (a the Taylor series method and (b) the method of undetermined coefficients. 4 a wee Baw net 24 y | wo Bersyey, Bax seosy, hen | 4 Dex? + xin 2 Bestsrsiny 90) 8.3. THE METHOD OF SUCCESSIVE APPROXIMATIONS. ‘A, The Method ‘We again consider the initial-value problem consisting of the differential equation aye Eafe» 6) and the intial condition Wo) = You 66) Wenow outine the Picard method of sucesve approximations for findings solution ofthis problem which valid om sme interval that includes the iil pote x | The first stp ofthe Picard methods quite unlike nything that we have done before and at first glance it appears to be athe rues. For the frst step actualy consist of N\ making a guess at the solution! That is, we choose a function 6 and call ita “zeroth {pprotimation” tothe actual solution, How do we make this pues? In other words\ { | | | | ' 83, THE MEMOD oF successive arrwoxiwanions 391 ‘what function do we choose? Actually, many diferent choices could be made. The only thing that we know about the actual solution is that in order to satily the initial condition (8.6) it must assume the value yo at x-=.x5. Therefore it would seem reasonable to choose for dp a function that assumes this value yg at x = x. Although this requirement is not essential, it certainly seems as sensible as anything cls. In particular, it soften convenient to choose for y the constant function that has the value yo forall. While this choice i certainly not essential tis often the simplest, most reasonable choice that quickly comes to mind. Insummary, then the fist step ofthe Picard method istochoose a function dy which will serve as a zeroth approximation. Having thus chosen a zeroth approximation $y, we now determine a frst ap- proximation 4 inthe following manner. We determine (x) so that (1) it satisfies the differential equation obtained from (8.1) by replacing y in f(x, » by Gol) and (2) it satisfies the inital condition (8.6). Thus @, is determined such that A tes00) = st, 60001 sn and 1050) = Yo. 638) ‘Wenow assume that f(x, do(x)] is continuous. Then 4 satisfies (8.37) and (8.38) if and only if 100 ~nt f Slt tt at 39) From this equation the first approximation ¢, is determined. ‘We now determine the second approximation ¢, in a similar manner. The function 43's determined such that A te,001 = st, 01001 (840) and :(%0) = Yor (4ly Ansring that fC 44(0 is comiouon, then satis (6.40) and (4) i and oniy i tst= of" see duonae (4 From this equation the second approximation @, is determined. ‘We now proceed in like manner to determine a third approximation $5, a fourth. approximation ¢, and so on. The nth approximation @, is determined from 8.09 = yo f $ll.4,-00)] dt, 643) where ¢,.. 8 the (n ~ Ist approximation. We thus obtain a sequence of functions abide392 APPROXIMATE METHODS OF SOLVING FRSTORDEE EQUATIONS where ischosen, gy is determined from (8.39) dz is determined from (8 42,...,andin eneral gs determined from (8.43) form = | Now just how does this sequence of functions relate tothe actual solution of the intii-value problem under consideration? It ean be proved, under certain general conditions ard for x restricted toa sufcient small interval about the initial point x, that (1)asm~ co the sequence of functions 4, defined by (843) form & 1 approaches a limit function 6, and (2) this limit fonction satisfies both che diferential equation 8.1) and te inital condition (8.6), Thats, under suitable restrictions the function ¢ defined by $= lim 4, is the exact solution ofthe initial-value problem under consideration, Furthermore, the error in approximating the exact solution @ by the ith approximation ¢ will be asbitrarly small provided that nis sufficiently largeand that xis sufficiently lose to the initial point xo, 8. An Example; Remarks on The Method We illustrate the Picard method of suecessive approximations by applying it to the initial-value problem of Examples 84 and 8.5. > Exainple 8.6 Use the method of successive approximations to find a sequence of functions that approaches the solution of the initial-value problem Bap, um vost am Solution. Our first step is to choose a function for the zeroth approximation do, ‘Since the initial value of y is 1, it would seem reasonable to choose for dy the constant funetion that has the value I for all x. Thus, we let d be such that ula) = {forall x. The nth approximation &, for = 1 is given F~ formula (843). Since fo) Thy inthe diferential equation (8.10), the formula (8.43) becomes in this ease daixl= 14 Sab, tM, med