Lectures On Ordinary Di Erential Equations

3y ago
4.27 MB
139 Pages
Last View : 7d ago
Last Download : 5m ago
Upload by : Ronan Garica

1DRAFTLectures on Ordinary Differential Equations(Oxford Physics Papers CP3/4)Alexander A. Schekochihin†The Rudolf Peierls Centre for Theoretical Physics, University of Oxford, Oxford OX1 3PU, UKMerton College, Oxford OX1 4JD, UK(compiled on 5 February 2021)These are the notes for my lectures on Ordinary Differential Equations for 1st-yearundergraduate physicists, taught since 2018 as part of Paper CP3 at Oxford. They alsoinclude lectures on Normal Modes (part of Paper CP4), taught sunce 2021. I will begrateful for any feedback from students, tutors or (critical) sympathisers. The latest,up-to-date version of this file is available at rSchekochihin/ODE/2018/ODELectureNotes.pdf.As typos, errors, and better ways of explaining things are found, I will update thesenotes in real time as lectures proceed. Please check back here for updates (the date tag ofthe version that you are reading appears above).Below I give references to some relevant bits of various books and online lecture notes.Having access to these materials is not essential for being able to follow these lectures,but reading (these or other) books above and beyond your lecturer’s notes is essential foryou to be able to claim that you have received an education.The problem sets suggested for your tutorials are at the end of this document. They toowill be updated in real time if errors are found or new/better questions arise.CONTENTS1. The Language of the Game1.1. What is an ODE?1.2. Integral Curves and the Cauchy Problem1.3. Existence and Uniqueness1.4. Parametric Solutions2. Some Methods for Solving First-Order ODEs2.1. Equations in Full Differentials2.1.1. Integrating Factor2.2. Separable Equations2.2.1. Separable Equations and Composite Solutions2.2.2. Reduction to Separable Form by Linear Change of Variables2.3. Homogeneous Equations2.3.1. Reduction to Homogeneous Form by Linear Change of Variables2.4. Linear Equations2.4.1. Solution via Integrating Factor2.4.2. Solution via Variation of Constant2.4.3. CF PI† E-mail: 16171819192021 A. Schekochihin2.5. Bernoulli Equations2.6. Riccati Equations2.7. Equations Unresolved With Respect to Derivative2.7.1. Cauchy Problem2.7.2. Solution via Introduction of ParameterThe Language of the Game: Going to Higher Order3.1. Mathematical Pendulum3.2. Laws of Motion3.3. All ODEs Are (Systems of) First-Order ODEs3.4. Existence and Uniqueness3.5. Phase Space and Phase Portrait3.5.1. Linear Pendulum3.5.2. Nonlinear Pendulum3.5.3. Local Linear Analysis3.5.4. Damped Pendulum (Introduction to Dissipative Systems)Linear ODEs: General Principles4.1. Existence and Uniqueness Theorem for Linear Equations4.2. Superposition Principle4.2.1. Superposition Principle for Inhomogeneous Equations4.3. General Solution of Homogeneous Equations4.3.1. (In)dependent Somewhere—(In)dependent Everywhere4.3.2. The Fundamental Matrix and the Wronskian4.3.3. How to Construct a Fundamental System4.3.4. How to Construct the Solution4.4. General Solution of Inhomogeneous Equations4.5. Green’s Function4.6. Buy One Get One Free4.6.1. Tips for Guessing GamesSecond-Order Linear ODE with Constant Coefficients5.1. Homogeneous Equation5.1.1. Damped Oscillator5.1.2. Homogeneous Equation: Degenerate Case5.1.3. Above and Beyond: n-th-Order Homogeneous Equation5.1.4. Scale-Invariant (Euler’s) Equation5.2. Inhomogeneous Equation5.2.1. Some Tips for Finding Particular Solutions5.2.2. Above and Beyond: Quasipolynominals and n-th-Order Inhomogeneous Equation5.3. Forced Oscillator5.3.1. Resonance5.3.2. Energy Budget of Forced Oscillator5.4. (Nonlinear) Digression: Rapidly Oscillating ForceSystems of Linear ODEs with Constant Coefficients6.1. Diagonalisable Systems With No Degenerate Eigenvalues6.1.1. General Solution of Inhomogeneous Equation6.2. Hermitian Systems6.3. Non-Hermitian Systems6.3.1. Solution by Triangulation6.3.2. Proof of Schur’s Triangulation Theorem6.3.3. Solution via Jordan 70

Oxford Physics Lectures: Ordinary Differential Equations37. Normal Modes of Linear Oscillatory Systems7.1. Coupled Identical Oscillators7.1.1. General Method7.1.2. Solution for Two Coupled Pendula7.1.3. Examples of Initial Conditions7.2. Energetics of Coupled Oscillators7.2.1. Energy of a System of Coupled Oscillators7.2.2. Restoring Forces Are Potential7.2.3. Energy in Terms of Normal Modes7.3. Generalisations and Complications7.3.1. Pendula of Unequal Length7.3.2. Pendula of Unequal Mass7.3.3. Damped Oscillators7.3.4. Forced Oscillators and Resonances7.3.5. A Worked Example: Can Coupled Pendula Be in Resonance WithEach Other?7.3.6. N Coupled Oscillators7.4. (Nonlinear) Digression: Coupled Oscillators and Hamiltonian Chaos8. Qualitative Solution of Systems of Autonomous Nonlinear ODEs8.1. Classification of 2D Equilibria8.1.1. Nodes: T 2 4D 08.1.2. Saddles: T 2 0 4D8.1.3. Pathologies: D 08.1.4. Effect of Nonlinearity8.1.5. Pathologies: T 2 4D8.1.6. Foci: 4D T 2 08.1.7. Centres: 4D T 2 08.1.8. Conservative Systems8.2. (Nonlinear) Digression: Hamiltonian Systems and Adiabatic Invariants8.3. Limit Cycles8.3.1. Poincaré–Bendixson Theorem8.4. Auto-Oscillations8.4.1. Liénard’s Theorem8.4.2. Relaxation Oscillations8.5. Outlook: Attractors, Dissipative Chaos, and Many Degrees of FreedomAppendix: Introduction to Lagrangian and Hamiltonian MechanicsA.1. Hamilton’s Action PrincipleA.2. Lagrangian of a Point MassA.3. Lagrangian of a System of Point MassesA.4. MomentumA.5. EnergyA.6. Hamilton’s Equations of MotionA.7. Liouville’s TheoremSources and Further 13114116Problem SetsPS-1: First-Order ODEsPS-2: Second-Order ODEs, Part IPS-3: Second-Order ODEs, Part IIPS-4: Systems of Linear ODEsiivixxiii

4A. A. SchekochihinPS-5: Systems of Nonlinear ODEsPS-6: Masses on Springs and Other Thingsxvxvii

Oxford Physics Lectures: Ordinary Differential Equations5To play the good family doctor who warns about reading something prematurely,simply because it would be premature for him his whole life long—I’m not the manfor that. And I find nothing more tactless and brutal than constantly tryingto nail talented youth down to its “immaturity,” with every other sentencea “that’s nothing for you yet.” Let him be the judge of that! Let himkeep an eye out for how he manages.Thomas Mann, Doctor Faustus1. The Language of the GameIt might not be an exaggeration to say that the ability of physical theory to make predictions and, consequently, both what we proudly call our “understanding of the world”and our technological civilisation hinge on our ability to solve differential equations—or,at least, to write them down and make intelligent progress in describing their solutions inqualitative, asymptotic or numerical terms. Thus, what you are about to start learningmay well be the most “practically” important bit of mathematics for you as physicists—sobe excited!I shall start by introducing some terms and notions that will form the language thatwe will speak and by stating some general results that enable us to speak it meaningfully.Then I shall move on to methods for solving and analysing various kinds of differentialequations.1.1. What is an ODE?[Literature: Pontryagin (1962, §1), Tikhonov et al. (1985, §§1.1-1.2), Arnold (2006, §§1.1-1.10),Yeomans (2014, §II.0), Tenenbaum & Pollard (1986, §3), Bender & Orszag (1999, §1.1)]A differential equation is an equation in which the unknowns are functions of one ormore variables and which contains both these functions and their derivatives. Physically,it is a relationship between quantities and their rates of change—in time, space orwhatever other “independent variable” we care to introduce.If more than one independent variable is involved, the differential equation is called apartial differential equation (PDE). Those will not concern me—you will encounter themnext term, in CP4, and also, in a proper way, in the second-year Mathematical Methods(Eßler 2009; Magorrian 2017; Lukas 2019).It makes sense that this should happen after you have—hopefully—learned from mehow to handle ordinary differential equations (ODEs), which are differential equationsthat involve functions of a single independent variable. They have the general formF (x, y, y 0 , . . . , y (n) ) 0 ,(1.1)where x is the independent variable, y is the function (or “dependent variable”1 ) and y 0 ,y 00 , . . . , y (n) are its first, second, . . . , nth derivatives.The order of an ODE is the order of the highest derivative that it contains.There can be more than one function involved: y1 (x), y2 (x), . . . , in which case therewould have to be more than one equation relating them and their derivatives (see §3.3).I will denote the independent variable x, but also often t, when I particularly want tothink of my ODE as describing the evolution of some physical quantity with time (or1We shall see in what follows that what is an “independent” and what is a “dependent” variablecan be negotiable as we manipulate ODEs. If y is a function of x, x is a function of y. See §1.4.

6A. A. SchekochihinFigure 1. Gottfried Wilhelm von Leibniz (1646-1716), great German philosopher, inventor ofcalculus and of monads (intrigued? look them up!), Newton’s (Fig. 11) bête noire.just when I feel like it). A derivative is the instantaneous rate of change of a quantitywith time (or the local rate of its change in space, or etc.). It is denoted, by convention,dy ẏdtordy y0 .dx(1.2)The tall-fraction notation, introduced by Leibniz (Fig. 1), underscores the derivative’smeaning as the change in y per change in t (or x) and also signals that derivatives can(with caution) be handled as fractions, e.g.,dx1 0dyy(1.3)is the derivative of x with respect to y.Example. Perhaps the most famous and most important ODE in the world isẏ ay,(1.4)where a is some constant. It describes a situation in which the rate of growth (whena is real and positive) or decay (when a 0) of some quantity is proportional to thequantity itself, e.g., the growth of a population whose rate of increase is proportional tothe number of individuals in it (i.e., a monogamous population; what is the ODE for apolygamous population?). This equation is first-order. It is also linear.Generally, an ODE is linear if it has the forman (x)y (n) an 1 y (n 1) · · · a2 (x)y 00 a1 (x)y 0 a0 (x)y f (x) ,(1.5)or, to use Einstein’s convention of implied summation over repeated indices, here meantto run from 0 to n,ai (x)y (i) f (x).(1.6)When f (x) 0, the linear ODE is called homogeneous [of which (1.4) is an example],otherwise it is inhomogeneous. Homogeneous linear ODEs have the important propertythat they do not change under an arbitrary rescaling y λy λ 6 0.2 We shall see (§4)that linear ODEs have many nice properties that make them much easier to solve thanmost nonlinear ones—and they also turn out to be useful in the analysis of the latter(§§3.5.3, 8).2Nonlinear ODEs can also have this property, in which case they are easier to solve: see Q2.6.

Oxford Physics Lectures: Ordinary Differential Equations7An example of nonlinear ODE isẏ ay 2 ,(1.7)which describes the evolution of a radically polygamous (or rather “omnigamous”)population, whose rate of increase is proportional to the number of all possible pairings.Let me return to (1.4). You can probably guess quite easily what its solution is:y Ceat ,(1.8)where C is an arbitrary constant.3 The emergence of C highlights an important propertyof ODEs: typically, they have an infinite number solutions, parametrised by arbitraryconstants.4 We shall see that the number of constants should be equal to the order ofthe equation, but some more work is needed to make this obvious.The solution (1.8) has been written in terms of the familiar exponential function. Butwhat is that? In fact, it is possible to introduce ex by definition as the function whosederivative is equal to the function itself.Exercise 1.1. Ponder or investigate how this can be done.Thus, in a sense, (1.8) is the solution of (1.4) by definition. This illustrates a generalprinciple: so called “elementary functions” can be thought of as nothing but solutions ofsome oft-encountered and useful ODEs. When an ODE cannot be solved in elementaryfunctions—or, more generally, in quadratures, which means in terms of integrals ofcombinations of elementary functions—and most ODEs cannot be so solved!—one hasfour options:(i) Introduce a new function, defined as the solution of the previously unsolved ODE,and study it in some way: e.g., work out a power series for it (see Q2.9), plot it, tabulateit, make an app for computing it. These, usually in application to second-order linearODEs, are called special functions (a good textbook on them is Lebedev 1972).(ii) Find approximate, or asymptotic, solutions (a graduate-level textbook: White 2010;a cult-status, wild romp: Bender & Orszag 1999; a short classic introduction: Erdélyi2003; see also Tikhonov et al. 1985, §7).(iii) Find numerical solutions (a short introduction: Tikhonov et al. 1985, §6.1; anultimate “how-to” guide: Press et al. 2002; a mathematical theory: Samarskii 2001).(iv) Study solutions qualitatively (§§3.5 and 8).There is another way in which (1.4) is special: it is resolved with respect to derivative,i.e., it is of the general formẏ f (t, y) .(1.9)This type of ODE allows for a vivid way of thinking of the multiplicity of an ODE’ssolutions, which we shall now discuss.3It is obvious that there should be a multiplicative constant in the solution of any homogeneousODE because of the possibility of an arbitrary rescaling of y.4In Q1.8, you get to play with an “inverse” problem: given an infinite, parametrised set offunctions, find an ODE whose solutions they are.

8A. A. Schekochihin(a) Integral curves of (1.9)(b) Integral curves of (1.4)Figure 2. Integral curves of (1.9), an ODE that is resolved with respect to derivative: (a) generalcase, (b) integral curves (1.8) of (1.4). The domain where f (t, y) is defined is (a) D, (b) the entireplane R2 . The ODE’s direction field is (1, f ), shown by arrows in (a). The initial conditionis (t0 , y0 ).1.2. Integral Curves and the Cauchy Problem[Literature: Pontryagin (1962, §1), Tikhonov et al. (1985, §1.1), Arnold (2006, §1.3),Tenenbaum & Pollard (1986, §§4-5), Bender & Orszag (1999, §1.2)]Let us suppose that the function f (t, y) is specified in some domain D within the R2plane (t, y). Through each point of this domain, let us draw a line whose pslope is f (t, y),i.e., whose direction vector is (1, f ) (if you like, you can normalise it by 1 f 2 ). Thisis called the direction field of the ODE (1.9). Then solutions of (1.9) are curves y(t) thatare at each point in D tangent to its direction field (Fig. 2a). They are called integralcurves. Solving an ODE means finding all its integral curves.For example, the integral curves of (1.4), parametrised by C in (1.8), are graphs of theexponential function in the (t, y) plane that intersect the y axis at y C (Fig. 2b). Wecan pick a single one of these curves, i.e., make the solution unique if we append to (1.4)an additional requirement that y(t 0) y0 for some given y0 . Then the desired curve(1.8) is the one with C y0 .More generally, an initial condition (IC)—which, if we were speaking in terms of aspatial variable x, rather than time t, we would call a boundary condition (BC)—is thestatement thaty(t0 ) y0for some (t0 , y0 ) D.(1.10)Equivalently, it is the statement that the integral curve that we seek must pass throughthe point (t0 , y0 ).The problem of finding the solution of (1.9) that satisfies the initial condition (1.10)is called the initial-value problem, or Cauchy problem (Fig. 3).Does the Cauchy problem always have a solution? If we can find one, is it the only oneor are there others? In other words, is there an integral curve that passes through everypoint (t0 , y0 ) D and can these curves ever intersect?

Oxford Physics Lectures: Ordinary Differential Equations9Figure 3. Baron Augustin-Louis Cauchy (1789-1857), great French mathematician, aristocratand rabid right-winger.1.3. Existence and Uniqueness[Literature: Pontryagin (1962, §§1, 2.4, 20), Tikhonov et al. (1985, §2.2), Arnold (2006,§§2.1-2.3), Tenenbaum & Pollard (1986, §58), Coddington (1990, Ch. 5)]The answer to these questions is given by the following existence and uniquenesstheorem.Theorem 1. Let f (t, y) and f / y exist and be continuous functions on some opendomain5 D R2 . Then(a) (t0 , y0 ) D, t such that the Cauchy problemẏ f (t, y),y(t0 ) y0has a solution in the interval I [t0 t, t0 t].(1.11)6(b) This solution is unique, i.e., if y1 (t) and y2 (t) are solutions of (1.11) on the intervalsI1 and I2 , respectively, then y1 (t) y2 (t) t I1 I2 (they are the same in theintersection of the intervals where they are solutions).Thus, we are guaranteed that we can solve the initial-value problem for at least sometime after (and before) t0 and that if we have found a solution, we can rest assured thatit is the only one. We are in business!Example. Let me illustrate how some of this can be broken and, therefore, why oneshould not be dismissive about formulating mathematical results precisely and watchingfor all the stated conditions to be satisfied. Consider the following Cauchy problem:ẏ y 2/3 ,y(0) 0.(1.12)Clearly, y(t) 0 is a solution. But y(t) (t/3)3 is also a solution, as can be verified bydirect substitution. The two integral curves intersect at t 0 (Fig. 4)! What has gone5D being an open domain means that around point D, there is a circle of some radius suchthat all points within the circle are also D. Thus, if you think of D as some bounded area, itcannot include its own boundary. Obviously, the whole space R2 is very much an open domain.6Formally, this means that a function y(t) such that y(t0 ) y0 and t I , (t, y(t)) D, ẏ(t)is continuous and ẏ(t) f (t, y(t)).

10A. A. SchekochihinFigure 4. Non-uniqueness: two solutions of (1.12), shown as bold lines, intersect at the origin.wrong? This is easy to see:f (t, y) y 2/3 f2 y 1/3 at y3 y 0.(1.13)Thus, f / y does not exist at y 0, the conditions of Theorem 1 are violated, and sothe Cauchy problem is under no obligation to have a unique solution.1.4. Parametric SolutionsBefore I move on, let me generalise slightly what I have just done. Casting aside thebigoted language of “dependent” (y) and “independent” (x or t) variables, let me treateverybody equally and write an ODE in the so-called symmetric form:P (x, y)dx Q(x, y)dy 0 .(1.14)We shall see (§2) that ODEs always end up in this form before getting integrated. Clearly,any ODE of the form (1.9) can be recast in the form (1.14) (Q 1, P f ), but notnecessarily vice versa because Q(x, y) may well vanish in some places, preventing onefrom dividing by it and simply recasting (1.14) asdyP (x, y) f (x, y).dxQ(x, y)(1.15)To be precise, what we shall assume is that P and Q are continuous functions in somedomain D R2 and that P (x, y) Q(x, y) 0 (x, y) D, i.e., that they do notvanish simulatneously. This means that in some parts of D, where Q 6 0, our equation(1.14) can be written as (1.15), an ODE resolved with respect to dy/dx, and in others,where P 6 0, it can be resolved with respect to dx/dy:Q(x, y)dx .dyP (x, y)(1.16)The symmetric form (1.14) is a compact way of writing the two equations (1.15) and(1.16) together.Solutions of (1.14) can in general be thought of as determined parametrically in termsof an auxiliary variable t: namely, (x(t), y(t)) is a solution of (1.14) if it is specifiedin some interval I R, and, t I , x(t) and y(t) are continuously differentiable,(x(t), y(t)) D,

Yeomans(2014, xII.0),Tenenbaum & Pollard(1986, x3),Bender & Orszag(1999, x1.1)] A di erential equation is an equation in which the unknowns are functions of one or more variables and which contains both these functions and their derivatives.

Related Documents:

Ordinary Di erential Equations by Morris Tenenbaum and Harry Pollard. Elementary Di erential Equations and Boundary Value Problems by William E. Boyce and Richard C. DiPrima. Introduction to Ordinary Di erential Equations by Shepley L. Ross. Di

69 of this semi-mechanistic approach, which we denote as Universal Di erential 70 Equations (UDEs) for universal approximators in di erential equations, are dif- 71 ferential equation models where part of the di erential equation contains an 72 embedded UA, such as a neural network, Chebyshev expansion, or a random 73 forest. 74 As a motivating example, the universal ordinary di erential .

1.2. Rewrite the linear 2 2 system of di erential equations (x0 y y0 3x y 4et as a linear second-order di erential equation. 1.3. Using the change of variables x u 2v; y 3u 4v show that the linear 2 2 system of di erential equations 8 : du dt 5u 8v dv dt u 2v can be rewritten as a linear second-order di erential equation. 5

Ordinary di erential equations This chapter contains three papers which are on the integer-order ordinary di erential equations for boundary value problem. 1.1 The symmetric positive solutions of four-point boundary value problems for nonlin-ear second-order di erential equations Abstract: In this paper, we are concerned with the existence of .

Definition 1.1 A di erential equation is an equation that relates a function to its derivative(s). The unknown is the function. A di erential equation is said to be ordinary (!הליגר) if the function is uni-variate, and more precisely if its domain is a connected subset of R. We ab

General theory of di erential equations of rst order 45 4.1. Slope elds (or direction elds) 45 4.1.1. Autonomous rst order di erential equations. 49 4.2. Existence and uniqueness of solutions for initial value problems 53 4.3. The method of successive approximations 59 4.4. Numerical methods for Di erential equations 62

topology and di erential geometry to the study of combinatorial spaces. Per-haps surprisingly, many of the standard ingredients of di erential topology and di erential geometry have combinatorial analogues. The combinatorial theories This work was partially supported by the National Science Foundation and the National Se-curity Agency. 177

Unit #15 - Di erential Equations Some problems and solutions selected or adapted from Hughes-Hallett Calculus. Basic Di erential Equations 1.Show that y x sin(x) ˇsatis es the initial value problem dy dx 1 cosx To verify anything is a solution to an equation, we sub it in and verify that the left and right hand sides are equal after