Page 3 :
This page intentionally left blank
Page 4 :
SCHAUM'S, OUTLINE OF, , Advanced, Mathematics for, , Engineers and, Sciuentists, Murray R. Spiegel, Ph.D., Former Professor and Chairman,, Mathematics Department, Rensselaer Polytechnic Institute, Hartford Graduate Center, , Schaum's Outline Series, , New York Chicago San Francisco Lisbon, London Madrid Mexico City Milan New Delhi, San Juan Seoul Singapore Sydney Toronto
Page 5 :
Copyright © 1971 by The McGraw-Hill Companies, Inc. All rights reserved. Except as permitted under the United States Copyright Act of, 1976, no part of this publication may be reproduced or distributed in any form or by any means, or stored in a database or retrieval system,, without the prior written permission of the publisher., ISBN: 978-0-07-170242-3, MHID: 0-07-170242-3, The material in this eBook also appears in the print version of this title: ISBN: 978-0-07-163540-0, MHID:0-07-163540-8., All trademarks are trademarks of their respective owners. Rather than put a trademark symbol after every occurrence of a trademarked name,, we use names in an editorial fashion only, and to the benefit of the trademark owner, with no intention of infringement of the trademark., Where such designations appear in this book, they have been printed with initial caps., McGraw-Hill eBooks are available at special quantity discounts to use as premiums and sales promotions, or for use in corporate training, programs. To contact a representative please e-mail us at bulksales@mcgraw-hill.com., TERMS OF USE, This is a copyrighted work and The McGraw-Hill Companies, Inc. (“McGrawHill”) and its licensors reserve all rights in and to the work., Use of this work is subject to these terms. Except as permitted under the Copyright Act of 1976 and the right to store and retrieve one copy, of the work, you may not decompile, disassemble, reverse engineer, reproduce, modify, create derivative works based upon, transmit,, distribute, disseminate, sell, publish or sublicense the work or any part of it without McGraw-Hill’s prior consent. You may use the work for, your own noncommercial and personal use; any other use of the work is strictly prohibited. Your right to use the work may be terminated if, you fail to comply with these terms., THE WORK IS PROVIDED “AS IS.” McGRAW-HILL AND ITS LICENSORS MAKE NO GUARANTEES OR WARRANTIES AS, TO THE ACCURACY, ADEQUACY OR COMPLETENESS OF OR RESULTS TO BE OBTAINED FROM USING THE WORK,, INCLUDING ANY INFORMATION THAT CAN BE ACCESSED THROUGH THE WORK VIA HYPERLINK OR OTHERWISE,, AND EXPRESSLY DISCLAIM ANY WARRANTY, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO IMPLIED, WARRANTIES OF MERCHANTABILITY OR FITNESS FOR A PARTICULAR PURPOSE. McGraw-Hill and its licensors do not warrant or, guarantee that the functions contained in the work will meet your requirements or that its operation will be uninterrupted or error free., Neither McGraw-Hill nor its licensors shall be liable to you or anyone else for any inaccuracy, error or omission, regardless of cause, in the, work or for any damages resulting therefrom. McGraw-Hill has no responsibility for the content of any information accessed through the, work. Under no circumstances shall McGraw-Hill and/or its licensors be liable for any indirect, incidental, special, punitive, consequential or, similar damages that result from the use of or inability to use the work, even if any of them has been advised of the possibility of such, damages. This limitation of liability shall apply to any claim or cause whatsoever whether such claim or cause arises in contract, tort or, otherwise.
Page 6 :
Schaum*s Outlines, and the Power of Computers..., The Ultimate Solution!, Now Available! An electronic, interactive version of Theory and, Problems of Electric Circuits from the Schauiris Outline Series., MathSoft, Inc. has joined with McGraw-Hill to offer you an electronic version, of the Theory and Problems of Electric Circuits from the Schaum's Outline, Series. Designed for students, educators, and professionals, this resource, provides comprehensive interactive on-screen access to the entire Table of, Contents including over 390 solved problems using Mathcad technical, calculation software for PC Windows and Macintosh., When used with Mathcad, this "live" electronic book makes your problem, solving easier with quick power to do a wide range of technical calculations., Enter your calculations, add graphs, math and explanatory text anywhere on the, page and you're done - Mathcad does the calculating work for you. Print your, results in presentation-quality output for truly informative documents, complete, with equations in real math notation. As with all of Mathcad's Electronic, Books, Electric Circuits will save you even more time by giving you hundreds of, interactive formulas and explanations you can immediately use in your own, work., Topics in Electric Circuits cover all the material in the Schaum's Outline, including circuit diagramming and analysis, current voltage and power relations, with related solution techniques, and DC and AC circuit analysis, including, transient analysis and Fourier Transforms. All topics are treated with "live", math, so you can experiment with all parameters and equations in the book or, in your documents., , To obtain the latest prices and terms and to order Mathcad and the electronic, version of Theory and Problems of Electric Circuits from the Schaunt's Outline, Series, call 1-800-628-4223 or 617-577-1017.
Page 7 :
CONTENTS, Page, , Chapter, , 1, , REVIEW OF FUNDAMENTAL CONCEPTS, , 1, , Chapter, , 2, , ORDINARY DIFFERENTIAL EQUATIONS, , 38, , Chapter, , 3, , LINEAR DIFFERENTIAL EQUATIONS, , 71, , Chapter, , 4, , LAPLACE TRANSFORMS, , 98, , Chapter, , 5, , VECTOR ANALYSIS, , Real numbers. Rules of algebra. Functions. Special types of functions. Limits., Continuity. Derivatives. Differentiation formulas. Integrals. Integration formulas. Sequences and series. Uniform convergence. Taylor series. Functions, of two or more variables. Partial derivatives. Taylor series for functions of, two or more variables. Linear equations and determinants. Maxima and, minima. Method of Lagrange multipliers. Leibnitz's rule for differentiating, an integral. Multiple integrals. Complex numbers., , Definition of a differential equation. Order of a differential equation. Arbitrary constants. Solution of a differential equation. Differential equation of a, family of curves. Special first order equations and solutions. Equations of, higher order. Existence and uniqueness of solutions. Applications of differential equations. Some special applications. Mechanics. Electric circuits., Orthogonal trajectories., Deflection of beams., Miscellaneous problems., Numerical methods for solving differential equations., , General linear differential equation of order n. Existence and uniqueness, theorem. Operator notation. Linear operators. Fundamental theorem on linear, differential equations. Linear dependence and Wronskians. Solutions of linear, equations with constant coefficients. Non-operator techniques. The complementary or homogeneous solution. The particular solution. Method of undetermined coefficients. Method of variation of parameters. Operator techniques., Method of reduction of order. Method of inverse operators. Linear equations, with variable coefficients. Simultaneous differential equations. Applications., , Definition of a Laplace transform. Laplace transforms of some elementary, functions. Sufficient conditions for existence of Laplace transforms. Inverse, Laplace transforms. Laplace transforms of derivatives. The unit step function., Some special theorems on Laplace transforms. Partial fractions. Solutions of, differential equations by Laplace transforms. Applications to physical problems. Laplace inversion formulas., , Vectors and scalars. Vector algebra. Laws of vector algebra. Unit vectors., Rectangular unit vectors. Components of a vector. Dot or scalar product., Cross or vector product. Triple products. Vector functions. Limits, continuity, and derivatives of vector functions. Geometric interpretation of a vector, derivative. Gradient, divergence and curl. Formulas involving V. Orthogonal, curvilinear coordinates. Jacobians. Gradient, divergence, curl and Laplacian, in orthogonal curvilinear. Special curvilinear coordinates., , 121
Page 8 :
CONTENTS, , Chapter, , 6, , Page, , MULTIPLE, LINE AND SURFACE INTEGRALS AND, INTEGRAL THEOREMS, , Double integrals. Iterated integrals. Triple integrals. Transformations of, multiple integrals. Line integrals. Vector notation for line integrals. Evaluation of line integrals. Properties of line integrals. Simple closed curves., Simply and multiply-connected regions. Green's theorem in the plane. Conditions for a line integral to be independent of the path. Surface integrals. The, divergence theorem. Stokes' theorem., , Chapter, , 7, , Chapter, , 8, , Chapter, , 9, , 147, , FOURIER SERIES, , 182, , FOURIER INTEGRALS, , 201, , GAMMA, BETA AND OTHER SPECIAL FUNCTIONS, , 210, , Periodic functions. Fourier series. Dirichlet conditions. Odd and even functions. Half range Fourier sine or cosine series. Parseval's identity. Differentiation and integration of Fourier series. Complex notation for Fourier series., Complex notation for Fourier series. Orthogonal functions., , The Fourier integral. Equivalent forms of Fourier's integral theorem. Fourier, transforms., Parseval's identities for Fourier integrals., The convolution, theorem., , The gamma function. Table of values and graph of the gamma function., Asymptotic formula for T(n). Miscellaneous results involving the gamma function. The beta function. Dirichlet integrals. Other special functions. Error, function. Exponential integral. Sine integral. Cosine integral. Fresnel sine, integral. Fresnel cosine integral. Asymptotic series or expansions., , Chapter 10, , BESSEL FUNCTIONS, 224, Bessel's differential equation. Bessel functions of the first kind. Bessel functions of the second kind. Generating function for Jn(x). Recurrence formulas., Functions related to Bessel functions. Hankel functions of first and second, kinds. Modified Bessel functions. Ber, bei, ker, kei functions. Equations, transformed into Bessel's equation. Asymptotic formulas for Bessel functions., Zeros of Bessel functions. Orthogonality of Bessel functions. Series of Bessel, functions., , Chapter 11, , LEGENDRE FUNCTIONS AND OTHER, ORTHOGONAL FUNCTIONS, , 242, , PARTIAL DIFFERENTIAL EQUATIONS, , 258, , Legendre's differential equation. Legendre polynomials. Generating function, for Legendre polynomials. Recurrence formulas. Legendre functions of the, second kind. Orthogonality of Legendre polynomials. Series of Legendre polynomials. Associated Legendre functions. Other special functions. Hermite, polynomials. Laguerre polynomials. Sturm-Liouville systems., , Chapter 12, , Some definitions involving partial differential equations. Linear partial differential equations. Some important partial differential equations. Heat conduction equation. Vibrating string equation. Laplace's equation. Longitudinal, vibrations of a beam. Transverse vibrations of a beam. Methods of solving, boundary-value problems. General solutions. Separation of variables. Laplace, transform methods.
Page 9 :
CONTENTS, Page, , Chapter 13, , Chapter 14, , COMPLEX VARIABLES AND CONFORMAL MAPPING, , 286, , COMPLEX INVERSION FORMULA FOR, LAPLACE TRANSFORMS, , 324, , Functions. Limits and continuity. Derivatives. Cauchy-Riemann equations., Integrals. Cauchy's theorem. Cauchy's integral formulas. Taylor's series., Singular points. Poles. Laurent's series. Residues. Residue theorem. Evaluation of definite integrals. Conformai mapping. Riemann's mapping theorem., Some general transformations. Mapping of a half plane on to a circle. The, Schwarz-Christoffel transformation. Solutions of Laplace's equation by conformal mapping., , The complex inversion formula. The Bromwich contour. Use of residue, theorem in finding inverse Laplace transforms. A sufficient condition for the, integral around T to approach zero. Modification of Bromwich contour in case, of branch points., Case of infinitely many singularities. Applications to, boundary-value problems., , Chapter 15, , Chapter 16, , MATRICES, , 342, , Definition of a matrix. Some special definitions and operations involving, matrices. Determinants. Theorems on determinants. Inverse of a matrix., Orthogonal and unitary matrices. Orthogonal vectors. Systems of linear, equations. Systems of n equations in n unknowns. Cramer's rule. Eigenvalues, and eigenvectors. Theorems on eigenvalues and eigenvectors., , CALCULUS OF VARIATIONS, , 375, , INDEX, , 399, , Maximum or minimum of an integral. Euler's equation. Constraints. The, variational notation. Generalizations. Hamilton's principle. Lagrange's equations. Sturm-Liouville systems and Rayleigh-Ritz methods. Operator interpretation of matrices.
Page 10 :
This page intentionally left blank
Page 11 :
Chapter 1, , : Review of, Fundamental Concepts, , REAL NUMBERS, At the very foundations of mathematics is the concept of a set or collection of objects, and, in particular, sets of numbers on which we base our quantitative work in science and, engineering. The student is already familiar with the following important sets of numbers., 1. Natural Numbers 1,2,3,4, ... or positive integers used in counting., 2. Integers 0, ±1, ±2, ±3, . . . . These numbers arose in order to provide meaning to, subtraction [inverse of addition] of any two natural numbers. Thus 2 — 6 = —4,, 8-8 = 0, etc., 3. Rational Numbers such as 2/3, —10/7, etc. arose in order to provide meaning to division, [inverse of multiplication] or quotient of any two integers with the exception that, division by zero is not defined., 4. Irrational Numbers such as \/2, *, etc. are numbers which cannot be expressed as the, quotient of two integers., Note that the set of natural numbers is a subset, i.e. a part, of the set of integers which in, turn is a subset of the set of rational numbers., The set of numbers which are either rational or irrational is called the set of real, numbers [to distinguish them from imaginary or complex numbers on page 11] and is composed of positive and negative numbers and zero. The real numbers can be represented as, points on a line as indicated in Fig. 1-1. For this reason we often use point and number, interchangeably., , Fig. 1-1, , The student is also familiar with the concept of inequality. Thus we say that the real, number a is greater than or less than b [symbolized by a > b or a <b] if a — b is a positive, or negative number respectively. For any real numbers a and b we must have a > b,, a = b or a < b., RULES OF ALGEBRA, If a, b, c are any real numbers, the following rules of algebra hold., 1., , a+b = b+a, , 2., , a + (b + c) = (a + b) + c, , 3., 4., 5., , ab = ba, a(bc) — (ab)c, a(b + c) — ab + ac, , Commutative law for addition, Associative law for addition, Commutative law for multiplication, Associative law for multiplication, Distributive law, , 1
Page 12 :
2, , REVIEW OF FUNDAMENTAL CONCEPTS, , [CHAP. 1, , It is from these rules [if we accept them as axioms or postulates] that we can prove the, usual rules of signs, as for example (-5)(3) = -15, (-2)(-3) = 6, etc., The student is also familiar with the usual rules of exponents:, (Í), FUNCTIONS, Another important concept is that of function. The student will recall that a function, / is a rule which assigns to each object x, also called member or element, of a set A an, element y of a set B. To indicate this correspondence we write y = /(«) where f(x) is, called the value of the function at x., Example 1. If /(») = a;2 - 3x + 2, then /(2) = 22- 3(2) + 2 = 0., , The student is also familiar with the process of "graphing functions" by obtaining, number pairs (a;, y) and considering these as points plotted on an xy coordinate system. In, general y = f(x) is represented graphically by a curve. Because y is usually determined, from x, we sometimes call x the independent variable and y the dependent variable., SPECIAL TYPES OF FUNCTIONS, 1. Polynomials f(x) = a<>xn + aix"'1 + azxn~2 + • • • +an. If a 0 ^0, n is called the degree, of the polynomial. The polynomial equation f(x) = 0 has exactly n roots provided we, count repetitions. For example a;3 - 3a;2 + 3a; -1 = 0 can be written (x - I)3 = 0 so, that the 3 roots are 1,1,1. Note that here we have used the binomial theorem, , where the binomial coefficients are given by, and where factorial n, i.e. nl, = n(n— !)(•» — 2) • • • 1 while 0! = 1 by definition., 2. Exponential Functions f(x) = ax. These functions obey the rules (1)., An important special case occurs where a = e = 2.7182818, 3. Logarithmic Functions f(x) — loga x. These functions are inverses of the exponential, functions, i.e. if ax = y then x = logay where a is called the base of the logarithm., Interchanging x and y gives y = logo x. If a = e, which is often called the natural, base of logarithms, we denote log«a; by mo:, called the natural logarithm of x. The, fundamental rules satisfied by natural logarithms [or logarithms to any base] are, (*), , 4. Trigonometric Functions sin», cosa;, tana;, cotx, seca;, ese a;., Some fundamental relationships among these functions are as follows.
Page 13 :
(6), (c), (d), , 3, , REVIEW OF FUNDAMENTAL CONCEPTS, , CHAP. 1], , sin2 x + cos2 x — 1,, sec2 x — tan2 x — 1,, esc2 x — cot2 x = 1, sin (—a;) = — sin x,, cos (—x) = cos x,, tan (—a;) = — tan x, sin (x ± y) = sin x cos y ± cos x sin y, cos (a; ± #) = cos x cos 2/ T sin a; sin #, , (e), , A cos a; + B sin a; =, where tan a = A/B, The trigonometric functions are periodic. For example sin* and cosa;, shown in, Fig. 1-2 and 1-3 respectively, have period 2ir., , Fig. 1-2, , Fig. 1-3, , 5. Inverse Trigonometric Functions sin"1 a;, cos"1 a;, tan"1 a;, cot"1 a;, sec"1 a;, csc^a;., These are inverses of the trigonometric functions. For example if sin x = y then, x = sin"1 y, or on interchanging x and y, y = sin"1 x., 6. Hyperbolic Functions. These are defined in terms of exponential functions as follows., , Some fundamental identities analogous to those for trigonometric functions are, (Z>) cosh2 x — sinh2 x = I, sech2a; + tanh2a; = 1, coth2 x — csch2 x = 1, (c) sinh (a; ± y) = sinh x cosh y ± cosh x sinh y, cosh (a; ± y) = cosh x cosh y ± sinh x sinh y, , tanh(x±y) —, The inverse hyperbolic functions, given by sinh"1 a;, cosh"1 a;, etc. can be expressed, in terms of logarithms [see Problem 1.9, for example]., LIMITS, , The function f(x) is said to have the limit I as x approaches a, abbreviated lim f(x) = I,, if given any number e > 0 we can find a number 8 > 0 such that |/(a;) — Z| < « whenever, 0< |*-a| <8., Note that \p\, i.e. the absolute value of p, is equal to p if p > 0, — p if p < 0 and 0 if, p = 0.
Page 14 :
4, , REVIEW OF FUNDAMENTAL CONCEPTS, , If, , [CHAP. 1, , then we have the following theorems on limits., , CONTINUITY, , The function f(x) is said to be continuous at a if lim f(x) = /(a)., z-»a, , 2, , Example 3. f(x) = x — 4x + 8 is continuous at x = 1., , However, if f(x) =, , then f(x) is not continuous [or is discontinuous] at x = 2 and x = 2 is called a, discontinuity of /(a;)., , If f(x) is continuous at each point of an interval such as Xi ^ x ^ x¡ or x\ < x ë a;2,, etc., it is said to be continuous in the interval., If fi(x) and fa(x) are continuous in an interval then fi(x) ± /z(a;), fi(x)fz(x), and, f\(x)lfz(x) where /2(a;) ¥= 0 are also continuous in the interval., DERIVATIVES, , The derivative of y = /(a;) at a point a; is defined as, (5), , where, The differential, , provided the limit exists., of y = f(x) is defined by, dy = /'(a;) dx where da; = Aa;, , (6), , The process of finding derivatives is called differentiation., By taking derivatives of, y' = dy/dx — f'(x) we can find second, third and higher order derivatives, denoted by, y" = d?yldxz = f"(x), y'" = d3y/dx3 = f'"(x), etc., Geometrically the derivative of a function f(x) at a point represents the slope of the, tangent line drawn to the curve y — f(x) at the point., If a function has a derivative at a point, then it is continuous at the point. However,, the converse is not necessarily true., DIFFERENTIATION FORMULAS, , In the following u, v represent functions of x while a, c, p represent constants. We, assume of course that the derivatives of u and v exist, i.e. u and v are differentiable.
Page 15 :
CHAP. 1], , REVIEW OF FUNDAMENTAL CONCEPTS, , 5, , In the special case where u = x, the above formulas are simplified since in such case, du/dx = 1., INTEGRALS, , If dy/dx = f(x), then we call y an indefinite integral or anti-derivative of f(x) and, denote it by, (7), , Since the derivative of a constant is zero, all indefinite integrals of f(x) can differ only by, a constant., The definite integral of f(x) between x = a and x = b is defined as, («), , provided this limit exists. Geometrically if f(x) =£ 0, this represents the area under the, curve y = f(x) bounded by the x axis and the ordinates at x — a and x = b. The integral, will exist if f(x) is continuous in a a x a b., Definite and indefinite integrals are related by the following theorem., Theorem 1-1 [Fundamental Theorem of Calculus]., , If f(x) = -p g(x), then, , Example 4., , The process of finding integrals is called integration., INTEGRATION FORMULAS, , In the following u,v represent functions of x while a,b,c,p represent constants. In, all cases we omit the constant of integration, which nevertheless is implied., 2, , 1, , 3., , or, , This is called integration by parts.
Page 16 :
6, , REVIEW OF FUNDAMENTAL CONCEPTS, , [CHAP. 1, , 4., where w = u(x) and w' = dw/dx expressed as a, function of w. This is called integration by substitution or transformation., 5., , 14, , 6., , 15, , 7., , 16, , 8., , 17, , 9., , 18, , 10., , 19, , 11., , 20, , 12., , 21, , 13., , 22, , SEQUENCES AND SERIES, A sequence, indicated by u\, u2, ... or briefly by (un), is a function defined on the set of, natural numbers. The sequence is said to have the limit I or to converge to I, if given any, « > 0 there exists a number N > 0 such that, for all n > N, and in such case, we write lim un = I. If the sequence does not converge we say that it diverges., n-* «, , Consider the sequence u\, Ui + u2, Ui + u2 + us, ... or Si, Si, Sa, ... where Sn =, Ui + uz + • • • +Un. We call <Sn) the sequence of partial sums of the sequence (un). The, symbol, Ui + uz + us + • • • or, or briefly, (9), is defined as synonymous with (Sn) and is called an infinite series. This series will converge, or diverge according as (Sn) converges or diverges. If it converges to S we call S the sum, of the series., The following are some important theorems concerning infinite series., Theorem 1-2. The series, Theorem 1-3. If, Theorem 1-4., , If, , converges if p > 1 and diverges if p a l ., , converges and, converges, then, , then, , converges., , converges., , In such case we say that 2wn converges absolutely or is absolutely convergent. A property, of such series is that the terms can be rearranged without affecting the sum., Theorem 1-5. If, , diverges a n d t h e n 2vn diverges.
Page 17 :
CHAP. 1], , REVIEW OF FUNDAMENTAL CONCEPTS, , Theorem 1-6. The series, , 7, , w h e r e c o n v e r g e s or diverges according as, exists or does not exist., , This theorem is often called the integral test., Theorem 1-7. The series, diverges if, However, if, series may or may not converge [see Problem 1.31]., , the, , Theorem 1-8. Suppose that, Then the series 2un converges (absolutely), if r < 1 and diverges if r > 1. If r = 1, no conclusion can be drawn., This theorem is often referred to as the ratio test., The above ideas can be extended to the case where the un are functions of x denoted, by Un(x). In such case the sequences or series will converge or diverge according to the, particular values of x. The set of values of x for which a sequence or series converges is, called the region of convergence, denoted by ^., Example 5. The series 1 + x + xz + x3 + • • • has a region of convergence 'K. [in this case an, interval] given by — 1 < * < 1 if we restrict ourselves to real values of x., , UNIFORM CONVERGENCE, , We can say that the series ui(x) + Uz(x) + • • • converges to the sum S(x) in a region "3^, if given e > 0 there exists a number N, which in general depends on both « and x, such, that |S(a;) — Sn(x}\ < f whenever n>N where Sn(x) = Ui(x)+ • • • + Um(x). If we can find, N depending only on e and not on x, we say that the series converges uniformly to S(x) in, % Uniformly convergent series have many important advantages as indicated in the, following theorems., Theorem 1-9. If Un(x), n = 1,2,3, ... are continuous in a ë x à & and 2un(x) is uniformly convergent to S(x) in a a x a b, then S(x) is continuous in, a = x = 6., Theorem 1-10. If ~S.Un(x) converges uniformly to S(x) in a ë x a b and un(x), n = 1,2,3, ..., are integrable in a ë x ë b, then, , Theorem 1-11. If Un(x], n — 1,2,3,... are continuous and have continuous derivatives in, a ë x ë b and if ~S,un(x) converges to S(x) while 2ui(x) is uniformly convergent in a a x ë b, then, , An important test for uniform convergence, often called the Weierstrass M test, is, given by the following., Theorem 1-12. If there is a set of positive constants Mn, n=l,2,3, ... such that, in "K and, converges, then, is uniformly convergent, [and also absolutely convergent] in <^.
Page 18 :
8, , REVIEW OF FUNDAMENTAL CONCEPTS, , [CHAP. 1, , TAYLOR SERIES, , The Taylor series for f(x) about x = a is defined as, (10), , where, , *o between a and x, , (11), , is called the remainder and where it is supposed that f(x) has derivatives of order n at, least. The case where n = 1 is often called the law of the mean or mean-value theorem, and can be written as, xo between a and x, (12), The infinite series corresponding to (10), also called the formal Taylor series for f(x),, will converge in some interval if limí¿n = 0 in this interval. Some important Taylor, n»+°o, series together with their intervals of convergence are as follows., 1., 2., 3., 4., 5., , A series of the form, is often called a power series. Such power series are, uniformly convergent in any interval which lies entirely within the interval of convergence, [see Problem 1.120],, FUNCTIONS OF TWO OR MORE VARIABLES, The concept of function of one variable given on page 2 can be extended to functions of, two or more variables. Thus for example z = f(x, y) defines a function / which assigns, to the number pair (a;, y) the number z., Example 6. If f(x, y) - x2 + Sxy + 2y*, then /(-1,2) = (-1)2 + 3(-l)(2) + 2(2)2 = 3., , The student is familiar with graphing 2 = f(x, y) in a 3-dimensional xyz coordinate, system to obtain a surface. We sometimes call x and y independent variables and z a, dependent variable. Occasionally we write z = z(x, y) rather than z — f(x, y), using the, symbol z in two different senses. However, no confusion should result., The ideas of limits and continuity for functions of two or more variables pattern closely, those for one variable., PARTIAL DERIVATIVES, , The partial derivatives of f(x, y) with respect to x and y are defined by, (13)
Page 19 :
CHAP. 1], , REVIEW OF FUNDAMENTAL CONCEPTS, , 9, , if these limits exist. We often write, Note that df/dx is simply the ordinary, derivative of / with respect to x keeping y constant, while df/dy is the ordinary derivative of, / with respect to y keeping x constant. Thus the usual differentiation formulas on pages, 4 and 5 apply., Example 7. If f(x, y) = 3x2 - 4xy + 2j/2 then, , Higher derivatives are defined similarly., derivatives, , For example, we have the second order, (U), , The derivatives in (13) are sometimes denoted by fx and fy. In such case fx(a,b), fv(a,b), denote these partial derivatives evaluated at (a, b). Similarly the derivatives in (14) are, denoted by fxx, fxv, fyx, fm respectively. The second and third results in (14) will be the same, if / has continuous partial derivatives of second order at least., The differential of f(x, y) is defined as, (15), , where, Generalizations of these results are easily made., TAYLOR SERIES FOR FUNCTIONS OF TWO OR MORE VARIABLES, , The ideas involved in Taylor series for functions of one variable can be generalized., For example, the Taylor series for f(x, y) about x = a, y — b can be written, , (16), LINEAR EQUATIONS AND DETERMINANTS, , Consider the system of linear equations, (17), , These represent two lines in the xy plane, and in general will meet in a point whose coordinates (x,y) are found by solving (17) simultaneously. We find, , (18), It is convenient to write these in determinant form as, (19), where we define a determinant of the second order or order 2 to be, , (20)
Page 20 :
10, , REVIEW OF FUNDAMENTAL CONCEPTS, , [CHAP. 1, , Note that the denominator for x and y in (19) is the determinant consisting of the coefficients, of * and y in (17). The numerator for x is found by replacing the first column of the, denominator by the constants c\, cz on the right side of (17). Similarly the numerator for, y is found by replacing the second column of the denominator by ci, c2. This procedure is, often called Cramer's rule. In case the denominator in (19) is zero, the two lines represented, by (17) do not meet in one point but are either coincident or parallel., The ideas are easily extended. Thus consider the equations, (21), , representing 3 planes. If they intersect in a point, the coordinates (x, y, z) of this point are, found from Cramer's rule to be, , (22), , where we define the determinant of order 3 by, =, , 01&2C3 + &iC2a3 + CiO2&3 — (biases + aiC2&3 + Ci62as), , (23), , The result (23) can be remembered by using the scheme of repeating the first two columns, as follows, , (24), , and taking the sums of the products of terms as indicated by the arrows marked + and —., The determinant can also be evaluated in terms of second order determinants as follows, (**), , where it is noted that a\, b\, c\ are the elements in the first row and the corresponding second, order determinants are those obtained from the given third order determinant by removing, the row and column in which the element appears., The general theory of determinants, of which the above results are special cases, is, considered in Chapter 15.
Page 21 :
CHAP. 1], , REVIEW OF FUNDAMENTAL CONCEPTS, , 11, , MAXIMA AND MINIMA, If for all x such that \x - a\ < 8 we have f(x) S /(a) [or f(x) g /(a)], we say that /(a), is a relative maximum [or relative minimum]. For f(x) to have a relative maximum or, minimum at x = a, we must have /'(a) = 0. Then if /"(a) < 0 it is a relative maximum, while if /"(a) > 0 it is a relative minimum. Note that possible points at which f(x) has, a relative maxima or minima are obtained by solving f'(x) = 0, i.e. by finding the values, of x where the slope of the graph of f(x) is equal to zero., Similarly f(x,y) has a relative maximum or minimum at x = a, y = b if fx(a, b) = 0,, fv(a,b) = 0. Thus possible points at which f(x,y) has relative maxima or minima are, obtained by solving simultaneously the equations, (26), Extensions to functions of more than two variables are similar., METHOD OF LAGRANGE MULTIPLIERS, Sometimes we wish to find the relative maxima or minima of f(x, y) = 0 subject to, some constraint condition </>(«, y) = 0. To do this we form the function h(x, y) = f(x, y) +, \4>(x, handset, \*i), The constant A. is called a Lagrange multiplier and the method is called the method of, Lagrange multipliers. Generalizations can be made [see Problems 1.54 and 1.150]., LEIBNITZ'S RULE FOR DIFFERENTIATING AN INTEGRAL, Let, , (28), , where / is supposed continuous and differentiable. Then Leibnitz's rule states that if a and, b are differentiable functions of a,, (29), , MULTIPLE INTEGRALS, A generalization of the integral for functions of one variable leads to the idea of, multiple integrals for functions of two or more variables. Because several ideas involved, in the theory may be new to some students, we postpone consideration of this topic to, Chapter 6., COMPLEX NUMBERS, Complex numbers arose in order to solve polynomial equations such as x2 +1 = 0 or, 2, x + x +1 = 0 which are not satisfied by real numbers. We assume that a complex number, has the form a + bi where a, b are real numbers and i, called the imaginary unit, has the, property that i2 = — 1. We define operations with complex numbers as follows., 1. Addition., , (a + bi) + (c + di) = (a + c) + (b + d)i, , 2. Subtraction., , (a + bi) — (c + di) = (a — c) + (b — d)i, , 3. Multiplication., , (a + bi)(c + di) = ac + adi + bci + bd& = (ac - bd) + (ad + bc)i, , 4. Division.
Page 22 :
REVIEW OF FUNDAMENTAL CONCEPTS, , 12, , [CHAP. 1, , Note that we have used the ordinary rules of algebra except that we replace i2 by —1, wherever it occurs. The commutative, associative and distributive laws of page 1 also, apply to complex numbers. We call a and b of a + bi the real and imaginary parts respectively. Two complex numbers are equal if and only if their real and imaginary parts are, respectively equal., A complex number z = x + iy can be considered as a, point P with coordinates (x,y) on a rectangular xy plane, called in this case the complex plane or Argand diagram, [Fig. 1-4]. If we construct the line from origin 0 to P and, let p be the distance OP and <f> the angle made by OP with, the positive x axis, we have from Fig. 1-4, , (30), and can write the complex number in so-called polar form as, , Fig. 1-4, (31), , We often call p the modulus or absolute value of z and denote it by \z\. The angle, called the amplitude or argument of z abbreviated argz. We can also write, where z = x — iy is called the conjugate of z = x + iy., , is, , If we write two complex numbers in polar form as, (32), , T H E N, , (33), (34), , Also if n is any real number, we have, (35), , which is often called De Moivre's theorem. We can use this to determine roots of complex, numbers. For example if « is a positive integer,, k = Q, 1, 2, . . . , n - l, , (36), , Using the series for ex, sin x, cos x on page 8, we are led to define, (37), , which are called Euler's formulas and which enable us to rewrite equations (31)-(36) in, terms of exponentials., Many of the ideas presented in this chapter involving real numbers can be extended, to complex numbers. These ideas are developed in Chapter 13.
Page 23 :
CHAP. 1], , REVIEW OF FUNDAMENTAL CONCEPTS, , 13, , Solved Problems, REAL NUMBERS AND LAWS OF ALGEBRA, 1.1. Prove that, is an irrational number., Assume the contrary, i.e. suppose that, where p and q are positive integers having, no common integer factor other than 1 [in such case we say that p/q is a fraction in lowest terms}., Squaring, 2 = p2/q2 or p2 = 2g2. Then p2 is even and so p must also be even. Thus p = 2m, where m is a positive integer. Substituting into p2 = 2<?2 we, find, so that q2 is even and, thus q is even, i.e. q = 2n where n is a positive integer. Since p and q are both even they have, 2 as a common factor, violating our assumption that thev have no common integer factor other than, 1. This contradiction shows that the assumption that, is rational is incorrect and proves that, is irrational., , 1.2., , Which is larger, Assume, It follows that, , 1.3., , Then raising each side to the 6th power we have 23 a 32 which is wrong., , Assuming that the real numbers a, b, c satisfy the rules of algebra on page 1, prove, that (6 + c)a = ba + ca., From rule 5 on page 1 we have a(b + c) = ab + ac. But by rule 3 it follows that a(b + c) —, (b + c)a, ab — ba, ac = ca. Thus (6 + c)a = ba + ca., , FUNCTIONS, 1.4. If f(x) = 2x3 - 3x + 5 find (a) /(-I), (b) f(0), (c) f(x + h)., (a) /(-I) = 2(-l)3 - 3(-l) + 5 = 2(-l) + 3 + 5 = 6, , 1.5., , (6), , /(O) = 2(0)8 _ 3(0) + 5 = 0 + 0 + 5 = 5, , (c), , f(x + h) - 2(x + h)3 - 3(œ + h) + 5 = 2(x3 + 3x2h + Sxh2 + h3) - 3x - 3h + 5, = 2a;« + 6x2h + 6xh2 + 2h3 - 3x - 3h + 5, , Using the rules of exponents (1) on page 2, prove the rules of logarithms (4) on, page 2., By definition, if ex — m then x = In m. Similarly if ev = n then y — In n., Since ex'e" = ex+y we have mn — ex+v or * + y = In (mri), i.e. In (mn) — In m + In n., Since e*/e» = e*-" we have m/n — ex~y or * — y — In (m/n), i.e. In (m/n) = In m — In n., Since (ex)v = exp we have m" = e*? or xp = In mv, i.e. In m? — p In m., , 1.6., , Prove that, Since cos (* + y) — cos * cos y — sin x sin y we have on letting y — x,, Also we have, , cos2« — sin2» = cos 2*, , (Í), , cos2» + sin2* = 1, , ia\, V*'/, , From (J) and (2) the required results follow by subtraction and addition respectively.
Page 24 :
14, , 1.7., , [CHAP. 1, , REVIEW OF FUNDAMENTAL CONCEPTS, , Prove that, , where, , We have, (1), Letting, , [see Fig. 1-5], (1) becomes, , as required., , 1.8., , Fig. 1-5, , Prove that (a) cosh2 x - sinh2 x = 1, (6) sech2 x + tanh2 x = 1., (a) By definition,, , (6) Dividing both sides of the result in (a) by cosh2 x, we have, , Since sinh «/cosh x — tanh x, 1/cosh x = sech x., , 1.9., , From this we obtain, , Prove that, By definition, if, then x — cosh, e2» —2»e» + 1 = 0. Solving this as a quadratic equation in'e», we find, , or, , Thus, , or, , Since, In, , ), , the required result follows. Note that we must have » S 1 if y = cosh"1 x is to be real., , LIMITS AND CONTINUITY, 1.10. If, , prove that, , (a) We must show that given any, when, Choose, so that, , we can, , find, , [depending on « in general] such that, , Then, , where we have used the result that, Take S as 1 or e/5, whichever is smaller., and the required result is proved., , Then we have, , whenever, , (6) There is no difference between the proof for this case and the proof in (a) since in both cases, we exclude * = 2.
Page 25 :
CHAP. 1], , 1.11., , REVIEW OF FUNDAMENTAL CONCEPTS, , Prove that if, , and, , then, , We must show that for any e > 0 we can find 8 > 0 such that, when, We have, Now by hypothesis, given, , 15, , we can, , find, , and, , U), , such that, (2), (Í), , Then from (1), (2) and (S) we have, , when, , where 8 is chosen as the smaller of Sj and S2., In a similar manner we can prove other theorems on limits such as, if, See ProblenTl.75., , 1.12. Prove that ( a ) f(x) = x2 i s continuous a t x = 2 while, continuous at x = 2., (a) Method 1., By Problem 1.10(a),, , IS NOT, , and so f(x) is continuous at x = 2., , Method 2., , We must show that given any, that, Problem 1.10., , when, , we can, , find, , [depending on e in general] such, The proof patterns that given in, , (6) Since /(2) = 0 we h a v e s o that /(*) is not continuous [or is discontinuous], at x = 2. We can also give a proof involving e, 8 methods by showing that given any, w e cannot, find, such t h a t w h e n, , DERIVATIVES, , 1.13. Prove that if u and v are differentiate functions of x, then, (a), (a) Write, Then by definition, if, , (b), , since, , (6), we have, , Denote u(x), v(x) briefly by u, v respectively.
Page 26 :
16, , REVIEW OF FUNDAMENTAL CONCEPTS, , [CHAP. 1, , 1.14. Prove that if f(x) has a derivative at the point a, then f(x) is continuous at a., We have, , Then if the derivative exists,, , Thus, , and the required result follows., , 1.15. Prove that if p is any positive integer and u is a differentiate function of x, then, By definition, , The result can also be shown to hold for all values of p., , 1.16.Assuming that, , prove that, , (b), , (<0, (a), , where we have used the result of Problem 1.6(a)., ( 6 ) From part ( a ) w e h, we find, , (C) By differentation formula 4, page 4, we have, , THEN LETTING, , SO THAT
Page 27 :
CHAP. 1], , REVIEW OF FUNDAMENTAL CONCEPTS, , 1.17. Prove that, Then by Problem 1.16(c),, , then, , Thus, , 1.18., , prove that, , Given that, , We have assumed here that, , which can be demonstrated by showing that In u is continuous., then, , 1.19., , Thus by part (a),, , Find, , If, , find, Consider y as a function of x and differentiate both sides with respect to x. Then, , or, , i.e., and, , 17
Page 28 :
18, , EEVIEW OF FUNDAMENTAL CONCEPTS, , [CHAP. 1, , then, , 1.21. Show that if, We have, , Then, , 1.22. Find the differentials of, (a), (&), , INTEGRALS, 1.23. Prove that, where constants of integration are omitted., (a), , Since, , (6), , Since, , (c), , If, , then by definition, and so F = G + H apart from a constant, , of integration and the required result follows., , (d), , 1.24. Find, (a), , Let, , (b) Let, , (c), , Let, , so that, , so that, , Then, , Then, , so that, , Then
Page 29 :
CHAP. 1], , REVIEW OF FUNDAMENTAL CONCEPTS, , 19, , 1.25. (a) Prove the formula for integration by parts, and (6) use the formula to find, (a) Since, , so that, (6) Let u = x, dv = sin2xdx., parts,, , Then du = dx, v = -lcos2x. Thus applying integration by, , 1.26. Evaluate (a), (a) Let u = x2 + 1 so that du = 2x dx. Then, , Thus, , (6) Let u = 3x so that du = 3 dx. Then, , Thus, In practice when evaluating definite integrals, we omit the constant of integration c since, it does not enter anyway., , 1.27. Find the area under the curve y = sin x between x = 0 and x = » [see Fie 1-2, page, Area =, , 1.28. (a) By using definition (8) page 5 for, the definite integral, obtain an approximate value for, , and, , give a geometric interpretation., (6) Show how the result in (a) can be, improved., (a) Let o = l, 6 = 2 and divide the interval from a to 6 of length b — a — 1 into, 10 equal parts so that n = 10, h =, [see Fig. 1-6]. We, have approximatel, , Fig.1-6
Page 30 :
REVIEW OF FUNDAMENTAL CONCEPTS, , 20, , [CHAP. 1, , The correct value is In a = .6932. Note that each term of the approximating sum repres-6-6ents, one of the 10 rectangular shaped areas shaded in Fig. 1-6., (6) The result obtained in (a) overestimates the true value since the upper edge of each rectangle, lies above the curve y = í/x. To obtain an underestimate we use the lower rectangles. In, this way we find approximately, , As a better estimate of the integral we take the arithmetic mean of the overestimate and the, underestimate, i.e. £(.7188 + .6688) = .6938 which compares very well with the correct value, .6932. Since the arithmetic mean of the areas of any upper rectangle and corresponding, lower rectangle is the area of a trapezoid, this method of averaging is often called the, trapezoidal rule., , SEQUENCES AND SERIES, 1.29. (a) Find the limit of the sequence .3, .33, .333, ... and (6) justify your conclusions,, (a) The nth term of the sequence can be written, , Now if, , then, so that by subtracting we have, , Thus the nth term is, (6) To give a proof that 1/3 is in fact the required limit we must show that given e > 0 we can, find N [depending on e] such that \un — £| < e for n > N. Now, , when, Thus we have found the required N and the statement that the limit is 1/3 is proved., , 1.30. Show that the series 1-1 + 1-1 + 1-1+, , ••• does not converge., , The sequence of partial sums is 1, 1 - 1, 1 — 1 + 1, 1 — 1 + 1 — 1, ... or 1, 0, 1, 0, ...., this sequence does not converge, the given series does not converge., , 1.31., , Since, , Use the integral test, page 7, to show t h a t ( a ) converges if p > 1 and, (&) diverges if p a l ., (a) Using f(ri) = 1/n* we have f(x) - 1/x" so that if p •£ 1,, , Now if p > 1 this limit exists and the corresponding series converges.
Page 31 :
REVIEW OF FUNDAMENTAL CONCEPTS, , CHAP. 1], , 21, , (6) If p < 1 the limit does not exist and the series diverges. If p = 1 then, , which does not exist and so the corresponding series for p = 1 diverges., This shows that, diverges even though the nth term approaches zero., [See Theorem 1-7, page 7.], , 1.32. Investigate the convergence of, The series of absolute values is, and converges by Problem 1.31, since p = 2. Thus the series converges absolutely and so converges by Theorem 1-4, page 6., , 1.33. Prove that the series, , converges for, n, , x, The «th term of the series is un = (—l)""1-^. Now, , Then by Theorem 1-8, page 7, the series converges if \x\ < 1, i.e. — ! < * < ! and diverges for, |*| > 1. For |*| = 1, i.e. x = ±1, no conclusion can be drawn. However, for x = 1 and —1 the, series becomes, which are absolutely convergent and thus convergent. It follows that the series converges for, -1 S * S 1., , UNIFORM CONVERGENCE, , 1.34. Investigate the uniform convergence of, for -1 ë a; ë 1., , Method 1., , Let, , If * = 0, then, if, then, [see Problem 1.110(a)]., , =, — 0., using the fact that the series is a geometric srries, , Then, Since each term of the series is continuous while the sum function S(x) is discontinuous at x = 0,, it follows by Theorem 1-9, page 7, that the series cannot be uniformly convergent in —1 S x £ 1., Method 2., , We have from the results of Method 1,, , Then, However, since we, cannot find an N which holds for all a; in -1 g * a 1 [consider x - 0 for example], it follows, that the series cannot be uniformly convergent in —1 S a; S 1.
Page 32 :
22, , REVIEW OF FUNDAMENTAL CONCEPTS, , [CHAP. 1, , 1.35. Prove Theorem 1-9, page 7., We must show that S(x) is continuous in a ë x S b., Now S(x) = Sn(x) + Rn(x), so that S(* + h) = Sn(x + h) + Rn(x + h) and thus, S(x + h)- S(x) = Sn(x + h)- Sn(x) + Rn(x + h)- Rn(x), , (1), , where we choose h so that both x and x + h lie in a a * S b [if * = b, for example, this will, require h < 0]., Since Sn(x) is a sum of a finite number of continuous functions, it must also be continuous., Then given « > 0, we can find 8 so that, whenever, (2), Since the series, by hypothesis, is uniformly convergent, we can choose N so that, (S), Then from (1), (2) and (5),, |S(* + /i)-S(aO| S, , \Sn(x + h) -Sn(x)\ + \Rn(x + h)\ + \Rn(x)\, , < e, , for \h\ < S, and so the continuity is established., , 1.36. Prove Theorem 1-10, page 7., If a function is continuous in a S x S b, its integral exists. Then since S(x), Sn(x) and Rn(x), are continuous,, To prove the theorem we must show that, , can be made arbitrarily small by choosing n large enough. This, however, follows at once, since by, the uniform convergence of the series we can make |iZn(*)| < t/(b — a) for n > N independent of, x in [a, b], and so, , This is equivalent to the statements, , 1.37. Prove the Weierstrass M test [Theorem 1-12, page 7]., Since 2MB converges, we have, Then, Since N is independent of x, the series must be uniformly convergent., , 1.38. Prove thai, We have, , is uniformly convergent for, Since, , given series is uniformly convergent for, , converges, it follows by the Weierstrass M test that the, [and in fact any finite interval].
Page 33 :
CHAP. 1], , REVIEW OF FUNDAMENTAL CONCEPTS, , 23, , TAYLOR SERIES, , 1.39. Obtain the formal Taylor series for sin x about x = 0., Let /(*) = sin «. Then, so that, Thus from, with o = 0, we have, , 1.40. Use series methods to obtain the approximate value of, From the expansion for ex on page 8 we have on replacing, , Then, Since this series is uniformly convergent for 0 = x = I, 1.120], we can integrate term by term to obtain, , [as can be proved by using Problem, , approx., , FUNCTIONS OF TWO OR MORE VARIABLES AND PARTIAL DERIVATIVES, 1.41. If, , find, , (e), (a), (6), , Then, , (c), , Then, , (d), , Then, , W, , Then, , (/), , Then, , (g), , Then, , Note that the results in (e) and (/) illustrate the fact that fxv = fyx since in this case / is, continuous and has continuous partial derivatives up to and including the second order.
Page 34 :
24, , REVIEW OF FUNDAMENTAL CONCEPTS, , 1.42. If f(x,y) = sin(x2 + 2y), find (a) /*, (6) /„, (c) /**, (d) /„, (e) /„*, (/)/«„., (a), (6), (c), , (d), (e), , (/), , Note that, , in (d) and (e)., , 1.43. If f(x,y) = Zx* + ±xy-2y2, find df., Method 1., Method 2., , 1.44. If 2 = f(ylx),, , show that, , Method 1., Let u — y/x so that, , Then, Method 2., Then since, , it follows on equating coefficients of dx and dy that, , and the result follows as in Method 1., , [CHAP. 1
Page 35 :
25, , REVIEW OF FUNDAMENTAL CONCEPTS, , CHAP. 1], , 1.45. If dz = M(x, y) dx + N(x, y) dy, prove that dM/dy = dN/dx, assumed to have continuous partial derivatives., Since, x and y are independent variables., Then since, , w e must h, , where M and N are, a, , v, , e, , s, , i, , n, , c, , e, , [if these derivatives are continuous], we have, or, , Similarly we can prove that if dM/dy - SNIdx then Max + Ndy can be written as the differential of z, i.e. dz, often called an exact differential., , 1.46. If z = z(u, v) where u = u(x, y), v = v(x, y), prove that (a), , (b), From z = z(u, v), u = u(x, y), v = v(x, y) we have, (S), Then using (2) and (3) in (1) we find on combining terms,, (*), , But considering that « is a function of x and y since u,v are functions of x and y, we must have, (5), Equating corresponding coefficients of dx and dy in (4) and (5) yields the required results., , TAYLOR SERIES FOR FUNCTIONS OF TWO OR MORE VARIABLES, 1.47. Verify the result (16) page 9 for the function of Problem 1.41., From Problem 1.41 we have /(2,-3) = -30, /s(2,-8) = 0, /v(2,-3) = 20, fxx(2, -3) = 6,, fxy(2,— 3) = fyx(2, — 3) = 4, /w(2, — 3) = —4. Also the higher derivatives are all zero. Then from, (18), page 9. we should have, I, Now the right side can be written, -30 + 20y + 60 + 3(*2 - 4* + 4) + 4(xy - 2y + Sx - 6) - 2(j/2 + By + 9) = 3»2 + 4*y - 2j/2, and the result is verified in this special case., , LINEAR EQUATIONS AND DETERMINANTS, 1.48. Verify that the solutions of (17), page 9, are given by (18)., We have, (1) a^x + bty = clt, (2) atfc + b& = C2, Multiply (1) by 62 to obtain, Multiply (2) by 6j to obtain, Then by subtraction, («162 — 61o2)w = cfo — 6tc2 which gives the required result for *. Similarly on multiplying (1) by a^, (2) by t^ and subtracting we obtain the result for y., , 1.49. Solve, , using determinants., , By Cramer's rule we have, , A check is supplied by substituting these results in the given equations.
Page 36 :
[CHAP. 1, , REVIEW OF FUNDAMENTAL CONCEPTS, , 26, , 1.50. Solve, , for z., , By Cramer's rule we have, , 1.51. (a) For what value of fc will the system of equations, , have solutions other than the trivial one x = O, y = 0., solutions., , (&) Find two non-trivial, , (a) By Cramer's rule the solution will be, , Now since the numerators are equal to zero, these can be non-trivial [i.e. nonzero] solutions, only if the denominator is also equal to zero, i.e., >, , (6) If k = 2 the equations become — x + y = 0, 2x — 2y = 0 and are identical, i.e. x = y. Then, solutions are x — 2, y = 2, x = 3, y — S for example. Actually there are infinitely many, such non-trivial solutions., , MAXIMA AND MINIMA. METHOD OF LAGRANGE MULTIPLIERS, 1.52. Find the relative maxima and minima of f(x) = x* - Sx3 + 22a;2 - 24a; + 20., The relative maxima and minima occur where, f ( x ) = 4a;3 - 24a;2 + 44a; - 24 = 0 or (x - l)(x - 2)(x - 3) = 0,, , i.e. x = 1,2,3, , 2, , Then since f"(x) = 12a; - 48a; + 44, we have /"(I) = 8 > 0, /"(2) = -4 < 0, /"(3) = 8 > 0. Thus, a relative minimum of 11 occurs at x = 1, a relative maximum of 12 occurs at x = 2 and a, relative minimum of 11 occurs at x = 3., , 1.53. Determine the dimensions of the largest rectangular parallelepiped which can be inscribed, in a hemisphere of radius a [Fig. 1-7]., The volume .of the parallelepiped is, V = (2x)(2y)(z) = 4*2/Z, and the equation of the surface of the hemisphere is, x2 + y2 + 22 = a2 orThe volume is a maximum where V2 — U =, 2, 2, X2y2z2 = xW(a? - x - j/ ) is a maximum. To find, this we solve simultaneously the equations, , Fig. 1-7
Page 37 :
CHAP. 1], , REVIEW OF FUNDAMENTAL CONCEPTS, , 27, , Since x ¥* O, y ¥= 0, these give, , x2 + 2j/2 = a?, , 2*2 + y2 = <&,, from which, Then the required volume is, , 1.54., , (a) Prove the method of Lagrange multipliers for the case where the function, f(x,y) is to be made a maximum or minimum [or extremum] subject to the constraint condition <f>(x, y) = 0., (6) Show how to generalize the result of (a) to the case where the function to be made, an extremum is f(x, y, z) and the constraint condition is $(x, y, z) = 0., (c) Illustrate the method of Lagrange multipliers by working Problem 1.53., (a) Assume that </>(x, y) = 0 defines y as a unique function of x, i.e. y = g(«), having a continuous, derivative g'(x). Then we must find the maximum or minimum [extremum] of, , f(x,y) =, , f(x,g(x)), , But as in elementary calculus this can be found by setting the derivative with respect to x, equal to zero, i.e., or, (J), Also from, , we have the identity <j>(x, g(x)) = 0 so that, or, , Eliminating, , (2), , between (1) and (2) we find, (*), , assuming, , Now if we define, , or, , W, (3) becomes, , (5), , But (4) and (5) are obtained by considering h(x, y) = f(x, y) + \<f>(x, y) and setting, («), , so that the method is proved. We call \ the Lagrange multiplier., (b) In this case we assume that, , can be solved to yield, , I so that, , But the extremum of this function of x and y can be found by setting the partial derivatives, with respect to x and y equal to zero, i.e., o r, , ( 7 ), , or, , (8), , Also the constraint condition yields the identity <f>(x, y, g(x, y)) = 0 so that by differentiation, with respect to x and y we have, o r, , (9), , or, , (io)
Page 38 :
28, , REVIEW OF FUNDAMENTAL CONCEPTS, , [CHAP. 1, , Eliminating gx and gy among the equations (7), (8), (9), (10) we find, assuming <f>z ¥* 0,, , (11), Then c a l l i n g o r, , (12), , equations (11) yield, f, , (13), , But (i#) and (i#) can be obtained by letting h(x, y, z) = f ( x , y, z) + \<j>(x, y, z) and setting the, derivative with respect to x,y,z equal to zero so that the required result is proved., (c) We must find the maximum of 4xyz subject to the constraint condition x2 + y2 + z2 — o2 = 0., To do this we form the function, h(x,y,z) = 4*2/2 + \(xz + y2 + z2 - a2), (14), and set the partial derivatives of h with respect to x, y, z equal to zero. Thus we obtain, , (IS), Multiplying these by x, y, z respectively and adding, we find, I2xyz + 2\(x2 + y2 + z2) = Í2xyz + 2Xa2 = 0, or X = —6xyz/a2. Then substituting this value of X in equations (15) and solving simultaneously, with x2 + y2 + z2 = a2, we find, x = y = z = a/v/S, as in Problem 1.53., , LEIBNITZ'S RULE, 1.55. If, , COMPLEX NUMBERS, 1.56. Perform the indicated operations., (a) (4 - 2i) + (-6 + 5i) = 4 - 2i - 6 + 5i = 4 - 6 + (-2 + 5)i = -2 + Si, (b) (-7 + Si) - (2 - 4i) = -7 + Si - 2 + 4i = -9 + 7i, (c) (3 - 2i)(l + 3f) = 3(1 + Si) - 2i(l + 3t) = 3 + 9i - 2i - 6i2 = 3 + 9i - 2i + 6 = 9 + 7i, (d), , («)
Page 39 :
29, , REVIEW OF FUNDAMENTAL CONCEPTS, , CHAP. 1], , (/), , to), 1.57., , If Zi and z2 are two complex numbers, prove that \ZiZz\ — \z\\ \zz\., Let «! = *! + iyi, z% = xz + iy2. Then, , 1.58. Solve a; 3 -2x-4 = 0., The possible rational roots are ±1, ±2, ±4. By trial we find x = 2 is a root. Then the given, equation can be written (x — 2)(*2 + 2x + 2) = 0. The solutions to the quadratic equation, For, , o = l,, , 6 = 2, e = 2, , this gives, , The set of solutions is 2, —1 + i, — 1 — i., , POLAR FORM OF COMPLEX NUMBERS, , 1.59. Express in polar form (a) 3 + Si, (b) -l + \fèi, (c) -1, (d) -2-2v/3i, , Fig. 1-8, (a) Amplitude 0 = 45° = ;r/4 radians. Modulus p -, , Then, , (6) A m p l i t u d e = 2ir/3 radians. Modulus, , Then, , -1 + \/3i = 2(cos2)r/3 + i sin2îr/3) = 2 cis2jr/3 = 26^/3, (c)A m p l i t u d e =, , v, , radians. M, , o, , d, , u, , l, , u, , s, , T, , h, , e, , n, , —1 = l(cos IT + i sin JT) = eis 7T = e^, (d) Amplitude, , 1.60. Evaluate (a), , = 4r/3 radians. Modulus, , (b), , (c) By Problem 1.59(6) and DeMoivre's theorem,, , Then, , x—
Page 40 :
[CHAP. 1, , REVIEW OF FUNDAMENTAL CONCEPTS, , 30, (6), , Then, , The results for k = 0,1,2 are, , The results for k = 3,4, 5,6, 7, ... give repetitions, of these. These complex roots are represented geometrically in the complex plane by points P1( P&P» on the, circle of Fig. 1-9., , Fig. 1-9, , 1.61. Assuming that the series for ex on page 8 holds for complex values of x, arrive at, Euler's formula e™ = cos <£ + i sin <£., If, , we have, , Similarly we find e-i* = cos <f> — i sin $., , Supplementary Problems, REAL NUMBERS AND LAWS OF ALGEBRA, 1.62., , Prove that (a) v% (&) V3, (c) V2 + A/3 are irrational numbers., , 1.63., , Arrange in order of increasing value: 11/9, \fí, VÏÏ ^/5/^/Z fy(N., , 1.64., , Explain why (a) 1/0, (6) 0/0 cannot represent unique numbers., , 1.65., , Explain why we must define (a) (-5)(3) = -15,, are to hold., , (6) (-2)(-3) = 6 if the rules of algebra on page 1, , FUNCTIONS, 1.66., , If /(») = (8«« + 2<B-6)/(«-4), find (o)/(2), (6) /(-I), («) /(3/2), (d) f(-x),, , (e)/(\/2)., , 1.67., , An odd function is one for which /(—a;) = —/(«) while an ei>ew function is one for which, /(—a;) = /(a;). Classify each of the following according as they are even or odd: (a) cos 2a;,, (6) sin 3ic, (c) tan x, (d) ex, (e) ex — e~x, (/) ex + e~x., , 1.68., , (a) Prove that e olnl) = &«. (6) Find e-2ta«.
Page 41 :
CHAP. 1], , REVIEW OF FUNDAMENTAL CONCEPTS, , 1.69., , Prove that (a) sin 3* = 3 sin * - 4 sin3 x, (b) cos Sx - 4 cos3 x - 3 cos x., , 1.70., , Prove that (a) coth2 x - csch2 x = 1, (6) cosh2 x + sinh2 * = cosh 2x, (c), , 1.71., , If cos x = 8/17, find (a) sin 2», (6) cos 2a, (c) sin (a/2)., , 1.72., , Find tanh (In 3)., , LIMITS AND CONTINUITY, 1.73. Find (a), 1.74., , If, , 1.75., , If, , (b), , (c), , 31, , and prove your conclusions., , prove that, and, , prove that, , (a), , (e), , (b), , (d), , 1.76., , If f(x) = \x\/x, prove that lim f(x) does not exist., , 1.77., , Using the definition prove that f(x) = 3 — x2 is continuous at (a) x = 1, (b) x = a., , 1.78., , If f(x) is continuous at a, prove that [/(œ)]2 is also continuous at a., , 1.79., , If f1(x) and fz(x) are continuous at a prove that (a) f1(x) + f2(x), (b) fi(x) —f2(x), (c) f1(x) f2(x),, (d) fi(x)/f2(x), fz(a)¥=0 are also continuous at a. Revise the statement if a is replaced by an, interval such as a = x = b or a < x < b., , 1.80., , If, , 1.81., , Prove the statement at the end of Problem 1.12(6)., , z-»0, , prove that f(x) is continuous (a) at x = 0, (6) in any interval., , DERIVATIVES, 1.82., , Using the definition, find the derivatives of (a) f(x) = x2 - 2x + 5 and (6) f ( x ) = (x — l)/(x + 1)., , 1.83., , Prove the differentiation formulas (a) 2, (b) 4 on page 4., , 1.84., , Assuming that, (c) 14 on page 5., , 1.85., , Prove formulas (a) 15, (b) 16, (c) 18, (d) 19, (e) 20 on page 5., , 1.86., , Find, , prove the formulas, , (a) 12, (6) 13,, , (/), , 1.87., , If e*» + y2 - cos x, find dy/dx., , 1.88., , If x — a(e — sin e), y = a(l — cos e), find dy/dx., , 1.89., , If y = (Sx + !)/(! - 2x), find d*y/dx2 at * = 2., , 1.90., , If y = (cj sin x + c2 cos x)/^fx,, 2, , 2, , show that xy" + 2y' + xy = 0 where clt c2 are any constants., , 1.91., , Find d y/dx, , for the function defined in Problem 1.88., , 1.92., , Find the equation of the tangent line to the curve xy2 + y — Sx — l at the point (1,1)., , 1.93., , Find the differentials of (a) y = x2 In x + Sx, (b) y = (2x - l)/(x + 2)., , 1.94., , (a) Show that if x > 0 and A« is numerically small compared with x, then, is approximately, equal to, (6) Use this result to find the approximate values o f a n d, (c) By obtaining a similar result for V* + A*, find an approximate value for, and, •>
Page 42 :
1.95., , [CHAP. 1, , REVIEW OP FUNDAMENTAL CONCEPTS, , 32, , Prove the law of the mean, page 8, and illustrate by an example., , INTEGRALS, 1.96., , Find the integrals, , (d), , (a), , (b), , (/), , (e), , (e), , (*), , (g), , 1.97., , Prove the integration formulas (a) 7, (b) 11, (c) 12, (d) 13, (e) 14 on page 6., , 1.98., , Find, , ( a, , (b), , (d), , (c), , (e), 1.99., , Evaluate (a), , (b), , (c), , 1.100. Prove the integration formulas (a) 15, (6) 16, (c) 17, (d) 18, (e) 19, (/) 20, (g) 21, (h) 22 on, page 6., 1.101., , Find, , 1.102. Find, , (a), , (c), , (b), , (d), , (a), , (c), , (6), , (d), , (e), , (e), , 1.103. (a) Find the area bounded by y = 4x — x2 and the x axis., (b) Find the area bounded by the curves y — x2 and y = x., 1.104., , Using the trapezoidal rule, find the approximate value of, , by dividing the interval from, , 0 to 1 into (a) 5, (b) 10 equal parts, (c) Show that the integral has the exact value ir/4 and compare, with the approximate values., 1.105., , Prove the trapezoidal rule which gives the approximate value of, , where h = (b — a)/n., 1.106. Assume that a function f(x) is approximated by the, parabolic function c0 + Cj» + c2x2 in the interval, a S x S a + 2h [see Fig. 1-10]. Show that we have, approximately, , The result is often called Simpson's rule., 1.107. (a) Show how Simpson's rule [Problem 1.106] can be, used to find the approximate value of, and (6) use the result to find the approximate values, of the integrals in Problems 1.28 and 1.104. Compare, the accuracy of the results with those of the, trapezoidal rule., , Fig. I.JQ
Page 43 :
CHAP. 1], , REVIEW OF FUNDAMENTAL CONCEPTS, , 33, , SEQUENCES AND SERIES, 1.108. Find the limit of the sequence whose nth term is un = (n- l)/(2w +1), justifying your answer., 1.109. Show that the series, 1.110., , converges and find its sum., , Prove that, , and (6) if, , then, , 1.111. Show that .181818 ... = .18 + .0018 + .000018 + • • • = 2/11., (d), , (c), , (b), , 1.112. Investigate t h e convergence o f t h e serie, (f), , (e), , 1.113. Prove that the series, , converges for, , 1.114. Prove (a) Theorem 1-3, (6) Theorem 1-4, (c) Theorem 1-5, page 6., 1.115. (a) Consider the series w t — u2 + u3 — M4 + • • • where uk > 0., then the series converges. (6) Thus show that, , Prove that if, , un+l£un and, is convergent, , but not absolutely convergent., 1.116., , Find the interval of convergence of (a), , 1.117. Find the interval of convergence of (a), , (b), (6), , UNIFORM CONVERGENCE, 1.118. Prove that the series *(1 — x) + x2(l — x) + xs(l — x) + • • • converges for — 1 < x a 1 but is not, uniformly convergent in this interval. Is it uniformly convergent in —1/2 S x ë 1/2? Explain., 1.119., , Investigate the uniform convergence of the series (a), , in, , (6), , for, , 1.120. Prove that a power series c0 + c^x — a) + c2(x — a)2 + • • • converges uniformly in any interval, which lies entirely within its interval of convergence. [Hint: Use the Weierstrass M test.], 1.121., , (a) Prove that if Sn(x) converges uniformly to S(x) in o = x a 6, then, , (6) Show that, and supply a possible explanation., 1.122. Prove Theorem 1-11, page 7., 1.123. (a) By using analogy with series, give a definition for uniform convergence and absolute convergence of the integral, (b) Prove a test for uniform [and absolute] convergence of the integral in (a) analogous to the, Weierstrass M test for series., (c) Prove theorems for integrals analogous to Theorems 1-9, 1-10 and 1-11 on page 7.
Page 44 :
34, , REVIEW OF FUNDAMENTAL CONCEPTS, , [CHAP. 1, , TAYLOR SERIES, 1.124. Obtain the formal Taylor series (a) 1, (6) 3, (c) 4, (d) 5 given on page 8., 1.125. Expand sin * in a Taylor series about a = W6 and use the result to find sin 31° approximately., Compare with the result from a table of sines., 1.126. Find an approximate value for (a), 1.127. Show that (a) sinh x =, PARTIAL DERIVATIVES, 1.128. If f ( x , y ) = x sin (y/x), find (a) f(2,v/2),, (/) /Sz(2,T/2), (O) fm(2,v/2)., 1.129. Verify that /„, = fyx, , for, , , (b), , (b) fx(2,ir/2),, , (e) fy(2,w/2),, , (d) f^(2,v/2),, , (e) f^(2, v/2),, , for the case where /(*, y) = (x - y)/(x + y) if * + y ¥> 0., , 1.130. If f ( x , y ) = xz tan-1 (y/x),, x and 2/?, , show t, , Arethereanyexcweptionalvaluesfor, , 1.131. If V(x,y,z) = (*2 + 3/2 + «2)-1/2, show that, , if x,y,z are not all zero., , 1.132. If f ( x , y ) = ycos(x-2y), find df if, 1.133. If f(x,y,z) = x*z - yz^ + x*, find df., 1.134. I f z = x2f(y/x), Problem 1.130., , ndcomparewith, , where / i s differentiate, show t, , 1.135. If, , prove that if U is a twice differentiable function of * and y,, , TAYLOR SERIES FOR FUNCTIONS OF TWO OR MORE VARIABLES, 1.136. Expand f(x, y) = 2xy + »2 + ya in a Taylor series about a = 1, 6 = 2 and verify the expansion., 1.137. Write Taylor's series for f(x, y, z) expanded about x — a, y = b, z = c and illustrate by expanding, f(x, y, z) = xz + y2 about x = 1, y = —1, 2 = 2., LINEAR EQUATIONS AND DETERMINANTS, 1.138. Solve the systems, , (a), , (6), , 1.139. Determine whether the system of equations, , has non-trivial solutions. If so give two such solutions. If not replace the first equation by, kx — y + z — 0 and determine fc so that the system has non-trivial solutions and give two such, solutions., 1.140. Show that if two rows (or columns) of a second or third order determinant are interchanged, the, sign of the determinant is changed., 1.141. Show that if the elements in two rows (or columns) of a second or third order determinant are, equal or proportional, the value of the determinant is zero.
Page 45 :
CHAP. 1], , REVIEW OF FUNDAMENTAL CONCEPTS, , 35, , 1.142. Discuss the system of equations, from a geometrical viewpoint., MAXIMA AND MINIMA. THE METHOD OF LAGRANGE MULTIPLIERS., 1.143. Find the relative maxima and minima of f(x) = 12 + 8«2 — x* and use the results to graph the, function., 1.144. Find the relative maxima and minima of (a) x2 In x, (b) a sec 6 + b esc e., 1.145. Find the area of the largest rectangle which can be inscribed in a semicircle of radius a with base, on its diameter., 1.146. If /'(a) = 0 and /"(a) = 0, is it possible that f(x) has a maximum or minimum at, Explain. [Hint: Consider f(x) = x* or f(x) = —x*.], 1.147. A box having only five rectangular sides is required to have a given volume V., dimensions so that the surface area will be a minimum., 1.148., , x = a?, , Determine its, , Find the shortest and largest distance from the origin to the curve x2 + xy + y2 = 16 and give a, geometric interpretation. [Hint: Use the method of Lagrange multipliers to find the maximum of, «2 + y2.], , 1.149. (a) Find the relative maximum and minimum of x2 + y2 + z2, given that 4a;2 + 9y2 + 16z2 = 576, and (6) interpret geometrically., 1.150. Generalize the method of Lagrange multipliers to the case where /(*, y, z) is to be made an extremum, [i.e. maximum or minimum] subject to two constraint conditions <t>i(x, y, z) — 0, <f>2(x, 2/> z ) = 0., [Hint: Let h(x, y, z) = f ( x , y, z) + Xj^z, y, z) + X202(&> V>z) and prove that dh/dx — 0, dh/dy = 0,, dh/dz — 0 where \i, X2 are two Lagrange multipliers.], LEIBNITZ'S RULE, , ddl/daby(a)integratingfirstand(b)usingLeibnitz's, , 1.151. I f, rule., 1.152., , Find dl/da if (a), , 1.153., , Prove Leibnitz's rule., integrals. Then let, , (b), [Hint:, , C o n s i d e r a n d write the result as the sum of three, , COMPLEX NUMBERS. POLAR FORM, 1.154.Perform each of the indicated operations: (a) 2(5 - 3i) - 3(-2 + i) + 5(i - 3), (b) (3 - 2i)3, (c), (/), , 1.155., , If Zj and z2 are complex numbers, prove (a), , 1.156., , Prove, , 1.157., , Find all solutions of 2x* - 3x3 - 7a;2 - 8x + 6 - 0., , (a), , (b), , (6), , giving any restrictions., (c), , 1.158. Let zl and z2 be represented by points Pl and P2 in the Argand diagram. Construct lines OP1 and, OP2, where O is the origin. Show that zl + z2 can be represented by the point P3, where OP3 is the, diagonal of a parallelogram having sides OPj and OP2. This is called the parallelogram law of, addition of complex numbers. Because of this and other properties, complex numbers can be considered as vectors in two dimensions. The general subject of vectors is given in Chapter 5., 1.159. Interpret geometrically the inequalities of Problem 1.156.
Page 46 :
36, 1.160., , REVIEW OF FUNDAMENTAL CONCEPTS, Express in polar form (o)3A/3 + 3¿, (b) -2-2i, (c), , (d) 5, (e) -5i., , 1.161. Evaluate (a), , (6), , 1.162. Determine all the indicated roots and represent them graphically:, (a), , (b), , 1.163. If, and, Interpret geometrically., 1.164., , (c), , prove (a) :, , Show that De Moivre's theorem is equivalent to (e**)n = ein<*., , Answers to Supplementary Problems, 1.63., , 1.66., , (a) -11/2, , (6)4/5 (c) -19/10 (d) (5 + 2x - 3x*)/(x + 4) (e) -(8 + 9\/2 )/14, , 1.67., , (a) even (b) odd (c) odd (d) neither (e) odd (/) even, , 1.68., , (6), , 1.72., , 4/5, , 1.73., , (a) 18 (6) 1 (c) 1/32, , 1.86., , (a), (b), (c), , 1.71., , (d), (e), (f), , 1.87., , 1.88., , 1.91., , -l/o(l - cos <9)2 or -esc4 (*/2)/4a, , 1.89., , -20/27, , 1.93., , (a) (x + 2x In * + 3) dx, , 1.96., , 1.98.), , 1.99., , (a) 240/289 (b) -161/289 (c) VÎT/l?, , (a) 14/3, , (6), , (b) 5 dx/(x + 2)2, , 1.92., , 3y - 2x = 1, , [CHAP. 1
Page 47 :
CHAP. 1], , REVIEW OF FUNDAMENTAL CONCEPTS, , 37, , 1.101., , 1.102., , 1.103. (a) 32/3, , (6) 1/6, , 1.110. (a) diverges, , 1.107. 1, , (6) converges, , (c) converges, , 1.116., , (a) -1 < x S 1, , (b) -lëx SI, , 1.117., , (a) -2SxS2, , (b) -1 •& x ë 1, , 1.128. (a), , (6), , (c), , (d), , (e) converges, , (/), , (e), , (a) x = -1, y = 2, , rel. max. 28 at x = ±2, , 1.144. (a) rel. min. -l/2e at x = e~i/ 2, 1.145. a2, , (b) rel. min. (a2's + 62/3)3/2 f or 0 = tan"1 (6/a)1/3, , 1<147, , 1.148. max.4\/2, min. 4\/6/3, 1.151. (a), (a) l-4i, , (6), (b) -9 - 46i, , (c), , (d) -1, , (e), , (/), , 1.157. ' 3, $, -l±i, 1.160. (a) 6cisW6, 1.161. (a), 1.162., , (g), , (6) x = 2, y - 0, z = -1, , 1.141. rel. min. 12 at x = 0,, , 1.154., , (/) diverges, , 1.133. (2xz + 4a;3) dx - t? dy + (x2 - Syz2) de, , 1.132., 1.138., , (d) diverges, , (a), (b), (c), , (d), , (6) 2^/2 cis 5W4, (b) -2i, , (c) 2cis5W3, , (d) 5cisO, , (e) 5cis3 ff /2
Page 48 :
Chapter 2, , Ordinary, Differential Equations, DEFINITION OF A DIFFERENTIAL EQUATION, A differential equation is an equation involving derivatives or differentials., lowing are some examples of differential equations., Example, , The fol-, , re, , Example 2., Example 3., Example 4., Example 5., , Equations such as those in Examples 1-4 involving only one independent variable are called, ordinary differential equations. Equations such as that of Example 5 with two or more, independent variables are called partial differential equations and are treated in Chapter 12., ORDER OF A DIFFERENTIAL EQUATION, An equation having a derivative of wth order but no higher is called an nth order differential equation. In Examples 1-5 above, the orders of the differential equations are, 2,1, 2,1, 2, respectively., ARBITRARY CONSTANTS, An arbitrary constant, often denoted by a letter at the beginning of the alphabet such as, A, B, C, clf cz, etc., may assume values independently of the variables involved. For example, in y — xz + cix + c2, c\ and c2 are arbitrary constants., The relation y = Ae~4x+B which can be written y = AeBe~*x = Ce~4x actually involves, only one arbitrary constant. We shall always assume that the minimum number of constants is present, i.e. the arbitrary constants are essential., SOLUTION OF A DIFFERENTIAL EQUATION, A solution of a differential equation is a relation between the variables which is free, of derivatives and which satisfies the differential equation identically., Example 6. y - x2 + c& + c2 is a solution of y" = 2 since by substitution we obtain the, identity 2 = 2., 38
Page 49 :
ORDINARY DIFFERENTIAL EQUATIONS, , CHAP. 2], , 39, , A general solution of an rath order differential equation is one involving n (essential), arbitrary constants., Example 7. Since y = x2 + «i* + c2 has two arbitrary constants and satisfies the second order, differential equation y" - 2, it is a general solution of y" = 2., , A particular solution is a solution obtained from the general solution by assigning, specific values to the arbitrary constants., Example 8. y = xz — 3* + 2 is a particular solution of y" = 2 and is obtained from the general, solution y = x2 + ctx + c2 by putting Cj = —3 and c2 = 2., , A singular solution is a solution which cannot be obtained from the general solution, by specifying values of the arbitrary constants., Example 9. The general solution of y = xy'- y'2 is y = ex - c2. However, as seen by substitution another solution is y = «2/4 which cannot be obtained from the general, solution for any constant c. This second solution is a singular solution. For a, relationship between the general and singular solutions see Problem 2.5., , DIFFERENTIAL EQUATION OF A FAMILY OF CURVES, A general solution of an nth order differential equation has n arbitrary constants (or, parameters) and represents geometrically an n parameter family of curves. Conversely, a relation with n arbitrary constants [sometimes called a primitive] has associated with it a, differential equation of order n [of which it is a general solution] called the differential, equation of the family. This differential equation is obtained by differentiating the primitive n times and then eliminating the n arbitrary constants among the n +1 resulting, equations. See Problems 2.6 and 2.7., SPECIAL FIRST ORDER EQUATIONS AND SOLUTIONS, Any first order differential equation can be put into the form, (1), , and the general solution of such an equation'contains one arbitrary constant. Many special, devices are available for finding general solutions of various types of first order differential, equations. In the following table some of these types are given., General Solution, method by which it can be obtained), , Differential Equation, M, 1. Separation of variables, , Divide by g i (y) f2 (x) ¥* 0 and integrate to obtain, , fi(x) ffi(y) dx + f2(x) ff2(y) dy, , = 0, , 2. Exact equation, M(x,y)dx + N(x,y)dy, where, , =, , 0, , The equation can be written as, M dx + Ndy = dU(x,y), = 0, v, where dU is an exact differential. Thus the solution is, U(x, y) = c or equivalentíy, , where dx indicates that the integration is to be performed, with respect to * keeping y constant.
Page 50 :
ORDINARY DIFFERENTIAL EQUATIONS, , 40, , General Solution, be obtained), , Differential Equation, 3. Integrating factor, , [CHAP. 2, , (or method by which it can, , The equation can be written as an exact differential, equation, where » is an appropriate integrating factor so that, , \yVigT*P, , and then method 2 applies., The following combinations are often useful in finding, integrating factors., («), , (&), , («), (d), , (e), See also Problem 2.17., 4. Linear equation, ,, , ff + P(*)V, , An, , = Q(x), , integrating factor is given by, /Pd)dx, , " =e, and the equation can then be written, , with solution, ye, , 5. Homogeneous equation, , ¡iy, (Pdx, , Let ylx — v or y — vx, and the equation becomes, , v + x ~ = F(v), , or, , xdv + (F(v) - v) dx - 0, , which is of Type 1 and has the solution, , where v = ylx. If F(v) = v, the solution is y - ex., 6. Bernoulli's equation, , Letting v = yl~n, the equation reduces to Type 4 with, solution, , If n = 0, the equation is of Type 4. If n = 1, it is of, Type 1.
Page 51 :
ORDINARY DIFFERENTIAL EQUATIONS, , CHAP. 2], , General Solution, (or method by which it can be obtained), , Differential Equation, 7. Equation solvable for y, V = g(«, P), , where p = y', , 41, , Differentiate both sides of the equation with respect, to a; to obtain, , or, , Then solve this last equation to obtain G(x, p, e) = 0., The required solution is obtained by eliminating p between G(x, p, c) = 0 and y = g(x, p)., An analogous method exists if the equation is solvable, for x., 8. Clairaut's equation, y = px + F(p), , where p = y', , 9. Miscellaneous equations, , The equation is of Type 7 and has solution, y = ex + F(c), The equation will also have a singular solution in general., (a) Letting, , the equation reduces to Type 1., , (6) Let x = X + h, y = Y + k and choose constants h, and k so that the equation reduces to Type 5. This is, possible if and only if a¡/a2 =£ /8,//32. If a^a^ =, Pi/Pz, the equation reduces to Type 9(a)., , EQUATIONS OF HIGHER ORDER, If a differential equation is of order m > 1 and has one of the variables x or y missing, explicitly from the equation, then it can be reduced to a differential equation of order, m — 1 by letting, or, (2), y' - P,, See Problems 2.25 and 2.26., EXISTENCE AND UNIQUENESS OF SOLUTIONS, It is often important to be able to predict directly from a differential equation and, associated conditions whether a solution exists and is unique. For the case of a first, order differential equation, y' = f ( x , y ), (3), an answer is supplied in the following, Existence and Uniqueness Theorem. If f(x, y) is continuous and has a continuous partial, derivative with respect to y at each point of the region R defined by \x — x0\ < 8,, \y — Vo\ < 8, then there exists in R one and only one solution to (3) which passes through, the point (x0,yn)., An immediate generalization of this theorem is possible for the nth order differential equation y<n> = f(x,y,y',...,y<n-»)., APPLICATIONS OF DIFFERENTIAL EQUATIONS, Many problems of science and engineering, when formulated mathematically, lead to, boundary-value problems, i.e. differential equations and associated conditions. Solutions, to these can be of great value to the scientist and engineer.
Page 52 :
42, , ORDINARY DIFFERENTIAL EQUATIONS, , [CHAP. 2, , In a mathematical formulation of a physical problem, a mathematical model is chosen, which often approximates the actual situation. For example, in treating rotation of the, earth around the sun, we may consider that the sun and earth are points., If a mathematical model and corresponding mathematical formulation leads to fairly, good agreement with that predicted by observation or experiment, then the model is good., Otherwise a new model may have to be chosen., SOME SPECIAL APPLICATIONS, , The following table lists some special applications of an elementary nature which arise, in science and engineering., I. MECHANICS, , The basic law of mechanics or dynamics is that of Newton, i.e., (4), , where m is the mass of the moving object, v is its velocity, t is the time and F is the net force, acting on the object. The quantity mv is often called the momentum., If m is constant, then this equation becomes, where a is the acceleration., On or near the earth's surface, mass m is connected with weight W by m = W/g, where g is the acceleration due to gravity., , (5), , or W - mg, , Various systems of units are available., (a) C.G.S. [centimeter (cm), gram (g), second (sec)]., If m is in grams and a in cm/sec2, then, F is in dynes., (b) M.K.S. [meter (m), kilogram (kg), second (sec)]. If m is in kilograms and a in m/sec2, then, F is in newtons (nt)., (c) F.P.S. [foot (ft), pound (Ib), second (sec)]. If W is in pounds a and g in ft/sec2, then F =, ma — Wa/g is in pounds., On the earth's surface g = 32 ft/sec2 = 980 cm/sec2 — 9.8 m/sec2 approximately., II. ELECTRIC CIRCUITS, , A simple series electric circuit [Fig. 2-1] may consist of, , (1) a battery or generator supplying an electromotive force or, e.m.f. (voltage or potential) of E volts,, (2) a resistor having a resistance of R ohms,, (3) an inductor having an inductance of L henries,, (4) a capacitor or condenser having a capacitance of C farads., The current I measured in amperes is the instantaneous time, rate of change of charge Q on the condenser measured in coulombs, i.e. 7 = dQ/dt., From basic principles of electricity we have, Potential drop across resistor = IR, Potential drop across inductor, , Fig. 2-1, , = L dl/dt, , Potential drop across capacitor = QIC, Kirchhoff's Laws., (a) The algebraic sum of the currents flowing into a junction point is zero., (6) The algebraic sum of the potential drops around a closed loop is zero., For single loop circuits law (a) implies that the current is the same throughout the loop.
Page 53 :
CHAP. 2], , III., , ORDINARY DIFFERENTIAL EQUATIONS, , 43, , ORTHOGONAL TRAJECTORIES, , A curve cutting each member of a one parameter family of curves at right angles is called an, orthogonal trajectory of the family. If dy/dx = F(x, y) is the differential equation of a one parameter family (not containing the parameter), then the differential equation for the family of, orthogonal trajectories is, , IV. DEFLECTION OF BEAMS, , A horizontal beam situated on the x axis of an, xy coordinate system and supported in various ways,, bends under the influence of vertical loads. The deflection curve of the beam, often called the elastic curve, and shown dashed in Fig. 2-2, is given by y = f(x), where y is measured as positive downward. This curve, may be determined from the equation, («), , where M(x) is the bending moment at x and is equal to, the algebraic sum of the moments of all forces to one, side of x, the moments being taken as positive for, forces in the positive y direction and negative otherwise., , Fig. 2-2, , For small deflections y' is small and the approximate equation, , Ely" = M(x), , (7), , is used. The quantity El where E is Young's modulus and / is the moment of inertia of a cross, section of the beam about its central axis, is called the flexural rigidity and is generally constant., V. MISCELLANEOUS PROBLEMS, , Various problems of science and engineering pertaining to temperature, heat flow, chemistry,, radioactivity, etc., can be mathematically formulated in terms of differential equations., , NUMERICAL METHODS FOR SOLVING DIFFERENTIAL, , EQUATIONS, , Given the boundary-value problem, (8), , it may not be possible to obtain an exact solution. In such case various methods are, available for obtaining an approximate or numerical solution. In the following we list, several such methods., 1. Step by step or Euler method. In this method we replace the differential equation of, (8) by the approximation, (9), so that, , y(x0 + h) = y(x<>) + hf(xo,y0), , (10), , By continuing in this manner we can then find y(x<> + 2h), y(x<> + 3h), etc. We choose, h sufficiently small so as to obtain good approximations., A modified procedure of this method can also be used. See Problem 2.45.
Page 54 :
44, , ORDINARY DIFFERENTIAL EQUATIONS, , [CHAP. 2, , 2. Taylor series method. By successive differentiation of the differential equation in (8), we can find y'(x0), y"(x0), y'"(x0),, Then the solution is given by the Taylor series, (11), , assuming that the series converges. If it does we can obtain y(x0 + h) to any desired, accuracy. See Problem 2.46., 3. Picard's method. By integrating the differential equation in (8) and using the boundary condition, we find, (12), Assuming the approximation yi(x) = y<>, we obtain from (12) a new approximation, (13), , Using this in (12) we obtain another approximation, (14), , Continuing in this manner we obtain a sequence of approximations y\, y*, 2/3, .... The, limit of this sequence, if it exists, is the required solution. However, by carrying out, the procedure a few times, good approximations can often be obtained. See Problem 2.47., 4. Runge-Kutta method., , This method consists of computing, , (15), , Then, , (16), , See Problem 2.48., These methods can also be adapted for higher order differential equations by writing, them as several first order equations. See Problems 2.49-2.51., , Solved Problems, CLASSIFICATION OF DIFFERENTIAL EQUATIONS, 2.1., , Classify each of the following differential equations by stating the order, the dependent and independent variables, and whether the equation is ordinary or partial., (a) x*y" + xy' + (x2 - n2)y = 0, , order 2, dep. var. y, indep. var. x, ordinary, , (6), , order 1, dep. var. x, indep. var. y, ordinary
Page 55 :
CHAP. 2], , ORDINARY DIFFERENTIAL EQUATIONS, , (c), , 45, , order 1, dep. var. y, indep. var. x, ordinary, Note that this equation is identical with the equation in (&)., , (d), , order 2, dep. var. u, indep. var. t, ordinary, , (e), , order 2, dep. var. Y, indep. var. x and t, partial, , (/) (a2 + 2j/2) dx + (Sx2 — 4j/2) dy = 0, , order 1, dep. var. y (or *) indep. var. x (or y), ordinary, , SOLUTIONS OF DIFFERENTIAL EQUATIONS, 22. Check whether each differential equation has the indicated solution. Which solutions, are general solutions?, (a) y' - x + y = 0; y = Ce~x + x - 1., Substitute y = Ce~x + x — 1, y' = -Ce"1 + 1 in the differential equation., , Then, , x, , y' - x + y = -Ce-* + 1 - x + Ce~ + x - 1 = 0, x, , Hence y — Ce~ + x — 1 is a solution., Since the number of arbitrary constants (one) equals the order of the differential equation, (one), it is a general solution., (&), , Differentiating xzy -y3 = c we have x2y' + 2xy - 3y2yi = 0, i.e. (x2 - 3y2)y' + 2xy = 0, or, , Then the solution is a general solution., , <«), , Substitute, sin í in the differential equation., , Then, , and so the given relation is a general solution since the order of the differential equation and, the number of arbitrary constants in the solution are both equal to 2., (d), in the differential equation, we find «3(2c)2 =, , Substituting, , 2(cx2)(2cx) or 4c2*3 = 4e2ic3. Thus v = ex2 is a solution. However, it is not a general solution since the number of arbitrary constants (one) is not equal to the order of the equation, (two)., , 2.3., , Determine the particular solution of the differential equation of Problem 2.2(c), satisfying the conditions 7(0) = 2, 7'(0) = -5., From Problem 2.2(c) the general solution is, 7 = I(t), At í = 0,, , -, , 1(0) = el + c2 = 2, , ctf* + c2e-*t + sin t, , i.e., , (Í), , Cj + c2 = 2
Page 56 :
[CHAP. 2, , ORDINARY DIFFERENTIAL EQUATIONS, , 46, , Differentiating with respect to t yields, I'(t) = c^* — 3c2e~at + cost, so that at * = 0, /'(O) = ct - 3c2 + 1 = -5, i.e., (2) ct - 3c2 = -6, Solving (1) and (2) simultaneously,, solution is, , 2.4., , we find Cj = 0,, , c2 = 2, , 7 - 2e-3* + sin t, , Show that the solution of the boundary-value problem, Q"(t) + 4Q'(t) + 20Q(t) = 16e-«, 0(0) = 2,, Q'(0) = 0, 2t, is Q(«) = e~ (l + sin4£ + cos4£)., We have, , and the required particular, , t i£ 0, , Q(t) - e~^(l + sin 4< + cos 4t), Q'(t) = e-2«(4 cos4« - 4 sin4«) - 2e~2t(l + sin 4t + cos4t), = e- 2t (2 cos 4* - 6 sin 4t - 2), Q"(t) = e-2t(-g sin4t - 24 cos 4t) - 2e~2«(2 cos4t - 6 sin 4* - 2), = e-2'(4sin4t-28cos4« + 4), , Q"(t) + 4Q'(t) + 20Q(t) = e-2«(4 sin 4t - 28 cos 4t + 4), + 4e- 2 «(2 cos 4t — 6 sin 4t - 2), + 20e-2'(l + sin 4t + cos 4t), = 16e-2t, Furthermore Q(0) = 2 and Q'(0) = 0. Thus the given relation is a solution to the boundary-value, problem., , Then, , 2.5., , Determine graphically a relationship, between the general solution y —, ex — c2 and the singular solution, y = x2/4 of the differential equation, y = xy' - y'2., Referring to Fig. 2-3, it is seen that, y = ex — c2 represents a family of straight, lines tangent to the parabola y — x2/4. The, parabola is the envelope of the family of, straight lines., The envelope of a family of curves, G(x, y, c) — 0, if it exists, can be found by solving simultaneously the equations dG/dc = 0, and G = 0. In this example G(x, y, c) =, y — cx + c2 and 8G/do = — * + 2c. Solving, simultaneously — * + 2c = 0 and y — ex +, c2 = 0, we find x - 2c, y = c2 or y = x2/4., , Fig. 2-3, , DIFFERENTIAL EQUATION OF A FAMILY OF CURVES, 2.6. (a) Graph various members of the one parameter family of curves y — ex3, (b) Obtain, the differential equation for the family in (a)., (a) The graphs of various members of the family are indicated in Fig. 2-4 below., (6) From y = cxs we have dy/dx = 3c*2. Then since c = y/x3, the required differential equation of the family is
Page 57 :
ORDINARY DIFFERENTIAL EQUATIONS, , CHAP. 2], , 47, , Fig. 2-4, , 2.7., , Find the differential equation of the two parameter family of conies ax2 + by2 = 1., Differentiating with respect to x we have, 2ax + 2byy' = 0, 2, , or, , a = —byy'/x, , or, , -bxyy' + by2 = 1, , 2, , Substituting into ax + by = 1 gives, (-byy'/x)xz + by2 = 1, where a has been eliminated., , Further differentiation gives, -b[xyy" + xy'2 + yy'] + 2byy' = 0, , Then the required differential equation obtained on division by 6 ^ 0 is, xyy" + xy'2 — yy' = 0, , 2.8., , (a) Find a general solution of the differential equation dy/dx = Sx2., (b) Graph the solutions obtained in (a)., (c) Determine the equation of the particular curve in (b) which passes through the, point (1,8)., (a) We have, dy = Bx2dx, , or, , 3, , so that y = x + c is the general solution., (6) The curves y — xs + c for various values of c are, shown in Fig. 2-5. The set of curves for all values, of c is a one parameter family of curves whose differential equation is dy/dx = 3a;2. Through each, point of the xy plane there passes one and only one, member of the family., (c) Since the curve passes through (1, 3), we have y = 3, when x = 1. Thus from y — xs + c, 3 = I3 + c or, e = 2 and the required curve has the equation, y = *3 + 2., , 2.9., , (a) Solve the boundary-value problem, y" = Bx -2, y(0) = 2, y(l), and (6) give a geometric interpretation., Oy.2, , Fig. 2-5, , = -3, , (a) Integrating once, we have y' = —, 2x + Cj. Since j/'(l) = -3, we have — 3 = f — 2 + ct and c, = —|. Thus, , [i.e. y' = -3 when * = 1],
Page 58 :
48, , ORDINARY DIFFERENTIAL EQUATIONS, , Integrating again,, , [CHAP. 2, , Since j/(0) = 2 [i.e., y = 2 when x = 0],, , we find c2 = 2. Then the required solution, , (6) Geometrically, represents that particular member of the family [whose, differential equation is y" = 3x — 2] which passes through the point (0,2) and which has a, slope of —3 at the point where x = 1., Note that if we integrate the differential equation twice, we have, so that the solution represents a two parameter family of curves. The particular member of, the family found above corresponds to ct = — §, c2 = 2., , SEPARATION OF VARIABLES, 2.10. (a) Find the general solution of (4a; + xyz) dx + (y + x2y) dy = 0., (6) Find that particular solution for which y(l) = 2., (a) The equation can be written as x(4 + y2) dx + y(l + x2) dy = 0 or, , Integrating,, , i.e., , In, , Thus the required general solution is (1 + x2)(4 + y2) — c., (b) For the particular solution where y(1) = 2, i.e. y = 2 when x — 1, put * = 1, y = 2 in, (1 + *2)(4 + j/2) = c to obtain c = 16. Thus (1 + o;2)(4 + y2) = 16., , 2.11. Solve the boundary-value problem, We have dy/dx = 8 — 3y so that on separating the variables,, , Thus, Putting x = 0 and y = 2, we have —^ In 2 = c1 and so the required solution is, , This can also be written as £ In (8 - 3y) - J In 2 = —x, In (8 - 3y) - In 2 = -Sx,, or finally, Check:, , so that, , Also, , In
Page 59 :
CHAP. 2], , ORDINARY DIFFERENTIAL EQUATIONS, , 49, , 2.12. Solve the preceding problem if y(Q) = 2 is replaced by y(Q) = 4., If we use the result —^ In (8 — Sy) = x + e1 of Problem 2.11, we find on formally putting, = o, y = 4 that the logarithm of a negative number is involved. The difficulty can be avoided, on noting that actually, x, , so that the solution is -£ In |8 - Sy\ = x - % In 4. Since 8 - Sy is negative, the result can then, be written as, , 2.13. Solve dy/dx = sec y tan x., Separating the variables, we have dy/seey — tanxdx, , or, Then integration yields, , or, , sin y = —In cos o; + e, , i.e., , In cos x + sin y = c, , EXACT DIFFERENTIAL EQUATIONS, , 2.14. (a) Show that (3a;2 + y cos a;) dx + (sin x — iy3) dy — 0 is an exact differential equation and (6) find its general solution., (a) Here M = 3a;2 + y cos x, N = sin x — 4y3 so that dM/dy = cos x = dN/dx, tion is exact., , and the equa-, , (6) Method 1 [Grouping of terms by inspection]. Write the equation as, , Sx2 dx + (y cos xdx + sin x dy) — 4y3 dy = 0, , i.e., , d(x*) + d(y sin x) - d(y*), , = 0, , 3, , or, , d(x + y sin x — y*) = 0, , Then on integrating,, , x3 + y sin x — y* = c, , Method 2 [Direct method]. Since M dx + N dy = dU an exact differential and dU = ^-dx +, dU, , — dy, we must have, , and, , or, , (j), , Integrating the first equation in (¿) partially with respect to x [keeping y constant], we have, , W, where F(y) is the constant of integration which may depend on y., Substituting (2) in the second equation of (J), we find, y cos x + F'(y) — sin x — 4ys, so that F(y) = -y* + Cj. Then from (2),, , or, , F'(y) = —4y3, , U = x3 + y sin x — y* + ct, Thus, Mdx + Ndy = dU = dix3 + y sin x - y* + cj, from which we must have, x3 + y sin x — y* = c, , - 0
Page 60 :
50, , ORDINARY DIFFERENTIAL EQUATIONS, , [CHAP. 2, , Method 3 [Formula]. From the result 2 on page 39 we see that the solution is, , i.e., or, , INTEGRATING FACTORS, 2.15. (a) Show that the equation (3xy2 + 2y) dx + (2x2y + x) dy = 0 is not exact but that, (6) it becomes exact after multiplication by x. (c) Thus solve the equation., ( a ) Here, tion is not exact., , o that the equa-, , (6) Multiplying by x, the equation becomes, (3xV + 2xy) dx + (2x3y + x2)dy, , = 0, , d the, from which M = Sxty2 + 2xy, N = 2x3y + x2. T, tion is exact., (c) By any of the methods of Problem 2.14 we find the required solution x2y + *3j/2 = c., , 2.16. Solve (a) (y + x*) dx-xdy = 0,, , equa-, , (6) (a;3 + xy" -y)dx + xdy = 0., , (a) Write as y dx — x dy + x* dx = 0. The combination y dx — x dy suggests integrating factors, I/*2, 1/y2, l/(x2 + y2) [see page 40, entry 3]. Only the first leads to favorable results, i.e., , so that, (6) Write as x(x2 + y2) dx + x dy — y dx = 0 and multiply by the integrating factor l/(x2 + y2), to obtain, , Then, , 2.17. (a) If M dx + N dy = 0 has an integrating factor /A which depends only on x, show, that n = eSK*"* where f(x) = (Mv - N,)/N, My = dM/dy, Nx = dN/dx. (b) Write, the condition that /*, depend only on y. (c) Use the result to solve Problem 2.15., (a) By hypothesis, fiM dx + pN dy = 0 is exact., Since ¡i depends only on x this can be written, , Thus, and so, , Then
Page 61 :
CHAP. 2], , ORDINARY DIFFERENTIAL EQUATIONS, , 51, , (6) By interchanging M and N, x and y we see by part (a) that there will be an integrating factor, H depending only on y if (Nx — MV)/M = g(y) and that in this case ¡i = e} 9 v, (e) In Problem 2.15, M - 3xy2 + 2y, N - 2x2y + x, Mv = 6xy + 2, Nx = 4xy + 1 and, depends only on *., , Thus by part (a),, , an integrating factor and we can proceed as in Problem 2.15(6)., , LINEAR EQUATIONS, 2.18., , Solve the equation, Write the equation as, Then, Since, = P(x) depends only on x, we see from Problem 2.17 that e -> p(x)<te, an integrating factor. Multiplying by this factor, the equation becomes, , which can be written as, Then on integrating we have, , or, 2.19., , Solve, Write the equation as, , , a linear equation of the form, , with P = —Z/x, Q — x2 cos 4x. As in Problem 2.18, an integrating factor is, Multiplying by x~2 we have, which can be written as, Then by integrating we find, , HOMOGENEOUS EQUATIONS, 2.20. Solve (2a;3 + y3) dx - Sxy2 dy = 0, Write as, The right side is seen to be a function of y/x either by writing it as, , is
Page 62 :
52, , ORDINARY DIFFERENTIAL EQUATIONS, , [CHAP. 2, , or by letting y = vx and showing that the right side depends only on v, i.e., , Thus the equation is homogeneous. From y = vx we obtain, , or, Separating the variables, we have, In (vs — 1) =, , Integrating,, or, , — 2 In x + Ci, , In [(v« -1)*2] = G!, 3, , 2, , c, , Thus (i> — 1)« = e i = c and the required solution on letting v = y/x is y3 — x3 = ex., Note that in the integration we assumed vs — 1 > 0. Similar reasoning analogous to that of, Problem 2.12 shows that the solution is also valid if v3 — 1 < 0., , 2.21., , Solve, The right side is a function of y/x. Thus letting y = vx,, , or, , Integrating,, Since, , the required solution is, , BERNOULLI'S EQUATION, 2.22. Solve, The equation written a, Q(x) = 1, n = 3., , s, , i, , s a Bernoulli equation [see page 4 0 ] with P(x) = l/x,, , Making the transformation yl~« = v, i.e. y~2 = v, we find on differentiating with respect, to x,, Thus the equation becomes, , i.e., Since, , the general solution of this is v = 2x + ex2, the solution of the required equation is
Page 63 :
CHAP. 2], , ORDINARY DIFFERENTIAL EQUATIONS, , 53, , EQUATIONS SOLVABLE FOR ONE VARIABLE, 2.23. Solve xp2 + 2px - y - 0 where, , p = y'., , The equation can be solved explicitly for y, i.e. y = xp2 + 2px., to x yields, , Differentiation with respect, , or, , M, , Case 1, p + 1 ¥• 0., , In this case (2) becomes on division by (p + 1),, , whose solution is xp2 — c., , Then, (2), , are the parametric equations of the general solution. By eliminating p from (2) we can obtain the, general solution in the form, .., 2, (y — e) -, , 4cx, , (3), , 2, , Case 2, p + 1 = 0. In this case p = —1, and substituting into xp + 2px — y = 0 we find x + y = 0, which is a solution of the differential equation, as can be checked. However, it cannot be obtained, from the general solution (3) by any choice of c. Thus x + y = 0 is a singular solution., , CLAIRAUT'S EQUATION, 2.24. Solve, , where p — y'., , Use the method of Problem 2.23., , Differentiating,, , from which, Case 1, , Case 2,, , ,, , I, , n, , this case p = c and so the general solution is, , In thiscase, , To eliminate p, note that, , and, , . The equation x2 + y2 = 1 satisfies the, , given differential equation but cannot be obtained from the general solution for any choice of c, and so is a singular solution., , EQUATIONS OF ORDER HIGHER THAN ONE, 2.25. Solve y" + 2y' = 4x., Since y is missing from the equation, let y' = p, y" = dp/dx., , Then the equation becomes, W, , a linear first order equation having integrating factor e^, written as, , zdx, , 2, , 2x, , = e *. Multiplying by e , (1) can be
Page 64 :
ORDINARY DIFFERENTIAL EQUATIONS, , 54, , [CHAP. 2, , so that, i.e., Integrating again, we find the required general solution, y = x* - x - fae-** + c2 = x2 - x + Ae~^ + B, , 2.26. Solve, , 1 + yy" + y'2 = 0., , en the given, , Since * i s missing from t h e equation,, equation becomes, Separating variables and integrating,, , i In (1 + p2) + In y =, , i.e., , z, , Cl, , or, , In [(1 + p%2] = c2, , 2, , from which (1 + jP)y = a and, Separating variables,, Integrating,, Squaring gives, which is the general solution., , MECHANICS PROBLEMS, 2.27. An object is thrown vertically upward from the ground with, initial velocity 1960 cm/sec. Neglecting air resistance,, (a) the maximum height reached and (b) the total time taken, to return to the starting point., , Let the object of mass m be located at distance x cm from the, ground after time í sec [see Fig. 2-6]. Choose the upward direction as, positive. By Newton's law,, Netforce = Weight, (1), , The initial conditions are x = 0, dx/dt = 1960 at t - 0., Solving (1) subject to the initial conditions, we find, x = 1960i-490t2, (o) The height is a maximum when dx/dt = 1960 — 980Í = 0 or t = 2., 490(2)2 = 1960. Thus the maximum height reached is 1960 cm., , Fig. 2-6, , (2), Then * = 1960(2) -, , (6) x = 0 when t(1960 — 490t) = 0 i.e. t = 0, 4. Then the required time to return is 4 sec., , 2.28., , A 192 lb object falls from rest at time t = 0 in a medium offering a resistance in Ib, numerically equal to twice its instantaneous velocity in ft/sec. Find (a) the velocity, and distance traveled at any time t > 0 and (b) the limiting velocity., Choose the positive direction downward., Netforce = Weight — Resistance, U)
Page 65 :
CHAP. 2], , ORDINARY DIFFERENTIAL EQUATIONS, , T, , 55, , (a) Solving (1) subject to the initial condition v = 0 at t = 0, we find for the velocity at any, tlme, , ', , v = 96(l-e~t/3), , (2), , Replace v by dxldt in (2) and use the initial condition x = 0 at t = 0 to obtain for the, distance traveled,, ^ = ^ + Bg_t/3 _ g), (6) The limiting velocity is, , lim 96(1 - e-'/s) = 96(ft/sec), , t-+00, , and can also be obtained by setting dv/dt = 32 — v/3 = 0., , 2.29. Solve Problem 2.28 if the medium offers a resistance in Ib numerically equal to 3v2., Net force, , =, , Weight — Resistance, , w, Then separating variables and integrating, we find, , (a) Since v = 0 at t = 0, c = 0. Then, (2), which is the velocity at any time., Integrating the second equation in (2) subject to x = 0 at t = 0, we find for the distance, traveled,, , (6) The limiting velocity is, , and can also be obtained by setting, , 2.30. A boat of mass m is traveling with velocity v0. At t — 0 the power is shut off., Assuming water resistance proportional to vn where n is a constant and v is the, instantaneous velocity, find v as a function of distance traveled., Let x = distance traveled after time t > 0., Net force, , =, , Forward thrust — Water resistance, , where k is the constant of proportionality. Then, , Case 1, n ¥= 2., , Integrating, using v = v0 at * = 0 where v0 is the velocity at t = 0,, v2-n, , Case 2, n = 2., , =, , v2-n, , _ lL (2 _ n)x, , Integrating, using v = v0 at * = 0,, V =, , V(>e-kx/m
Page 66 :
[CHAP. 2, , ORDINARY DIFFERENTIAL EQUATIONS, , 56, , 2.31. A uniform chain of length a is placed on a, horizontal frictionless table so that a length b, of the chain dangles over the side. How long, will it take for the chain to slide off the table?, Suppose that at time t a length x of the chain is, dangling over the side [Fig. 2-7]. Assume that the density (mass per unit length) of the chain is a. Then, Fig. 2-7, Net force = Mass being accelerated • Acceleration, (1), Then since, , (1) becomes, (2), , Integrating (2), using x = b when v = 0,, (S), Separating the variables in (S) and integrating again using x = b when í = 0,, , Since the chain slides off when x = a, the time T taken is, , ELECTRIC CIRCUIT PROBLEMS, 2.32. A resistor of R = 10 ohms, an inductor of, L = 2 henries and a battery of E volts are connected in series with a switch S [Fig. 2-8]. At, t = 0 the switch is closed and the current 7 = 0., Find 7 for t > 0 if (a) E = 40, (&) E = 2Qe~3t,, (c) E = 50 sin 5i., By Kirchhoff's laws,, Potential drop across R + Potential drop across L + Potential drop across E, , =, , 0, , or, , (a) If E = 40,, , Solving this subject to 7 = 0 at t = 0, we have / = 4(1 — e~5t)., , (b) If E = 20e-3t, — + 51 = loe-at or multiplying by the integrating factor e«, j- (7e«) = 10e2t, from which we obtain 7 = 5(e~3t — «-««) on using / = 0 at í = 0., (c) If E = 50 sin 5t,, , sin 5t, j¿ (/«s») = 25e« sin 5i., , Integrating,, , Since / = 0 at t = 0, c = f and so / = |(sin 5t - cosSi) + fe-s*., The term £«-«, which approaches zero as t increases, is called the transient current., The remaining terms £(sin 5t — cos 5i) comprise the steady-state current.
Page 67 :
57, , ORDINARY DIFFERENTIAL EQUATIONS, , CHAP. 2], , 2.33. A resistor of R = 5 ohms and a condenser of C = .02 farads are connected in series, with a battery of E = 100 volts [Fig. 2-9]. If at t = 0 the charge Q on the condenser, is 5 coulombs, find Q and the current 7 for t > 0., dQ, Potential drop across R — 51 — 55Jt, -^, , Potential drop across C = -rjr, , =, , *>OQ, , Potential drop across E — —E, Then by Kirchhoff's laws,, If E = 100 volts,, -jT (e10*Q) = 20elot. Then integrating and solving subject, to Q = 5 at t = 0, we find Q = 2 + 3e-">« and / = dQ/dt = -30e-i°*., , Fig. 2-9, , Fig. 2-10, , 2.34. An inductor of L henries and a condenser of C farads are connected in series, [Fig. 2-10]. If Q = Qo and 7 = 0 at t = 0, find (a) Q and (6) 7 at t > 0., Potential drop across, Potential drop across, Then, Since, , (Í), we have, , so that (1) becomes, or, , (2), , Integrating the second equation of (2) yields, (3), Since / = 0 when Q = Q0, we have Ct = Q«/2C. Thus (3) becomes on solving for /,, (4), Separating variables in (4) and, or, , (5), , Since Q = Q0 for t = 0, we find C2 = v/2. Thus from (5),, or, and, , Note that the charge oscillates with amplitude Q0, period Zv-^LC and frequency (I/period), The current oscillates with amplitude Q0/\/Zc and the same period and frequency as the, charge.
Page 68 :
[CHAP. 2, , ORDINARY DIFFERENTIAL EQUATIONS, , 58, , GEOMETRY PROBLEMS, 2.35. (a) Find the orthogonal trajectories of the family of curves y = ex2 and (b) give a, geometrical interpretation., (a) The differential equation of the family is, , Since the slope of each member of the orthogonal family must be the negative reciprocal of, this slope, we see that the slope of the orthogonal family is, , Solving this we find that the equation of the orthogonal trajectories is, , x* + 2y2 = k, (b) The family y = ex2 is a family of parabolas while the orthogonal family *2 + 2j/2 = fc is a, family of ellipses [see Fig. 2-llj., , Fig. 2-11, , Fig. 2-12, , 2.36. Determine a curve such that the length of its tangent included between the x and y, axes is a constant a > 0., Let (x, y) be any point P on the required curve and (X, Y) any point Q on the tangent line AB, [Fig. 2-12]., The equation of line AB passing through (x, y) with slope y' is, Y-y = y'(X-x), Set X — 0, Y — 0 to obtain the y and x intercepts, OA = y — xy',, , OB = x — yly' = —(y — xy')/y', , Then the length of AB, apart from sign, is, Since this must equal ±a, we have on solving for y,, where y' = p., To solve (1) differentiate both sides with respect to x, so that, , or, , (¿)
Page 69 :
CHAP. 2], , ORDINARY DIFFERENTIAL EQUATIONS, , Case 1, dp/dx — 0., , In this case p = e and the general solution is, , y, Case 2, dp/dx ¥• 0., , 59, , = ex ±, , (2), , In this case, using (1) we find, , x =, Then, , x*'*, , a;2/3 + j/2/3 = a2/3, , so that, , (5), , which is a singular solution., The curve (3) which is a hypocycloid [Fig. 2-13] and which is, the envelope of the family of lines (2), is the required curve., , Fig. 2-13, , FLOW PROBLEMS, , 2.37. A cylindrical tank has 40 gallons (gal) of a salt solution containing 2 Ib dissolved salt, per gallon, i.e. 2 Ib salt/gal. A salt solution of concentration 3 Ib salt/gal flows into, the tank at 4 gal/min. How much salt is in the tank at any time if the well-stirred, mixture flows out at 4 gal/min?, Let the tank contain A Ib salt after í minutes., Rate of change of amount of salt, , =, , Then, , Rate of entrance, , dA, A, Solving the equation -JT = 12 - — subject to A = 40 gal •, , find A = 120 - 40«-«/io., , 2.38. A right circular cone [Fig. 2-14] is filled with water., In what time will the water empty through an orifice, O of cross-sectional area a at the vertex? Assume, velocity of exit is v =, where h is the instantaneous height ("head") of the water level above 0, and K is the discharge coefficient., At time í the water level is at h. At time t + dt, dt > 0,, the water level is at h + dh where dh < 0. We have, Change in volume of water, z, , —¡rr dh, , =, , Amount of water leaving, , =, , avdt, , —, , From similar triangles OAB and OEF, r = Rh/H., , Solving, subject to h = H at t — 0, we have, , t, , -, , Then time for emptying, when h = 0, is T =, , Then, , =, , — Rate of exit, , 8, , 0 Ib salt at t = 0, we
Page 70 :
60, , ORDINARY DIFFERENTIAL EQUATIONS, , [CHAP. 2, , CHEMISTRY PROBLEMS, , 2.39. Radium decays at a rate proportional to the instantaneous amount present at any, time. If the half life of radium is T years, determine the amount present after, í years., Let A gm of radium be present after í years. Then, Time rate of change of A <* A, a, , or, , A, , = — kA, , Solving subject to A = A0 at t = 0, we find A = A0e~kt., The half life is the time T when the amount present is half the original amount, i.e. A0/2., Then, = A 0 e-w, , or, , e-kT, , Another method. e~kT = \,, , = ^, , = (£)i/T, , e-k, , A, , = A0(e~kY = A0(i)t/T, , A = A 0 e-* In2 / T ., , and, , k -, , and, , 2.40. Chemical A dissolves in solution at a rate proportional to the instantaneous amount, of undissolved chemical and to the difference in concentration between the actual, solution Co and saturated solution Cs. A porous inert solid containing 10 Ib of A is, agitated with 100 gallons water and after an hour 4 Ib of A is dissolved. If a saturated, solution contains .2 Ib of A per gallon, find (a) the amount of A which is undissolved, after 2 hr and (6) the time to dissolve 80% of A., Let x Ib of A be undissolved after t hours. We have, « x(Cs-Ca), Then, , or, , = kx, , = =, , Kx(x, , +10), , Separating the variables and integrating,, , i.e., , =, , «i + Cj, , Using the conditions ( = 0, x = 10 and t = 1, x — 6, we find, x =, , (a) When í = 2 hr, * = 3.91 Ib of A undissolved., (6) When * = 2 Ib, (f )' = £ and t = 3.82 hr., , TEMPERATURE PROBLEM, , 2.41. Newton's law of cooling states that the time rate of change in temperature of an, object varies as the difference in temperature between object and surroundings. If, an object cools from 80°C to 60°C in 20 minutes, find the temperature in 40 minutes, if the surrounding temperature is 20°C., Let U = temperature of object after í minutes. Then, <* U-20, kt, , or, , = fc(C7-20), , Solving, U = 20 + ce . At t = O, U = 80 so that c = 60 and U - 20 + 60e«., U = 60 so that ewk - 2/3, ek = (2/3)1/20. Then, U - 20 + 60e« = 20 + 60(e")* = 20 + 60(|)«/2<>, When í = 40, U = 20 + 60(2/3)2 = 46.7°C., , At í = 20,
Page 71 :
CHAP. 2], , ORDINARY DIFFERENTIAL EQUATIONS, , 61, , BENDING OF BEAMS, , 2.42. A beam of length L is simply supported at, both ends [Fig. 2-15]. (a) Find the deflection, if the beam has constant weight W per unit, length and (b) determine the maximum, deflection., , Fig. 2-15, , (a) The total weight of the beam is WL, so each end supports weight %WL. Let x be the distance, from the left end A of the beam. To find the bending moment M at x, consider forces to the, left of x., (1) Force \WL at A has moment -(%WL)x., (2) Force due to weight of beam to left of x has magnitude Wx and moment, Wx(x/Z) - $Wx*, Then the total bending moment at * is %Wx2 — \WLx. Thus, Ely" = %Wx*-$WLx, Solving this subject to y' = 0 at x = LIZ [from symmetry] and y = 0 at x = 0, we find, y = •, , (x* - 2Lx* + L*x), , This could also be obtained using conditions y — 0 at x = 0 and x — L., (b) The maximum deflection occurs, at x = L/2 and is 5WL4/S84EI., Note that if forces to the right of x had been considered, the bending moment would be, -%WL(L -x) + W(L -x), , as above., , = %Wx* - \WLx, , 2.43. A cantilever beam [Fig. 2-16] has one end, horizontally imbedded in concrete and a force, W acting on the other end. Find (a) the deflection and (6) maximum deflection of the, beam assuming its weight to be negligible. :, (a) Considering the portion of the beam to the right, of x, the bending moment at x is W(L — x). Then, Ely" = W(L-x), Fig. 2-16, Solving this subject to y' = 0 at * = 0 and y = 0 at x - 0, we find, (3Lx2-x*), , y =, , (b) The maximum deflection occurring at x = L is WL3/3EI., , NUMERICAL METHODS, , 2.44. If dy/dx = 2x + y, y(0) - 1, (a) find the approximate value of y(.5) using the step, by step or Euler method with h = .1 and (6) compare with the exact solution., (a) To obtain the entries in the first line of the table below, use the fact that a:0 = 0, y0 — 1.0000, so that y' - 2x0 + y0 - 1.0000., x, , yy, , y', y' = 2x, 2x + y, , .0, .1, .2, .3, .4, .5, , 1.0000, 1.1000, 1.2300, 1.2300, 1.3930, 1.5923, 1.8315, , 1.0000, 1.3000, 1.6300, 1.6300, 1.9930, 1.9930, 2.3923, 2.3923
Page 72 :
62, , ORDINARY DIFFERENTIAL EQUATIONS, , [CHAP. 2, , To obtain the entries in the second line use h — .1 and equation (10), page 43, to obtain, y(x0 + h), , =, =, =, , y(x0) + hf(x0,y0), value of y in first line + (.l)(slope in first line), 1.0000 + (.1)(1.0000) = 1.1000, , and the corresponding slope is, , y' = 2(.l) + 1.1000 = 1.3000, , Similarly we have, value of y in third line =, =, =, and, slope in third line =, , value of y in second line + (.1) (slope in second line), 1.1000 + .1(1.3000), 1.2300, 2(.2) + 1.2300 = 1.6300, , The remainder of the entries are obtained by continuing in this manner and we find finally, 2/(.5) = 1.8315., (6) The equation is linear and has integrating factor e~x. Solving, we find y = 3ex — 2x — 2., Then when * = .5, y = 3e-5 — 3 = 3(1.6487) — 3 = 1.9461. Better accuracy can be obtained by, using smaller values of h or proceeding as in Problem 2.45 which is a modification of the, method., , 2.45. Show how to improve the accuracy of the method of Problem 2.44., The method which we shall use is essentially the same as that of Problem 2.44 except that, we obtain improved values of y and y', which we denote by y\,y'\', y2,y21 etc., as indicated in the, table below. Success is achieved when there is very little difference between the improved values, of y [see columns headed 3/3 and 3/4 respectively in the table]., The first line in the table below is the same as the first line in the table of Problem 2.44 and, is found in exactly the same way., Similarly the entries corresponding to y1 and y[ in line two of the table below are the same, as those in line two of the table of Problem 2.44., We now refer only to the table given below. As seen from this table, the slopes corresponding to * = .0 and x = .1 are 1.0000 and 1.3000 respectively. Then the average [mean], slope is ¿(1.0000 +1.3000) = 1.1500. Using this modified slope, we find for the corresponding value, °f y, , 2/2 = 1.0000 + .1(1.1500) = 1.1150, , (1), , The slope corresponding to this value of y is given by, y'2 = 2(.l) + 1.1150 = 1.3150, using the given differential equation., , (2), , From the improved slope (2) we now obtain an improved average slope, ¿(1.0000 + 1.3150) = 1.1575, , (S), , and from this an improved value of y given by, ya = 1.0000 + (.1)(1.1575) = 1.1158, , (¿), , which in turn gives an improved slope equal to, ya = 2(.l) + 1.1158 = 1.3158, , (5), , The improved average slope using (5) is then, ¿(1.0000 + 1.3158) = 1.1579, and so, , i/4 = 1.0000 + (.1)(1.1579) = 1.1158, , (6), , Since this agrees with (4), the process ends., We now use yt = 1.1158 and y'z = 1.3158 to obtain the entries in the third line., and, , yl (line 3) =, , 1.1158 + (.1)(1.3158) = 1.2474, , y[ (line 3) =, , 2(.2) + 1.2474 =, , 1.6474, , Thus
Page 73 :
CHAP. 2], , ORDINARY DIFFERENTIAL EQUATIONS, , 63, , Continuing in this manner we finally obtain y(.5) = 1.9483, the last entry of the last line in, the table. This agrees closely with the true value 1.9461., *, , Vi, , y'i, , Vz, , VÍ, , Vi, , Vs, , 2/4, , .0, .1, .2, .3, .4, .5, , 1.0000, 1.1000, 1.2474, 1.4313, 1.6557, 1.9247, , 1.0000, 1.3000, 1.6474, 2.0313, 2.4557, 2.9247, , 1.1150, 1.2640, 1.4496, 1.6759, 1.9471, , 1.3150, 1.6640, 2.0496, 2.4759, 2.9471, , 1.1158, 1.2648, 1.4505, 1.6769, 1.9482, , 1.3158, 1.6648, 2.0505, 2.4769, 2.9482, , 1.1158, 1.2648, 1.4506, 1.6770, 1.9483, , 2.46. Solve Problem 2.44 by the Taylor series method., By successively differentiating the given differential equation, we have, y' = Zx + y, y" = 2 + y', y'" = y", yW = y'", y<^ = j/«v>,, Thus, , y(0) = 1,, , y'(0) = 1,, , y" = 3,, , y'" = .3,, , i/«v> = 3,, , y<v> = 3,, , and so, , Then when x = .5,, 3/(.5), , 2.47. Solve Problem 2.44 by Picard's method., We have on integrating the differential equation using the boundary condition,, V(x), , Letting yl = 1 as a first approximation we find the second approximation, V<L(X), Then a third approximation is, Va(x), , Proceeding in this manner we find the successive approximations, y4(x), , Vs(x), , -, , ..., ...
Page 74 :
ORDINARY DIFFERENTIAL EQUATIONS, , g4, , [CHAP. 2, , Ve(x), , Putting x = .5 in this last approximation, we find, 3/6(.5) = 1 + .5 + .375 + .0625 + .0078 + .0008 = 1.9461, , 2.48. Solve Problem 2.44 by the Runge-Kutta method., Using h = .5, x0 = 0, y0 — 1, f ( x , y) = Zx + y in the equations on page 44,, fcj = .5[2(0) + 1] = .5, fc2 = .5[2(.25) + 1.25] = .875, ka = .5[2(.25) + 1.4375] = .96875, fc4 = .5[2(.5) + 1.9675] = 1.48375, Then, , y(.5) = 1 + J(.5 + 1.750 + 1.9375 + 1.48375) = 1.9452, , Better approximations can be obtained by using two or more applications of the method with, smaller values of h., , 2.49. (a) Show how to solve numerically the system of equations, dyldx = f(x,y,v),, dv/dx = g(x,y,v); y(x0) — i/o,, by the Runge-Kutta method., (b) Use (a) to show how to solve numerically, , v(x0) = Vo, , (a) Using the results on page 44, we write, , so that, (6) The equation dïy/dx2 = g(x, y, dy/dx), , is equivalent to the system, , dyldx = v,, , dv/dx = g(x,y,v), , and comparing with the equations of (a) we have f ( x , y, v) = v., y(x0 + h) can be written in terms of Z1( 12, /3 as, , Then the expression for, , 2.50. If, f 2y = x; y(Q) = 1, y'(Q) use Problem 2.49 to find an approximate, value of y(.5)., The differential equation is equivalent to the system, dy/dx, , = v,, , dv/dx = x — 2y + 3v, , and with the notation of Problem 2.49(6) we have g(x, y,v) — x — 2y + Sv, x0 — 0, y0 = 1, V0 — 0., Thus if h = .5 we have from Problem 2.49(a),, li = .5(-2) = -1,, , Z2 = .5(-3.25) = -1.625,, , 13 = .5(-3.9375) = -1.96875
Page 75 :
CHAP. 2], , ORDINARY DIFFERENTIAL EQUATIONS, , Then, , 65, , 3/(.5) = 1 + ^ (-1 - 1.625 - 1.96875) = .6172, D, , The equation can be solved exactly and the solution is, y(.S) = e-» - fe + 1 = 1.6487 - f (2.7183) + 1 = .6100., , y — ex — %e2x + ^* + f, , so that, , 2.51. Work Problem 2.50 by the Taylor series method., We have y' — v, v' — x — 2y + 3v so that by successive differentiation,, , - v',, , y", , v" - I - 2y' + 3v',, , y'" = v",, , V" = -2y" + 3v",, , ..., , and we find corresponding to x — 0,, y', , =0, v' = -2, y" = -2, v" = -5, y'" = -5, v'" = -11,, , y™ = -11,, , = -23, i> (v > = -47, 2/ (vi) = -47, f (vi) = -95,, , yW> = -95,, , yw, , i>(iv) = -23, ..., , Then, , Thus we have approximately, 2/(.5) = 1 - .25 - .104167 - .028646 - .005990 - .001020 - .000147, =, , .61003, , Supplementary Problems, CLASSIFICATION OF DIFFERENTIAL EQUATIONS, 2.52. Classify each of the following differential equations by stating the order, the dependent and, independent variables and whether the equation is ordinary or partial., , (c), , xyy', , =, , (y")3, , (e), , y", , (/), , d(uv), , SOLUTIONS OF DIFFERENTIAL EQUATIONS, 2.53. Check whether each differential equation has the indicated solution., general solutions., (a), , y" — 2y' + y = x;, , (b), , tl'(t) + I(t) - «2;, , ( c ) y = xy' +, , (d), , [1 + i/'2]3/2, , =, =, , v*dv, , Determine which solutions are, , y = (c¡ + c2x)ex + x + 2, t? - 3tl - e, , 27x* + 256# = 0, , f ley = 0;, , V = el sin 2t + c2 cos 2t + c3t sin 2í + c4í cos 2<, , 2.54., , (a) Show that y = e~x(cl cos x + C2 sin x) is a general solution of y" + 2y' + 2y = 0. (6) Determine the particular solution such that j/(0) = —2, y'(0) = 5., , 2.55., , (a) Show that y - ex + 3c2 + c and (* + I)2 + 12y = 0 are solutions of the differential equation, y — xy' + 3y'2 + y'. (b) What is the name given to each of these solutions? (c) Explain the relationship between these solutions and illustrate graphically.
Page 76 :
66, , ORDINARY DIFFERENTIAL EQUATIONS, , [CHAP. 2, , DIFFERENTIAL EQUATION OF A FAMILY OF CURVES, 2.56., Find differential equations for the following families of curves: (a) »2 + cyz = 1, (6) y2 = ax + b., 2.57., , Find the differential equation for (a) the family of straight lines which intersect at the point, (2,1) and (b) the family of circles tangent to the x axis and having unit radius., , MISCELLANEOUS TYPES OF DIFFERENTIAL EQUATIONS, 2.58., (a) Find the general solution of dy/dx = 8x3 — 4* +1. (b) Determine the particular solution such, that j/(l) = 3., 2.59., , Solve each of the following boundary-value problems., (a), W, (c), , 2.60., , Solve (a), , (b) xy' + 3y = 0; j/(l) = 2, (c), , (e) (l-»2)j/' = 43/; »(0) = 1., 2.61., , Solve, , 2.62., , Each of the following differential equations has an integrating factor depending on only one, variable. Find the integrating factor and solve the equation., (a) (4j/ - *2) dx + x dy = 0, (6) (2*j/2 - y) dx + (2x - x*y) dy = 0, (c) 2dx + (2x -3y-3)dy = 0; j/(2) = 0, (d) (2y sin x + 3y* sin x cos x) dx — (4ys cos2 x + cos «) dy = 0, , 2.63., , Solve each of the following differential equations given that each has an integrating factor of the, form xfyt;, (a) (Sy-2xy*)dx + (4x-3x*v*)dy = 0, (b) (2xy* + 2y) dx + (x*y* + 2x) dy = 0, , 2.64., , Solve, , (a), (6), , 2.65., , Solve, , (, , a, , ), , j, , /, , (, , l, , ) = 2 , ( b ) xy'- 4y = *,, , (, , (d) -%ax + 2y cot x = esc x., 2.66., , Solve (a), , 2.67., , Solve, , (a), , 2.68., , Solve, , (a), , 2.69., , Solve, , 2.70., , Solve, , (6), , (c), (6) 2a;2j/' = xy + j/«., (6), , (a) j/'2 + (y ~ l)y' - y = 0,, , (6) (xy' + y)2 = «-«., , (d)
Page 77 :
CHAP. 2], , ORDINARY DIFFERENTIAL EQUATIONS, , 2.71., , Solve each of the following, determining any singular solutions. In each case p = y'., (a) y = px + 2p2, (&) xp2 = 2y(p + 2), (c) (xp - y)2 = P2-1, (d) x2y = x*p - yp2., , 2.72., , Solve (a) xy" - Sy' = x2, (b) yy" + 2y'2 = 0, (c) y" + iy = 0, (d) (1 + j/'2)« = (y")2., , 67, , EXISTENCE AND UNIQUENESS OF SOLUTIONS, 2.73. Use the theorem on page 41 to discuss the existence and uniqueness of solutions for each of the, following: (a)y' = 2xy, y(0) = 1, (b) y = xy'- y'2; y(2) = 1, (c) y' = (y + x)/(y - x); j/(l) = 1., MISCELLANEOUS APPLICATIONS, 2.74. An object moves along the x axis, acted upon by a constant force. If its initial velocity in the, positive direction is 40 meters/sec while 5 seconds later it is 20 meters/sec, find (a) the velocity at, any time, (6) the position at any time assuming the object starts from the origin x = 0., 2.75., , A 64 Ib object falls from rest. The limiting velocity is 4 ft/sec. Find the velocity after t seconds, assuming a force of resistance proportional to (a) v, (b) v2, (c) -\fv., , 2.76., , A particle is located at x = a (a > 0) at t = 0. It moves toward x = 0 in such a way that its, velocity is always proportional to xn where M is a constant. Show that the particle will reach, x — 0 if and only if n < 1., , 2.77., , An electric circuit contains an 8 ohm resistor in series with an inductor of .5 henries and a battery, of E volts. At * = 0 the current is zero. Find the current at any time í > 0 and the maximum, current if (a) E - 64, (6) E = %te-™, (c) E = 32e-"»., , 2.78., , (a) Solve for the current in the circuit of Problem 2.77 if E = 64 sin St. (b) What is the transient, current and steady-state current?, , 2.79., , An electric circuit contains a 20 ohm resistor in series with a capacitor of .05 farads and a battery, of E volts. At t — 0 there is no charge on the capacitor. Find the charge and current at any, time t > 0 if (a) E - 60, (6) E = lOOíe-2*., , 2.80., , Find the charge and current in the circuit of Problem 2.79 if E — 100 cos 2i., , 2.81., , Find the orthogonal trajectories of the family of curves, (a) xy = c, (b) x2 + y2 = ex, (c) y2 = ex2 - 2y, , 2.82., , Find the equation of that curve passing through (0,1) which is orthogonal to each member of the, family x2 + y2 = cex., , 2.83., , A curve passing through (1, 2) has the property that the length of the perpendicular drawn from, the origin to the normal at any point of the curve is always equal numerically to the ordinate of, the point. Find its equation., , 2.84., , The tangent to any point of a certain curve forms with the coordinate axes a triangle having, constant area A. Find the equation of the curve., , 2.85., , A tank contains 100 gallons of water. A salt solution containing 2 Ib of salt per gallon flows in at, the rate of 3 gallons per minute and the well-stirred mixture flows out at the same rate, (a) How, much salt is in the tank at any time? (b) When will the tank have 100 Ib of salt?, , 2.86., , In Problem 2.85, how much salt is in the tank at any time if the mixture flows out at (a) 2 gal/min,, (6) 4 gal/min., , 2.87., , A right circular cylinder of radius 8 ft and height 16 ft whose axis is vertical is filled with water., How long will it take for all the water to escape through a 2 in2 orifice at the bottom of the tank, assuming the velocity of escape v in terms of the instantaneous height h is given by v = .6^/2gh., , 2.88., , Solve Problem 2.87 if the cylinder has its axis horizontal.
Page 78 :
68, , ORDINARY DIFFERENTIAL EQUATIONS, , [CHAP. 2, , 2.89., , The rate at which bacteria multiply is proportional to the instantaneous number present., original number doubles in 2 hours, in how many hours will it triple?, , If the, , 2.90., , After 2 days, 10 grams of a radioactive chemical is present. Three days later 5 grams is present., How much of the chemical was present initially assuming the rate of disintegration is proportional, to the instantaneous amount which is present?, , 2.91., , Find the half life of a radioactive substance if three quarters of it is present after 8 hours., , 2.92., , Chemical A is transformed into chemical B at a rate proportional to the instantaneous amount of, A which is untransformed. If 20% of chemical A is transformed in 2 hours, (a) what percentage, of A is transformed in 6 hours and (b) when will 80% of A be transformed?, , 2.93., , It takes 15 minutes for an object to warm up from 10°C to 20°C in a room whose temperature is, 30°C. Assuming Newton's law of cooling, how long would it take to warm up from 20°C to 25°C?, , 2.94., , At 1:00 P.M. the temperature of a tank of water is 200°F. At 1:30 P.M. its temperature is 160°F., Assuming the surrounding temperature is maintained at 80 °F, (a) what is the temperature at, 2:00 P.M. and (6) at what time will the temperature be 100°F? Assume Newton's law of cooling., , 2.95., , A beam of length L ft and negligible weight is simply supported at the ends and has a concentrated load W Ib at the center. Find (a) the deflection and (6) the maximum deflection., , 2.96., , A cantilever beam of length L ft has a weight of w Ib/ft., maximum deflection., , Find (a) the deflection and (6) the, , NUMERICAL METHODS, 2.97. Use the step-by-step or Euler method to solve numerically each of the following., (a) dy/dx = y, y(0) = 1; find y(.2) using h - .05., (6) dy/dx = x + y, y(0) = 0; find y(.5) using h = .1., (c) dy/dx = Sx — y, y(l) = 0; find y(.5) using h = .1., 2.98., , Work Problem 2.97 using the modified step-by-step or Euler method., , 2.99., , Work Problem 2.97 using (a) the Taylor series method, (b) Picard's method, (c) the Runge-Kutta, method., , 2.100. (a) Given dy/dx - (x + y)2, y(0) = 1, find an approximate value for y(.2) by using an appropriate, numerical method. (6) Compare with the exact solution obtained by using the transformation, x + y — i)2., 2.101. If tPy/daP + y = x, j/(0) = 1, j/'(0) = 0, find y(.5) using the (a) Euler method, (6) Taylor series, method, (c) Picard method, (d) Runge-Kutta method., 2.102. If, , h y — 2x, y(l) = 0, j/'(l) = 0 find y(A) using an appropriate numerical method.
Page 79 :
CHAP. 2], , ORDINARY DIFFERENTIAL EQUATIONS, , 69, , Answers to Supplementary Problems, 2.52., , (a) order 3, dep. var. U, ind. var. t, ordinary. (V) order 2, dep. var. T, ind. var. x,y,z, partial., (c) order 2, dep. var. y, ind. var. x, ordinary, (d) order 3, dep. var. s, ind. var. t, ordinary, (e) order, 2, dep. var. y, ind. var. x, ordinary. (/) order 1, dep. var. u (or v), ind. var. v (or u), ordinary., , 2.53., , All solutions except for (c) are general solutions., , 2.54., , (6) y - e~x(3 sin x — 2 cos*), , 2.55., , The first is a general solution and the second a singular solution., , 2.56., , (a) (1 - x2)y' + xy = 0, , 2.57., , (a) y' = (y-l)/(x-2), , 2.58., , (a) y - 2x* - 2x2 + x + c, , 2.59., , (a) y = 9*7/3 - 2x - 7, ,,,, 1., . x4 2x . 5, to, W » = 2 *+ 24~T+8, , 2.60., , (a) y = ce~xí, (b) y = 2/x*, , 2.61., , (a) x2 + 4xy — 5y2 = c, (b) 2xzy2 - 3x + 2y3 = -3, , 2.62., , (a) a;3; «-ty - £ic6 = c, (6) 1/3/3; a^ — x = cy2, , 2.63., , (a) 2x3y* — x*ys = c; int. factor x2y3, , 2.64., , (a) 2 tan~i(y/x) + 4» + j/2 = c, , 2.65., , (a) xy = x* + 1, (6) » + 3# = ex*, , 2.66., , (a) x2-xy = cy, , (c) x2 - 3xy - 2y2 = c, , (b) x = ce-*'<-* + y>, , (d) ln(5a;2-4i»;3/ + 2/2) + 2tan-i, , (6) yy" + y'* = 0, , (b) (y - 1)2(1 + y'2) = 1, (b) y - 2x* - 2x2 + x + 2, , (c) s = 2 + 12t - 2ts - 16 sin t, , (e) ^ = (1 + a;)2/(l - x)2, , (c) I = 5(l-e-2t), (d), , (c) yex — xe~y = c, (d) x2 cos y + j/3 tan 2x - 4x = c, (c) e«; e»(2a; - 3y) = 4, (d) cos x; y cos2 a; + y* cos3 » = c, (b) 3xy2 + 2 = cx3ys; int. factor x~*y~*, , (6) a:2 - 2/2 = ce~2*, , (¿) y ~ e3x — 2x — 1, (d) y sin2 » = c — cos », , 2.67., , (a) xy(c-x2) = 2, , (b) xe*lv* = c, , 2.68., , (a) (* - 2)2 + 4(« - 2)(» - 1) - fo - I)2 = c, (6) *2 _ 2xy + y2 + 2x + 2y - c, , 2.69., , (as + y + 1)(1 - *) = 1, , 2.70., , (a) (y - x + c)(y - cex) = 0, , 2.71., , (a), (b), (c), (d), , (b) x(xy - c)2 - 4, , y = ex + 2c2; sing, solution y = -x2/8., (x — c)2 = cy; sing, solutions y = 0, 4x + y = 0., (ex — y)2 = e2 — 1; sing, solution x2 — y2 = 1., y2 = ex2 - c2; sing, solutions y = ±x2/2., , =c
Page 80 :
70, 2.72., , ORDINARY DIFFERENTIAL EQUATIONS, (a) y = -$xs + CjX4 + C2, , (c) y = ct sin 2x + C2 cos 2», , (6) j/s =, , (d) (x - a)2 + (y - Í»)2 = 1, , Clx + c2, , or, , [CHAP. 2, y = a sin (2* + 6), , (6) x = 40t - 2i2, , 2.74., , (a) v = 40 - 40i, , 2.75., , (a) v = 4(1-e-»'), , 2.77., , (a) / = 8(1 —e~16t), max. current = Samp, (6) / = 8(2e-16t, max. current = e~2/8 = .01692 amp, (c) / — 8(e~ 8t — e~16i), max. current = 2 amp, , 2.78., , (a) / = 6.4 sin 8t - 3.2 cos 8i + 3.2e-i«, (6) Transient current = 3.2e~16t, steady-state current = 6.4 sin 8i — 3.2 cos 8t, , 2.79., , (a) Q = 3(1-«-*), / = 3e~', (6) Q = Se-* - 5e-2« - 5te~**, I = lOie-2' + 5e~* - 5e~*, , 2.80., , Q = 2 sin 2t + cos 2i - e~*, / = 4 cos 2i - 2 sin 2i + e~*, , 2.81., , (a) w2 - 1/2 = fc (6) a;2 + j/2 = %, , 2.82., , * = j/ tan, , 2.85., , (a) 200(1 - e~-o«), , 2.86., , (a), , 2.87., , 6.72 hr, , 2.90., , 10\^4 or 21.54 gm, , 2.92., , (a) 48.8% (6) after 14.43 hr, , 2.94., , (a) 133.3°F, , (6) about 3:12 P.M., , 2.95., , (a) y =, , (3L2* - 4œ») for 0 g * S LIZ and is symmetrical for LIZ S « ë L, , 2.96., , (a) y =, , (x* - 4Lx* + 6L2»2), , 2.97., , (a) Exact value = 1.2214, , (6) v = 41, , = 4tanh8i, , (c) y + 1 = feeV4(x2+2!/+!'2), , 2.83. a2 + j/2 = 5w, , 2.84. 2*j/ = A, , (6) after 23.1 minutes, , (100 + t) -, , (6) 3(100 -1) - 300, , 2.88. 5.69 hr, , 2.100. Exact value = 1.5087, 2.101. Exact value = .89815, 2.102. Exact value = .14776, , , 0 S t S 100, , 2.89. 3.17 hr, , 2.91., , 19.26 hr, 2.93. 30 min, , (6), , (6) Exact value = 2.2974, , (c) Exact value = —1.6, , (V) WLS/48EI
Page 81 :
Chapter 3, Linear, Differential Equations, , GENERAL LINEAR DIFFERENTIAL EQUATION OF ORDER n, The general linear differential equation of order n has the form, , (1), A differential equation which cannot be written in this form is called nonlinear., = sin a; is a second order linear equation., , Example 1., Example 2 . =, , e~x i s a second order nonlinear equation., /, , \, , If R(x), the right side of (J), is replaced by zero the resulting equation is called the complementary, reduced or homogeneous equation. If R(x) ¥= 0, the equation is called the, complete or nonhomogeneous equation., Example 3. If, , = sin», , is the complete equation, then, , -, , Zxy = 0 is the corresponding complementary, reduced or homogeneous equation., , If ao(x), ai(x), .. .,an(x) are all constants, (1) is said to have constant coefficients,, it is said to have variable coefficients., , otherwise, , EXISTENCE AND UNIQUENESS THEOREM, If a0(x), ai(x), ..., an(x) and R(x) are continuous in the interval \x — Xo\ < 8 and, a<>(x) ¥= 0, then there exists one and only one solution to (Í) which satisfies the conditions, • • • ' »("~"(*o) = »o"~", , »(*o) = 2V »'(*„) = V'v, , (*), , OPERATOR NOTATION, It is sometimes convenient to adopt the notation Dy, D2y, ..., Dny to denote, The symbols D, D2, ... are called differential operators and have properties analogous to those of algebraic quantities. Using this notation, we shall agree to, write (1) as, [a»(x)D« + ai(x)D»-i + • • • + a*-i(x)D + an(x)]y = R(x), (3), or briefly, , <j)(D)y, , = R(x), , where <t>(D) = a0(x)Dn + a1(x)Dn~1 + • • • + o»-i(a;)Z> + o»(a;) is called an operator polynomial, in D., Example 4., , - 2xy = sin * can be written, 71, , (xD2 + 3D - 2x)y - sin x.
Page 82 :
72, , LINEAR DIFFERENTIAL EQUATIONS, , [CHAP. 3, , LINEAR OPERATORS, , An operator L is called a linear operator if for any constants A, B and functions u, v to, which L can be applied, we have, L(Au + Bv) = AL(u)+BL(v), The operators D,D2,... and <j>(D) are linear operators [see Problem 3.3]., FUNDAMENTAL THEOREM ON LINEAR DIFFERENTIAL EQUATIONS, , In order to find the general solution of, 4>(D)y = R(x), (4), where R(x) ¥* 0, let Yc(x) be the general solution of the complementary reduced or homogeneous equation, my = Q, (5), We often refer to Yc(x) as the complementary or homogeneous solution. Then we have the, following important theorem, sometimes referred to as the superposition principle or, theorem., Theorem 3-1. The general solution of (4) is obtained by adding the complementary solution Yc(x) to a particular solution Yp(x) of (4), i.e., y = Yc(x) + Yp(x), Example 5. The general solution of (Z>2 — 3D + 2)y = 0 is y - c^* + C2e2x and a particular, solution of (D2 — 3D + 2)y = 4x2 is 2x2 + 6x + 1. Then the general solution of, (Tft - 3D + 2)y = 4a;2 is y = c^* + c2e*x + Zx2 + 6x + 1., , Because of this theorem it is clear that we shall have to consider separately the problems, of finding general solutions of homogeneous equations and particular solutions of nonhomogeneous equations., LINEAR DEPENDENCE AND WRONSKIANS, , A set of n functions yi(x), yz(x), ..., yn(x) is said to be linearly dependent over an interval, if there exist n constants Ci, cz, ..., cn, not all zero, such that, c\yi(x) + c2y2(x) + • • • + cnyn(x) = 0, identically over the interval. Otherwise the set of functions is said to be linearly, independent., Example 6. 2eSx, 5eSx, e~4x are linearly dependent over any interval since we can find constants, c, i> "2. cs n°t all zero such that c^Ze3*) + c2(5e3x) + cs(e~*x) = 0 identically; for instance, Cj = —5, c2 = 2, c3 = 0., Example 7. ex and xex are linearly independent since Cje* + c2xex = 0 identically if and only if, «i = 0, C2 = 0., , Theorem 3-2. The set of functions 3/1(0;), y2(x), ..., yn(x) [assumed differentiate] is linearly, independent on an interval if and only if the determinant, 2/1 (*), , W(yl,y2,...,yr), , =, , y2(x), , 2/;(z), , 2/2 (<*), , y?-»(x), , y?-»(x), , • • • yn(*), • • • y'n(x}, , ... y?-»(x), called the Wronskian of yi,..., yn is different from zero on the interval.
Page 83 :
CHAP. 3], , LINEAR DIFFERENTIAL EQUATIONS, , 73, , This last theorem is important in connection with solutions of the homogeneous or, reduced equation as seen in the following, Theorem 3-3., , [Superposition Principle]. If yi(x),y2(x), ...,yn(x) are n linearly independent solutions of the wth order linear equation <j>(D)y = 0, then, y = ciyi(x) + c2y2(x) + • •• + cnyn(x), where ci, c2, ...,cn are n arbitrary constants, is the general solution of, 4>(D)y = 0., , SOLUTIONS OF LINEAR EQUATIONS WITH CONSTANT COEFFICIENTS, The remarks made so far have applied to the general equation (1). Particular simplifications occur when the equation has constant coefficients and we now turn to this case., Two general procedures are available in this case, namely those which do not involve, operator techniques and those which do. For each case methods exist for finding complementary and particular solutions and use is then made of the fundamental Theorem 3-1., NON-OPERATOR TECHNIQUES, I. THE COMPLEMENTARY OR HOMOGENEOUS SOLUTION, Let y = emx, m = constant, in (a0Dn + aiDn~l + • • • +an)y = 0 to obtain, a0mn + aim"-1 H, , + a« = 0, , (6), , which is called the auxiliary equation or characteristic equation. This can be factored, m, , °, , a0(w - mi)(m - m2) • • • (m - mn) - 0, , (7), , which has roots mi, m2, ..., mn. Three cases must be considered., Case 1. Roots all real and distinct., Then emix, em*x, ..., em"x are n linearly independent solutions so that by Theorem 3-3, the required solution is, y = ciemix + c2em*x + • • • + cnem«x, (8), Case 2. Some roots are complex., If ao, 01, ..., an are real, then when a + bi is a root of (6) so also is a — bi [where, a, b are real]. Then a solution corresponding to the roots a + bi and a — bi is, y = eax(ci cos bx + c2 sin bx), (9), in, where use is made of Euler's formula e — cos u + c sin u., Case 3. Some roots are repeated., If mi is a root of multiplicity k, then a solution is given by, y = (GI + c2x + c3x2 + • • • + ckxk~l)em^x, II. THE PARTICULAR SOLUTION, Two important methods for finding a particular solution of 4>(D)y = R(x), available., , (10), , are, , 1. Method of Undetermined Coefficients., In this method we assume a trial solution containing unknown constants (indicated by a, b, c, ...) which are to be determined by substitution in the given equation., The trial solution to be assumed in each case depends on the special form of R(x), and is shown in the following table. In each case /, g, p, q are given constants and, k is a given positive integer.
Page 84 :
74, , [CHAP. 3, , LINEAR DIFFERENTIAL EQUATIONS, , R(x), , Assumed Trial Solution, , fepx, , ae"*, , f cos px + g sin px, , a cos px + b sin px, , /„** + ft**-» + • • • + / *, , auxk + a1xk~1 + • • • + ak, , ew(/ cos px + g sin qx), , eix(a cos px + b sin px), , ««(/„*" + /! a*- 1+ ••• +/ fc ), , e"x(a0xk + a&K-1 H, , (/o!>;'c+ •'•+/*) cos pa, + too*"" H, 1- fffc) sin />», , (a0xk + • • • + ak) cos px, + (b0xk + ••• +bk) sin px, e"x(a^Kk + • • • + ak) cos px, + e«x(b0xk H, + bk) sin px, , e^Oo»* H, 1- /k) cos pa, x, k, + e^ (g^x¡ H, h srfc) sin px, Sums oí any or some of the above entries., , h ak), , Sums of the corresponding trial solutions., , The above method holds in case no term in the assumed trial solution appears in, the complementary solution. If any term of the assumed trial solution does appear, in the complementary solution, we must multiply this trial solution by the smallest, positive integral power of x which is large enough so that none of the terms which, are then present appear in the complementary solution., 2. Method of Variation of Parameters., Let the complementary solution of $(D)y = R(x) be, y = ciyi(x) + c2y2(x) + • • • + cnyn(x), Replace the arbitrary constants Ci, c2, ..., en by functions Ki(x), K^x),..., Kn(x), and seek to determine these so that y = K\y\ + Kzyz + • • • + Knyn is a solution of, <t>(D)y = R(x). Since, to determine these n functions, we must impose n restrictions, on them and since one of these is that the differential equation be satisfied, it follows, that the remaining n -1 may be taken at will. The conditions which lead to the, greatest simplicity are given by the equations, K'^ + K'^+.-.+K'^ = 0, , %+ %+•••+*X = o, tf><"-2) + K'2y?-» + • • • + £><,»-» = 0, #><«-» + #><»-» + • • • + Kfâ-» = R(x)/a0, , (11), , where the last equation represents the condition that the given differential equation be satisfied., Since the determinant of the above system of equations is the Wronskian of, 2/1,2/2, ..., 2/n which is supposed to be different from zero, the equations can be solved, for K(,Kz, ...,K'n. From these Ki,Kz, .. .,Kn can be found by integration leading, to the required solution. The method is applicable whenever the complementary, solution can be found, including cases where oo,..., a» are not constants.
Page 85 : CHAP. 3], , LINEAR DIFFERENTIAL EQUATIONS, , 75, , OPERATOR TECHNIQUES, When Oo, ai, ..., an are constants, the equation <j>(D)y = R(x) can be written in factored, formas, (12), ao(D - mi)(D - m2) • • • (D — mn)y = R(x), where mi, ...,mn are constants and where the order of the factors (D - mi), ..., (D — mn), is immaterial. This is not true if 00,0,1,.. .,an are not constants [see Problem 3.2]. The, constants mi, ...,mn are the same as the roots of the auxiliary equation (6) or (7), and so, the complementary solution can be written as before. To obtain particular solutions, the, following operator methods will be found useful., 1. Method of Reduction of Order., Let ao(D-m2) • • • (D-mJy = Yi. Then (12) becomes (D-mi)Yi = R(x) which, can be solved for YI. Then let a<>(D -ms) ••• (D- mn)y = Yz so that (D - m2)Yz = Yi, which can be solved for Y2. By continuing in this manner, y can be obtained. This, method yields the general solution if all arbitrary constants are kept, while if arbitrary, constants are omitted it yields a particular solution., 2. Method of Inverse Operators., Let -Tn\, R(x) be defined as a particular solution yp such that <¡>(D)yp = R(x). We, <p\uan, >, call 1/£(D), inverse operator. By reference to the entries in the following table, the, labor involved in finding particular solutions of <¡>(D)y = R(x) is often diminished considerably. In using these we often employ the theorem that, , which is simply a statement of the fact that 1/^>(D) is a linear operator., Table of Inverse Operator Techniques, , A, , B, , This can also be evaluated by expanding, the inverse operator into partial fractions, and then using entry A., , C, , D
Page 86 :
[CHAP. 3, , LINEAR DIFFERENTIAL EQUATIONS, , 76, , Table of Inverse Operator Techniques (cont.), , E., , "Re" means "real part of" and "Im" means, "imaginary part of"., , if $(ip) ^ o, otherwise use entry C., , F., xv, , = (c0 + CID +, , h ckDk + • • -)XP, , by expanding 1/*<D) in powers of D., (p = positive integer), , + cpDi>)xI>, , (c0 + CjZH, sinee DP xP, , ^, , = 0 for n > 0., , G., called the "operator shift theorem"., H., , LINEAR EQUATIONS WITH VARIABLE COEFFICIENTS, Various methods are available for solving differential equations of the form (1) where, «to, 0,1,..., On are not constants. In the following we list a few important methods., I., , MISCELLANEOUS TRANSFORMATIONS OF VARIABLES, 1. Cauchy or Euler Equation. This equation has the form, (b0xnDn + bixn-lDn~l + • • • + bn-!xD + bn)y = R(x), where 60, & i , . . . , bn are constants. It can be solved by letting x — e* and using, the results, xD = Dt, x2D2 = Dt(Dt-l),, x3D* = Dt(Dt-l)(Dt-2), ..., where Dt = d/dt, thus reducing the equation to one with constant coefficients., The case where R(x) = 0 can be solved by letting y = x" and determining the, constant p., 2. Case where one solution is known. If one solution y = Y(x) of 4>(D)y - R(x), is known, then the substitution y = vY(x) transforms the differential equation, into one of order n -1 in v'. If n - 2, the equation can then be solved exactly., See Problem 3.34., 3. Reduction to canonical form. The general second order linear equation, y" + p(x)y' + q(x)y = r(x), , (IS), , can be transformed into the canonical form, v" + f(x)v, , = g(x), , (14)
Page 87 :
CHAP. 3], , LINEAR DIFFERENTIAL EQUATIONS, , 77, , where, , (15), , by using the substitution, , (16), , Thus if (14) can be solved, so can (13). See Problems 3.35 and 3.36., II., , EXACT EQUATIONS, The equation [ao(x)Dn + ai(x)Dn-1 + ••• + an(x)]y = R(x) is called exact if, a0(x)D» + ai(iK)Z)»-1 + • • • + an(x), , = D[p0(x)Dn-1 + ••• + pn-i(x)}, , For example, [oo(x)D2 + ai(x)D + a2(x)]y = R(x) is exact if and only if a¿' + aí +, a2 = 0 identically. See Problems 3.37 and 3.38., III. VARIATION OF PARAMETERS, This method, identical to that on page 74, can be used when the complementary, solution is known., IV. OPERATOR FACTORIZATION, If <f>(D) can be factored into factors each having the form p(x)D + q(x), then the, method of reduction of order (page 75) can be used. See Problem 3.39., V., , SERIES METHODS, The equation at>(x)y" + ai(x)y' + a2(x)y = 0 where a<>, oi, o2 are polynomials, can, often be solved by assuming that, y = xe(c0 + dx + C2X2+•••), , =, , CkX"**, , where ck = 0, k < 0, , (17), , where fi and Ck are constants. Substituting (17) into the differential equation leads, to an equation for /?, called the indicial equation, and equations for the constants, Co, ci, ... in the form of a recursion formula. By solving for p and the other constants,, a series solution can often be obtained. The series is called a Frobenius series and the, method is often called the method of Frobenius. See Problem 3.40., , SIMULTANEOUS DIFFERENTIAL EQUATIONS, A system of differential equations with two or more dependent variables and one, independent variable can be solved by eliminating all but one of the dependent variables,, thus obtaining a single ordinary differential equation. Solutions obtained should be checked, by substitution into the original differential equations to insure that the proper number of, arbitrary constants are present. See Problem 3.41., , APPLICATIONS, Problems in mechanics, electricity and other fields of science and engineering often lead, to linear differential equations and can be solved by the above methods. See Problems, 3.42-3.47.
Page 89 :
CHAP. 3], , LINEAR DIFFERENTIAL EQUATIONS, , 79, , LINEAR DEPENDENCE AND WRONSKIANS, , 3.6., , Show that the functions cos 2x, sin2 x, cos2 x are linearly dependent., We must show that there are constants clf e2, cs not all zero such that e1 cos 2x + c2 sin2 x +, cs cos2 * = 0 identically. Since cos 2w = cos2 x — sin2 x, we can choose ct = 1, c2 = 1, c3 = —1, and the required result follows., , 3.7., , Prove that if the Wronskian of the set of functions y\,... yn is different from zero, on an interval, then the functions are linearly independent on the interval., Suppose the contrary, i.e. that the functions are linearly dependent on the interval. Then there, are n constants c l 5 ..., cn, not all zero, such that, Ci2/i + • • • + enyn = O, , (Í), , identically. By successive differentiation we then have identically, Ciy'i, , + • • • + cny'n, , = 0, , ClJ/ (n-i), , + ... + cny<»-» = 0, , Now in order for the system of n equations (Í) and (ê) to have solutions for c1( .. .,cn which are, not all zero, we must have, , Since W ¥° 0 by hypothesis, the contradiction shows that the functions cannot be linearly, dependent and so must be linearly independent., , 3.8., , Prove that if the functions yi, ..., yn are linearly independent on an interval, then, the Wronskian is different from zero., Suppose the contrary, i.e. the Wronskian is zero for a particular value x0 of the interval., Consider the system of equations, , Since the Wronskian is zero we see that this system has a solution c,, ..., cn not all zero. Now let, V = ctf^x) + • • • + cnyn(x), , (2), , If J/i(*), • • ->yn(x) are solutions of <t>(D)y = 0, it follows from Problem 4 that (2) is also a solution, of <f>(D)y — 0 which as is seen from equations (1) satisfies the conditions y(x0) — 0, y'(x0) = 0, ...,, j/Cn-i)(a;0) = 0. But y = 0 satisfies <f>(D)y = 0 and these same conditions. Thus by the uniqueness, theorem of page 41 it follows that the only solution is y = 0, i.e., Ci3/i(*) + • • • + cnyn(x) = 0, from which we see that j/1( ..., yn are linearly dependent. Since ylt ..., yn are linearly independent by hypothesis, the contradiction shows that the Wronskian cannot be zero for ic0 and the, required result is proved., This problem and Problem 3.7 together provide a proof of Theorem 3-2.
Page 90 :
80, , 3.9., , LINEAR DIFFERENTIAL EQUATIONS, , [CHAP. 3, , Let y\(x), Vi(x) be solutions of y" + p(x)y' + q(x)y = 0., (a) Prove that the Wronskian is W = yiyí — y¿y[ = ce~Svax., (b) Under what conditions on c will y\ and 2/2 be linearly independent., (a) Since yt and yz are solutions,, y" + py{ + qy! = 0,, Multiplying these equations by j/2, , y'z + py'z + qyz = 0, , an, , d 2/i respectively and subtracting,, , yiv'z -y^y" + p(2/ií/2-2/22/í) - o, , Then using W = y^a'z — VzV\, , and, , (i), , noting that î/iî/2'— 2/zî/i' — dWIdx, (1) becomes, , + pW = 0, which has the solution, , W = ViVz —V&i = ce~Spdx, , (2), , This result is sometimes called Abel's identity,, (b) Since e-/»1»1 is never zero, W = 0 if and only if c = 0 and W ¥* 0 if and only if e ^ 0., Thus if c ¥= 0, j/! and j/2 are linear independent., , 3.10. Use Problem 3.9 to prove that if y\ is a known solution of y" + p(x)y' + q(x)y = 0,, then a linearly independent solution is, so that the general solution is y = c\y\ + ctyz., From (2) of Problem 3.9 we have on dividing by yz. *£ 0,, , Then integrating and solving for yz leads to the required linearly independent solution., , THE REDUCED OR HOMOGENEOUS EQUATION, 3.11. (a) Find three linearly independent solutions of (D3 - 9D)y = 0 and (b) write the, general solution., (a) Assume that y = emx is a solution where m is a constant. Then (Z>3 — 9D)emx = (w3 — 9m)emx, is zero when ms — 9m = 0, i.e. m(m — 3)(m + 3) = 0 or m = 0,3, —3., Then e°x = 1, esx, e~3x are solutions. The Wronskian is, e-ax, 1, esx, _, 3e«* -3e~Sx, 31, 0 3e -Se-te, = 54, 9e8i 9e-3i, 3, g, -3x, e, 0 9e *, so that the functions are linearly independent by Theorem 3-2., (6) The general solution is y - e^ + e2e3* + c3e~Sx = Cj + c2e3x + cae~Sx., , 3.12. Solve (a) 2y" - 5y' + 2y = 0. (6) (2DS - D2 - 5D - 2)y = 0., (a) The auxiliary equation is 2m2-5«t 2+ 2 = 2x, 0 or, Then the general solution is y = Cje*/ + C2e ., , (2m — l)(w-2) = 0 so that m = 1/2,2., , (6) The auxiliary equation is 2m3 - m2 - 5m - 2 = 0 or (2m + l)(m + l)(m - 2) = 0 so that, m — —1/2, —1,2. Then the general solution is y = e^-*'2 + c2e~* + ese^.
Page 91 :
CHAP. 3], , 3.13., , LINEAR DIFFERENTIAL EQUATIONS, , 81, , Solve y" + 9y = 0 or (D* + V)y = 0., The auxiliary equation is m2 + 9 = 0 and m = ±Si. Then the general solution is, = Ae3to + Be~3ix = A(cos Sx + i sin Sx) + B(cos Sx — i sin Sx) which can be written y =, Cj cos Sx + c2 sin Sx. Since cos Sx, sin 3* are linearly independent, this is the general solution., , y, , 3.14. Solve (D2 + 6D + 25)y = 0., The auxiliary equation is m2 + 6m + 25 = 0 and so m —, Then the general solution is, y = Ae<- 3 + 4i >* + Be<-8"*1)a: =, , e~Sx(Ae*ix + Be-***), , = e~íx(eí cos 4a; + c2 sin 4»), , 3.15. Solve (Z>4-16)|/ = 0., The auxiliary equation m4 -16 = 0 or (m2 + 4)(m2 - 4) = 0 has roots ±2i, ±2., general solution is y = el cos 2x + c2 sin Zx + cse2x + c4e~2x., , Then the, , 3.16. Solve y" - 8y' + 16y = 0., The equation can be written (D2 — 8D + 16)y = 0 or (D — 4)2y = 0. The auxiliary equation, (m — I)2 = 0 has roots m = 4,4 so that e4*, e4x are solutions corresponding to these repeated roots., However, they are clearly not linearly independent and so we cannot say that y = Cje4x + C2e4j: is, the general solution, since it can be written (cj + c2)e41 = ce41 which involves only one arbitrary, constant. There are two methods which can be used to get the general solution., Method 1., Write the given equation as (D - 4)(I> - 4)j/ = 0 and let (D — 4)y = Yl so that (D - a)Y1 - 0, and so Yl = A^. Thus (D — 4)y = Aje4*. Solving, we find y = (et + ctffie*1., This method, called the method of reduction of order, illustrates a general procedure for, repeated roots. Using it we would find, for example, that if 4 were a triple repeated root, the, general solution would be y — (Cj + e2x + C3x2)e*x., Method 2., Since one solution yt = e*x is known, we can use Problem 3.10 to find the linearly independent, solution, 2/2, , =, , Then the general solution is y — eje41 + c2xe*x = (cj + czx)e*x., , 3.17. Solve (D + 2)S(D - 3)4(D2 + 2D + 5)y = 0., The auxiliary equation (m + 2)a(m - 3)4(w2 + 2m + 5) = 0, —1 ± 2i. Then the general solution is, , has roots -2, -2, -2, 3, 3, 3, 3,, , y = (cj + C2x + caxz)e-2x + (c4 + csx + cex2 + e?xs)e3x + e~x(cg cos 2x + c9 sin 2x), , THE COMPLETE OR NONHOMOGENEOUS EQUATION., UNDETERMINED COEFFICIENTS, 3.18. Solve (D2 + 2D + 4)y = 8xz + l2e~x., The complementary solution is e~x(cl cos ^x + e 2 smv / 3^)., To obtain a particular solution assume corresponding to 8*2 and 12e~x the trial solutions, ax + bx + c and de~x respectively, since none of these terms are present in the complementary, solution. Then substituting y = ax2 + bx + c + de~x in the given equation, we find, 2, , ¿lax2 + (4a + 46)* + (2a + 26 + 4c) + Sde'* -, , 8*2 + I2e~x
Page 92 :
82, , LINEAR DIFFERENTIAL EQUATIONS, , [CHAP. 3, , Equating corresponding coefficients on both sides of the equation,, 4a = 8, 4a + 46 = 0, 2a + 26 + 4c = 0, 3d = 12, Then a = 2, 6 = —2, c = 0, d = 4 and the particular solution is 2x2 — 2x + 4e~x., required general solution is, , Thus the, , 3.19. Solve Problem 3.18 if the term 10 sin Bx is added to the right side., Corresponding to the additional term 10 sin Sx we assume the additional trial solution, h cos Sx + k sin 3» which does not appear in the complementary solution. Substituting this into, the equation (D2 + 2D + i)y = 10 sin Sx,, (6fc - 5h) cos Sx - (5k + 6h) sin Sx - 10 sin 3¡», Equating coefficients,, or, , 6k - 5h = 0, 5k + 6h - -10, , ., , 3.20. Solve, , Then the required general solution is, , (D2 + 4)y = 8sin2«., , The complementary solution is c1 cos 2x + c2 sin 2x., For a particular solution we would normally assume a trial solution a cos 2x + b sin 2x. However, since the ternis appear in the complementary solution, we multiply by x to obtain the trial, solution x(a cos 2x + 6 sin 2x). Then substituting in the given equation,, —4a cos 2x — 46 sin 2x — 8 sin 2x, so that 4o = 0, —46 = 8 and o = 0, 6 = —2. Then the required general solution is, y = c1 cos 2x + c2 sin 2x — 2x sin 2x, , 3.21., , Solve, , (D5 - 3D4 + 3D3 - D2)y = x2 + 2x + Se*., , The auxiliary equation is m5 — Sm4 + 3m3 — -m2 = 0 or, 0, 0, 1, 1, 1 and the complementary solution is, Cj + c2x + (c3 + c4x + C5x2)ex, , m2(m — I)3 = 0., , Thus, , m—, , Corresponding to the polynomial x2 + 2x, we would normally assume a trial solution ax2 + bx + c., However, some of these terms appear in the complementary solution. Multiplying by x, the trial, solution would be x(ax2 + bx + c) = ax3 + bx2 + ex, but one of these terms is still in the complementary solution. Finally multiplying by x again to obtain ax4 + bx3 + ex2, we see that this, has no term in the complementary solution and so is the needed trial solution., Similarly, corresponding to Sex we would normally assume a trial solution dex. But since, this term as well as dxex and dx2ex are in the complementary solution, we must use as trial solution, dx3ex., Thus the assumed trial solution is ax* + bx3 + ex2 + dx3ex. Substituting this in the given, equation, we find, -12aa;2 + (72a - 6b)x + (186 - 72a - 2c) + 6dex = x2 + 2x + Sex, from which, and the general solution is, , THE COMPLETE OR NONHOMOGENEOUS EQUATION., VARIATION OF PARAMETERS, 3.22. Solve y" + y = sec x or (D2 + l)y = sec x., The complementary solution is Cj cos x + c2 sin *. Then we assume the general solution to be
Page 93 :
CHAP. 3], , LINEAR DIFFERENTIAL EQUATIONS, , 83, , y = Kl cos x + K2 sin x where Klt K2 are suitable functions of x to be determined. Differentiation, yields, y' = —K-1 sin x + K2 cos x + Kl cos x + K2 sin x, (1), Since there are two functions KltK2, we must arrive at two conditions for determining them., However, one of these conditions is that the differential equation must be satisfied. Thus we are at, liberty to impose arbitrarily the second condition. We choose this condition to be the one which, simplifies (1) most, namely, K( cos x + K2 sin x = 0, (2), Then (1) becomes, Differentiating again, , y', , =, , —K1 sin x + K2 cos x, , = —Kl cos x — K2 sin x — K1 sin x + K2 cos x, , y", , Thus, , y" + y, , From the two equations, , = —K[ sin x + K2 cos *, , K[ cos x + K2 sin x, —K 1 sin x + K2 cos x, , we find K[ = 1, K2 = —tan x., required general solution is, , y, , 3.23. Solve, , =, , sec a;, , (3), , = 0, = sec x, , Thus by integrating, K1 = x + c¡, K2 = —In sec x + c2 and the, , = GI sin x + c2 cos x + x sin x — cos x In sec x, , (D3 + 4D)y = 4 cot 2x., , The complementary solution is ci + c2 cos 2x + c3 sin 2x. Then we are led to the following, equations for determining the functions Klt K2, Ks in the general solution, , y, , =, , KI + K2 cos 2x + K3 sin 2x, , KI + K2 cos 2x + K's sin 2x = 0, 0 - 2K'2 sin 2x + 2K'Z cos 2x = 0, 0 - 4K2 cos 2x - 4K'3 sin 2x = 4 cot 2œ, Solving this system, we find K( = cot 2x, K2 — —cos2 2x/sin 2a;, K's — —cos 2x so that, From this we find the general solution, , OPERATOR TECHNIQUES, 3.24. Evaluate, Method 1., Let, , Then by definition (D — 2)y — e*x or, , — 2y = e4x. Solving this as a, , first order linear equation with integrating factor e~2x, we obtain y = ^e4x + ce2*. Since we are, interested only in particular solutions,, , Method 2 [using formula A, page 75, derived as in method 1].
Page 95 :
CHAP. 3], , 3.28., , LINEAR DIFFERENTIAL EQUATIONS, , 85, , Evaluate, We can show by a method similar to that in Problem 3.27 that D2 can formally be replaced, by -22 = -4. Thus, , 3.29., , Prove that, , (a) Use mathematical induction. The result is true for n = 1, since D[eJ>xR(x)] = e^xDR(x) +, pei>xR(x) = eP*(D + p)R(x). Assume the result true for n = k. Then Dk[et>*R(x)], Differentiating both sides,, epx(£) + p)kR(x)., , Thus if the result is true for n = k, it is true for n = fc + 1; but since it is true for n = I,, it must be true for n = 2, ... and thus all n., (6), , (c) Let, , = y, , so that <f>(D)y = eJ>xR(x)., , Then, , or, , from which, , 3.30., , Solve, , (Ds + D)y = e-*x cos2x., , Using Problem 3.29,, , Since the complementary solution is ct + c2 cos x + c3 sin x, the required general solution is, , y, , = c1 + cz cos x + c3 sin x +, , (9 sin 2x + 1 cos 2x)
Page 96 :
86, , LINEAR DIFFERENTIAL EQUATIONS, , [CHAP. 3, , 3.31. Evaluate, By formal long division in ascending powers of D,, , CAUCHY OR EULER EQUATION, 3.32. Let D - d/dx and Dt = d/dt. Prove that if x = e*, then, (a), Then xDy = e*Dy = Dty or xD = Dt., (6), , Then x2D*y = e*D*y = (Dt - Dt)y or xW = D{ -Dt = Dt(Dt - 1)., , 3.33. Solve, By the transformation, , x = e* the equation becomes, , Then the general solution is, Another method., Letting y = xv in the complementary equation, , we find, , Thus x2 and x~2 are solutions and the complementary solution is, We can now use the method of variation of parameters to find the required general solution., , CASE WHERE ONE SOLUTION IS KNOWN, 3.34. Solve, given that y — x is a solution., Let y = xv. Then y' — xv' + v, y" — xv" + 2v' and the given equation becomes, , Integration yields, , i.e., or, Then, and so the general solution is, , We can also use the result of Problem 3.10 to solve the equation.
Page 97 :
CHAP. 3], , LINEAR DIFFERENTIAL EQUATIONS, , 87, , REDUCTION TO CANONICAL FORM, 3.35. By letting y = uv and choosing u appropriately, obtain a differential equation corresponding to y" + p(x)y' + q(x}y — r(x) with the term involving the first derivative, removed., Substituting y = uv in the given equation, we find, uv" + (2ur + pu)v' + (u" + pu' + qu)v = r, Let 2u' + pu = 0 so that u = e~$ W>t* ., , (1), , Then (1) becomes, (2), , as required. Equation (2) is called the canonical form of the given equation., , 3.36. Solve, , 4x2y" + 4xy' + (x2 - l)y = 0., , Comparing with Problem 3.35, p = l/x, q = (x2 — l)/4a;2, r = 0., V", , + %v = 0, , Then the canonical form is, , or, , Thus, , EXACT EQUATIONS, 3.37. Show that [a0(x)D2 + ai(x)D + a2(x)]y = R(x) is exact if and only if a'd - a( + a2 = 0, identically., By definition of an exact equation, there are functions PQ(X), PI(X) such that, [ooD2 + aiD + a¿y = D[p0D + Pl]y = [p0Z>2 + (p£ + Pl)D + p(]y, Then we must have, , a0 = pv,, , a± — p'u + plt, , a2 — p{, , Eliminating p0, p1 from these 3 equations, we find a'¿ — a¡ + ct2 = 0., Conversely if «£,' — a{ + az = 0, then, [ct0I>2 + aj) + a2]y = [a0D2 + aj) + a{ - a'0']y = D[a0D + at - a'9}y, and the equation is exact., , 3.38. Solve, , (1 - x2)y" - Zxy' - y = 1., , Comparing with Problem 3.37, a0 = 1 — x2, a l = —3*, a2 = —1 and a'0' — a{ + a2 = 0 so that, the equation is exact and can be written, , D[(l - xz)D - x]y = 1, Integrating this equation and solving the resulting first order linear differential equation, , we find that, , OPERATOR FACTORIZATION, 3.39. (a) Show that the operator xD2 + (2x + 3)Z) + 4 = (D + 2)(xD + 2)., (6) Use (a) to solve xy" + (2x + S)y' + 4y = e2x., (a) (D + 2)(xD + 2)y = (D + 2)(xDy + 2y) = D(xDy + 2y) + 2(xDy + 2y), = xD*y + Dy+ 2Dy + 2xDy + 4y = [xD* + (2x + 3)D 4- 4]j/
Page 98 :
88, , LINEAR DIFFERENTIAL EQUATIONS, (6) The given equation can be written, Now solving, and, , [CHAP. 3, , we find, , Then, , SERIES METHODS, , 3.40. Solve Problem 3.36 by using the method of Frobenius., where we define, , omitting the summation limits., , Thus, , In order to write the series on the right in terms of coefficients of xk+P, we must replace the index, of summation k in the third series by fc — 2 [note that this does not affect the limits — «o and °° of, the index of summation!. Then the series on the right can be written, Since this must be zero, each coefficient must be zero so that, {4(fc + /3)2-l}c fc + c fc _ 2 = 0, , ', , (1), , 2, , Let fc = 0. Then since c_ 2 = 0, (1) becomes (4/8 — I)c0 = 0 which is called the indicial equation., Assuming c0 ¥= 0, this leads to 4/Î2 — 1 = 0, ¿8 = ±1/2. We have two cases 0 = 1/2, —1/2 and, we shall consider the smaller value —1/2 first., Case 1, p = —1/2., , In this case (i) becomes, , Putting k = 1,2,3,4, ... in (2), we find, , and from these it is clear that c0 and Cj are undetermined while cz, es, c 4 ,... are found in terms of, c0 and c1 as follows:, , Then the corresponding solution is
Page 99 :
LINEAR DIFFERENTIAL EQUATIONS, , CHAP. 3], , 89, , which is equivalent to the solution obtained in Problem 3.36., Note that for this equation it is not necessary to consider Case 2, p = 1/2 since we have, already obtained the required solution. Note also that if we had considered ft — 1/2 first, we, would not have obtained the general solution. This is typical in general when the roots of the, indicial equation differ by any integer except zero. In other equations both cases would have to, be considered, each leading to a series solution. The general solution would then be obtained, from these series on multiplying each by an arbitrary constant and adding., , SIMULTANEOUS EQUATIONS, 3.41. Solve the system of equations, , Write the system as, (.D2 + 3)« + Dy = e~t,, , -Wx + (D2 + B)y = sin2t, , (1), , Method 1, using determinants., , Formal application of Cramer's rule for solving linear equations can be used and we find, , where we have expanded the determinants so that operators precede functions., Note that the results are equivalent to, (Z>2 + i)(D2 + g)x = 4e-«-2cos2t,, (D2 + 1)(Z>2 + 9)j/ = -sin2i-4e-', Solving these equations we find, , We can show that the total number of arbitrary constants in the solution is the same as the degree, of the polynomial in D obtained from the determinant, #2 + 3, D, -W, Z>2 + 3, , = D* + 1002 + 9, , i.e. 4. Thus there is a relationship between the constants Cj, e2, cs, c4 and cs, ce, c7, cg. To determine, this relationship we must substitute the values of x and y obtained above in the original equations., If we do, we find, c5 = 2c2, ce = -2c1( c7 = -2c4, c8 = 2c8, Thus the required solution is, , Method 2., , We can also eliminate one of the variables, for example x, by operating on the first of equations, (Í) with 4Z>, the second with D2 + 3 and adding. To solve the resulting equations, we can then use, the method of undetermined coefficients.
Page 100 :
90, , [CHAP. 3, , LINEAR DIFFERENTIAL EQUATIONS, , APPLICATIONS, 3.42. A particle P of mass 2 (gm) moves on the x axis attracted toward origin O with a, force numerically equal to Sx. If it is initially at rest at x = 10 (cm), find its position at any later time assuming (a) no other forces act, (&) a damping force numerically equal to 8 times the instantaneous velocity acts., (a) Choose the positive direction to the right [Fig. 3-1]. When, x > 0, the net force is to the left [i.e. is negative] and so, is —8*. When x < 0, the net force is to the right [i.e. is, positive] and so is also —8». Thus by Newton's law,, and, , Fig. 3-1, , x = «! cos 2i + c2 sin 2t, , Since x - 10, dx/dt = 0 at í = 0, we find x = 10 cos 2f., The graph of the motion is shown in Fig. 3-2. The amplitude [maximum displacement from, 0] is 10 (cm). The period [time for a complete cycle] is ir (sec). The frequency [number of cycles, per second] is 1/n- (cycles per second). The motion is often called simple harmonic motion., , Fig. 3-2, , Fig. 3-3, , (6) The damping force is given by —8 dx/dt, regardless of where the particle is. Thus for example, if x < 0 and dx/dt > 0, then the particle is to the left of O and moving to the right so the, damping force must be to the left, i.e. negative. Thus by Newton's law,, or, 2, , the general solution of which is x — e~ t(ct + ezt). Since x = 10, dx/dt = 0 when t = 0,, ct = 10, c2 = 20 so that x = 10e~2t(l + 2i). The motion is non-oscillatory. The particle, approaches 0 but never reaches it [see Fig. 3-3]., , 3.43. A 20 lb weight suspended from the end of a vertical spring stretches it 6 inches., Assuming no external forces, find the position of the weight at any time if initially, the weight is (a) pulled down 2 inches and released, (6) pulled down 3 inches and, given an initial velocity of 2 ft/sec downward. Find the period and amplitude in, each case., Let A and B [Fig. 3-4] represent the position of the end of the, spring before and after the weight W is put on. B is called the, equilibrium position. Call y the displacement of W at any position C, from the equilibrium position. Assume that y is positive in the downward direction., By Hooke's law, 20 lb stretches the spring .5 ft, 40 lb stretches it, 1 ft and so 40(.5 + y) lb stretches it (.5 + y) ft. Thus when W is at C,, the tension in the spring is 40(.5 + y) lb., By Newton's law,, Mass • Acceleration, or, , =, =, =, , Net force on W, Weight downward — Tension upward, 20, , 40(.5 + y), , Fig. 3-4
Page 101 :
CHAP. 3], , LINEAR DIFFERENTIAL EQUATIONS, which becomes, , (a) Since, amplitude is, , + 64j/ = 0, , or, , V — GI cos 8t + C2 sin 8i, , at t - 0, we have, ft and the period is 2W8 = W4 sec., , (6) Since, , 91, , , c2 = 0 and so y - £ cos 8i. The, and, , Then the amplitude is, , ft approx. and the period is W4 sec., , 3.44. Solve Problem 3.43 taking into account an external damping force given in pounds by, PV where v is the instantaneous velocity in ft/sec and (a) /3 = 8, (b) ft = 10, (c) /3 = 12.5., The equation of motion with damping force fiv = /? dy/dt, , (a) If /3 = 8,, , is, , Solving subject to the conditions y = 1/6,, , dy/dt = 0, , at í = 0, we find, , V, , =, , The motion is damped oscillatory with period 2^/4.8 = 5?r/12 sec., (6) If, , ¿S = 10,, , Solving subject to the conditions as in (a), y =, , i. The motion is called critically damped since any smaller value of /3 would, produce oscillatory motion., ( c ) I f j8 = 1 2, overdamped., , ., , 5, , ,, , a, , n, , d we, , findThe, , motion i s called, , 3.45. (a) Work Problem 3.43 (a) if an external force given by F(t) = 40 cos 8i is applied, for t > 0 and (b) give a physical interpretation of what happens as í increases., (a) The equation of motion in this case becomes, , or, , The solution of this, subject to y = 1/6, dy/dt = 0 at t = 0, is, y — % cos 8i + 4t sin 8*, (b) As t increases, the term 4i sin 8t increases numerically without bound and physically the, spring will ultimately break. This illustrates the phenomenon of resonance and shows what, can happen when the frequency of the applied force is equal to the natural frequency of the, system., , 3.46. A rod AOB [Fig. 3-5 below] rotates in a vertical plane about a point O on it with, constant angular velocity «. A particle P of mass m is constrained to move along, the rod. Assuming no frictional forces, find (a) a differential equation of motion of, P, (b) the position of P at any time and (c) the condition under which P describes, simple harmonic motion., (a) Let r be the distance of P from O at time t and suppose that the rod is horizontal at t - 0., We have
Page 102 :
LINEAR DIFFERENTIAL EQUATIONS, , 92, , [CHAP. 3, , Vertical, , Fig. 3-5, , Net force on P, , =, , Centrifugal force + Component force due to gravity, , or, , where, , (J), , r — r0, dr/dt = v0, , at, , t =0, , (2), , i.e. r0 and v0 are the initial displacement and velocity of P., (b) Solving (1) subject to (2), we find, , (c) Simple harmonic motion along the rod results if and only if r0 = 0 and v0 — g/2o., , 3.47. An inductor of 2 henries, resistor of 16 ohms and capacitor of .02 farads are connected in series with a battery of e.m.f. E = 100 sin3í. At t = 0 the charge on the, capacitor and current in the circuit are zero. Find the (a) charge and (6) current, at £ > 0., Letting Q and / be the instantaneous charge and current at time t, we find by Kirchhoff's laws, , or since / = dQ/dt,, , Solving this subject to Q - 0, dQ/dt - 0 at t = 0, we find, (a), (6), , The first term is the steady-state current and the second, which becomes negligible as time, increases, is called the transient current.
Page 103 :
CHAP. 3], , LINEAR DIFFERENTIAL EQUATIONS, , 93, , 3.48. Given the electric network of Fig. 3-6. Find, the currents in the various branches if the, initial currents are zero., Kirchhoff's second law states that the algebraic, sum of the voltage drops around &ny closed loop is, zero. Let us traverse loops KLMNK and JKNPJ in, a counterclockwise direction as shown. In traversing, these loops we consider voltage drops as positive, when we go against the current. A voltage rise is, considered as the negative of a voltage drop., Fig. 3-6, Let 7 be the current in NPJKN. This current divides at junction point K into /i and /2 so that, / = Ii + 72. This is equivalent to Kirchhoff's first law., Applying Kirchhoff's second law to loops JKNPJ and KLMNK respectively, we have, dl,, 201 - 120 + 2^ + 10/n, , = O, , d/j, d/2, -lO/i - 2^ + 4^ + 20/2 = 0, , Putting / = /! + /2 and using the operator D = d/dt, these become, (D + 15)/! + 10/2 = 60, -(D + 5)/! + (2D + 10)/2 = 0, , (Í), («), , (S), (4), , Solving these subject to I1 — I2 = 0 at t = 0, we find, /! = 3(1-«-*>«),, , /2 = f(l-e-2«i),, , / = |(i_ e -20t), , Supplementary Problems, OPERATORS, 3.49. Write each of the following in operator form: (a) y" + 4j/' + 5j/ = e~x, (b) 2y'" - x — 2y' + y,, (c) xy' + 2y = 1., 3.50., , (a) Evaluate (Z>3 + 1) sin 2x and (D + l)(D2 - D + 1) sin 2x. (b) Are the operators I>3 + 1 and, (D + 1)(I>2 - D + 1) equivalent? Explain., , 3.51., , (a) Evaluate (xD - S)(D + 2){x2 + x - 4} and (D + 2)(xD - B){x2 + x - 4}. (6) Are the operators, (xD - S)(D + 2) and (D + 2)(xD - 3) equivalent? Explain., , 3.52., , (a) Evaluate (xD)(xD)(xD){x2 + 2ex} which can be written (xD)3{x* + 2ex}. (b) Is this the same as, xsDs{x2 + 2ex}"! (c) Are the operators (xD)t and x3D* equivalent? Explain., , 3.53., , Under what conditions will the operators (D + a)(D + b) and (D + b)(D + a) be equivalent? Explain., , LINEAR DEPENDENCE AND WRONSKIANS, 3.54. (a) Show that the functions x2, Bx + 2, x — \, 2x + 5 are linearly dependent. (6) Are the functions, x2, 3x + 2, x — i linearly dependent?, 3.55., , Investigate the linear dependence of ex, xex, x2ex., , 3.56., , Show that y = x is a solution of xy" + xy' — y = 0 and find the general solution., , 3.57., , If î/t is a solution of y" + p(x)y' + q(x)y = r(x), explain how to find the general solution. Illustrate, by an example.
Page 104 :
LINEAR DIFFERENTIAL EQUATIONS, , 94, , [CHAP. 3, , THE REDUCED OR HOMOGENEOUS SOLUTION, 3.58., (a) Find three linearly independent solutions of (D + 2)(D - 1)(D - S)y = 0 and (6) write the, general solution., 3.59., , Solve, , (a) y" + 8y' + I2y = 0, (b) (D* - 4Z> - l)y = 0., , 3.60., , Solve, , (a) (D* + 25)y = 0; y(Q) = 2, y'(0) = -5,, , 3.61., , Solve, , (a) y" + 4y' + % = 0, (6) ley" - 8y' + y = O, (c) (D + 6)4(1) - 3)2j, = 0, , 3.62., , Solve D*(D + 1)2(D2 + W + 5)2(Z>2 + 4)j/ = 0., , (6) y" - 8y' + 20y = 0, (c) (D» + 8)y - 0., , (d) (D» + D«)y = 0, (e) (£>» + 64)2j/ = 0, , THE COMPLETE OR NONHOMOGENEOUS EQUATION, 3.63., Solve (a) y" - 5y' + 6y = 50 sin 4w, (6) (Z>3 - 8)y = 16» + 18e~x + 64 cos 2x - 32., 3.64., , Solve, , (a) y" + By' + 2y = 4e~2*, (6) (D» + 3£>2)j/ = 180ws + 24a., , 3.65., , Solve, , (I>6 - 2I>5 + D*)y = 12Qx + 8e*., , 3.66., , Solve using variation of parameters, (a) y" + 2y' — 3y = xe~x, (b) y" + 4y = esc 2x, , (c) xy" — y' = x, (d) (D3 + D)y = 4 tan x, , OPERATOR TECHNIQUES, 3.67., Evaluate each of the following:, , 3.68., , Solve using operator techniques: (a) (D2 + W + 4)y = I8ex - 8 sin 2x, (b) (D + 1)(D - 3)y = esx + x2,, (c) (D + 2)2(D - 2)y = «-2* cos x., , 3.69., , Prove entries (a) A, (b) B, (c) E, (d) H in the table on pages 75 and 76., , CAUCHY OR EULER EQUATION, 3.70., Solve (a) x2y" + xy' - X2j/ = 0, (6) x2y" - 2xy' + 2y = a3, (c) (2a;2i>2 + 5xD + l)y = In a;, , (d) x3y"' + 3x*y" + xy' + 8y = 7*-»/*, (e) r2R" + 2rR' - n(n + 1)B = 0, , MISCELLANEOUS METHODS, 3.71. (a) Show that (xD2 - xD + 1) = (D - l)(xD - 1). (6) Thus solve xy" - xy' + y = a;2., 3.72., , Work Problem 3.71(6) by noting that y = x is a solution of the homogeneous equation and letting, y = vx., , 3.73., , Solve y" + 2xy' + x2y = 0 by reducing the equation to canonical form., , 3.74., , Show that xy" + (x + 2)y' + y = 0 is exact and find its solution.
Page 105 :
3.75., , 95, , LINEAR DIFFERENTIAL EQUATIONS, , CHAP. 3], , Use the method of Frobenius to solve the equations, (a) y" + y - 0, (6) xy" - Zy' + xy = 0, (c) y" = xy., , SIMULTANEOUS EQUATIONS, , 3.76., , Solve the system, , 3.77., , Solve the system, , 3.78., , Solve the system (Z>2 + 2)x + Dy = 2 sin t + 3 cos * + 5e-*, Dx + (D2 -l)y = 3 cost- 5 sin t - e~*, if x = 2, y = -3, Dx = 0, Dy = 4 when * = 0., , 3.79., , Solve, , if * = 2, y = 0 when t = 0., , if x = y = z = 0 at i = 0., , APPLICATIONS, , 3.80., , A particle moves on the x axis, attracted toward the origin 0 with a force proportional to its, instantaneous distance from O. If the particle starts from rest at x — 5 cm and reaches, x = 2.5 cm for the first time after 2 seconds, find (a) the position at any time t after it starts,, (6) the magnitude of its velocity at x = 0, (c) the amplitude, period and frequency of the vibration, and (d) the instantaneous acceleration., , 3.81., , The position of a particle moving along the x axis is determined by the equation -j-c + 4 — + 8a; =, 20 cos 2i. If the particle starts from rest at x — 0, find (a) a; as a function of t, (b) the amplitude,, period and frequency after a long time., , 3.82., , A 60 Ib weight hung on a vertical spring stretches it 2 inches. The weight is then pulled down 4, inches and released, (a) Find the position of the weight at any time if a damping force numerically, equal to 15 times the instantaneous velocity is acting. (6) Is the motion oscillatory, overdamped, or critically damped?, , 3.83., , The weight on a vertical spring undergoes forced vibrations according to the equation, , ^72/v*, , dor, , where x is the displacement from the equilibrium position and a > 0 is a constant. If x —, dx/dt = 0 when t = 0, find (a) a; as a function of t, (b) the period of the external force for, which resonance occurs., 3.84., , The charge on the capacitor in the network of Fig. 3-7 is 2 coulombs. If the switch K is closed, at time t = 0, find the charge and the current at any time t > 0 when (a) E = 100 volts,, (6) E = 100 sin 4f., , Fig- 3-7, 3.85., , Fig. 3-8, , Find the currents /, 7t and /2 in the network of Fig. 3-8 and also the charge Q on the capacitor if, (a) E = 360, (b) E = 600e~5t sin 3f. Assume the charge and currents are zero at t = 0.
Page 106 :
96, , LINEAR DIFFERENTIAL EQUATIONS, , Answers to Supplementary Problems, 3.49., , (a) (D2 + 4D + 5)y = e~x,, , 3.50., , (a) Both equal sin 2x - 8 cos 2x., , 3.51., , (a) -2*2 - 8x + 21. -2a;2 - 6» + 22, (6) No, , 3.52., , (a) Sx2 + 2a;3ea: + Ba^e* + 2xex, , 3.56., , y, , 3.59., , (a) y, , 3.60., , (a) j/, , (6) (2Z>« + 2D- l)y = x,, , (6) Yes, , 3.55., , =, , (c) y, , 3.61., , (a) y, , (c) y, (d) V, , (•) y, 3.62., , j/ = ct 4, , 3.63., , (a) j/ =, (6) y =, , 3.64., , (o) y =, , 3.65., , y =, , 3.66., , (a) j/ =, , C l -I, , (6) V =, (c) y =, (<*) y =, , 3.67., , (a), (6), (c), , 3.68., , (o) y =, (6) y =, (c) !/ =, , 3.70., , (a) y =, , (d) V =, (e) R =, 3.71., 3.73., , (c) (xD + 2)y = 1, , 3.74., , Functions are linearly independent., , [CHAP. 3
Page 107 :
CHAP. 3], , LINEAR DIFFERENTIAL EQUATIONS, , 97, , 3.75., , 3.76., 3.77., 3.78., , 3.79., , 3.80., , 3.81., , 3.82., , 3.83., , 3.84., , 3.85., , (a) x = 5 cos Ort/6), (b) 5)r/6(cm/sec), (c) amp. = 5 cm, period = 12 sec, freq. = Jj cycle per second,, (d) -25,r2/36(cm/sec2)
Page 108 :
Chapter 4, Laplace Transforms, , DEFINITION OF A LAPLACE TRANSFORM, The Laplace transform of a function /(£) is defined as, , (1), and is said to exist or not according as the integral in (1) exists [converges] or does not exist, [diverges]. In this chapter we assume that s is real. Later [Chapter 14] we shall find it, convenient to take s as complex., Often in practice there will be a real number s0 such that the integral (1) exists for, s > so and does not exist for s a s0. The set of values s > so for which (1) exists is called, the range of convergence or existence of X? {/(*)}• It may happen, however, that (1) does, not exist for any value of s [see Problem 4.50]., The symbol ^ in (1) is called the Laplace transform operator. We can show that =C is a, linear operator, i.e., •C {ci/i(t) + c2/2(f)} = ci.C {/i(i)} + C2.Ç {/i(*)}, (*), LAPLACE TRANSFORMS OF SOME ELEMENTARY FUNCTIONS, In the following table we give Laplace transforms of some special elementary functions, together with the range of existence or convergence. Often, however, we shall omit this, range of existence since in most instances it can easily be supplied when needed., , /(*), /(*>, 1, , ., , < {/(*)} = f<«), , je {/(*)} = F(t), , 1, , s> 0, , 2., , tn, , n = 1, 2, 3, . . ., , s>0, , 3., , f, , p > -1, , 8>0, , 4., , ea*, , s>a, , 5., , cos at, , s>0, , 6., , sin at, , s >0, , 7., , cosh at, , s > \a\, , 8., , sinh at, , s > \a\, , 98
Page 109 :
CHAP. 4], , LAPLACE TRANSFORMS, , 99, , In entry 3, T(p +1) is the gamma function defined by, , (3), A study of this function is provided in Chapter 9. For our present purposes, however, we, need only the following properties:, (4), The first is called a recursion formula for the gamma function. Note that if p is any positive integer n then r(« +1) = n\, thus explaining the relationship of entries 2 and 3 of the, table., SUFFICIENT CONDITIONS FOR EXISTENCE OF LAPLACE TRANSFORMS, In order to be able to state sufficient conditions on f(t) under which we can guarantee, the existence of £ {/(i)}, we introduce the concepts of piecewise continuity and exponential, order as follows., 1. Piecewise continuity. A function f(t) is said, to be piecewise continuous in an interval if, (i) the interval can be divided into a finite, number of subintervals in each of which f(t), is continuous and (ii) the limits of /(í) as t, approaches the endpoints of each subinterval, are finite. Another way of stating this is to, say that a piecewise continuous function is one, that has only a finite number of finite discontinuities. An example of a piecewise continuous function is shown in Fig. 4-1., , Fig. 4-1, , 2. Exponential order. A function f(t) is said to be of exponential order for t > T if we, can find constants M and « such that |/(i)| a Meat for t > T., Using these we have the following theorem,, Theorem 4-1., , If f(t) is piecewise continuous in every finite interval 0 ë t ë T and is of, exponential order for t > T, then ^ {/(i)} exists for s > a., , It should be emphasized that these conditions are only sufficient [and not necessary],, i.e. if the conditions are not satisfied =C {/(i)} may still exist. For example, <£ {*~1/2} exists, even though i~1/2 is not piecewise continuous in 0 a t ë T., An interesting theorem which is related to Theorem 4-1 is the following, Theorem 4-2. If /(i) satisfies the conditions of Theorem 4-1, then, It follows that if lim F(s) •/= 0, then f(t) cannot satisfy the conditions of, Theorem 4-1., INVERSE LAPLACE TRANSFORMS, If £ (f(t)} = F(s), then we call /(i) the inverse Laplace transform of F(s) and write, XT' (F(s)} = f(t)., Example 1. Since •€{*}=-=,, we have ^, 8, , -, , 1, , = t.
Page 110 :
100, , LAPLACE TRANSFORMS, , [CHAP. 4, , While it is clear that whenever a Laplace transform exists it is unique, the same is not, true for an inverse Laplace transform., •<, _, we can show that ^ {/(i)} = 1/s2- However, this func(JLO t — 2, tion f(t) differs from that of Example 1 at t — 2 although both have the same, Laplace transform. It follows that .C""1 (1/s2) can represent two (or more) different, functions., , Example 2. If, , /(i) =, , We can show that if two functions have the same Laplace transform, then they cannot, differ from each other on any interval of positive length no matter how small. This is, sometimes called Lerch's theorem. The theorem implies that if two functions have the, same Laplace transform, then they are for all practical purposes the same and so in, practice we can take the inverse Laplace transform as essentially unique. In particular if, two continuous functions have the same Laplace transform, they must be identical., The symbol jC_~l is called the inverse Laplace transform operator and is a linear, operator, i.e., JTl(*Fi(8) + C2F2(s)} = d/i(í) + c2/2(í), LAPLACE TRANSFORMS OF DERIVATIVES, We shall find that Laplace transforms provide useful means for solving linear differential equations. For this reason it will be necessary for us to find Laplace transforms, q£ derivatives. The following theorems are fundamental., Theorem 4-3. Let /(i) be continuous and have a piecewise continuous derivative /'(i) in, every finite interval O á í a T. Suppose also that /(i) is of exponential, order for t > T. Then, This can be extended as follows., Theorem 4-4., , •C {/'(«)} = «^ {/(«)}-/(O), , Let f(t) be such that /<n~°(i) is continuous and /(n)(£) piecewise continuous, in every finite interval 0 a t a T. Suppose also that f(t),f'(t),...,/<"-"(*), are of exponential order for t > T. Then, •C {/°°(*)} = sn=C {/(«)} - *"-V(0) - s"-2/'(0), /<»-»(0), , THE UNIT STEP FUNCTION, The unit step function, also called Heaviside's, unit step function, is defined as, 1f(i-a), , =, , fO t < a, 1, \l t>a, , and is shown graphically in Fig. 4-2., It is possible to express various discontinuous, functions in terms of the unit step function., Fig. 4-2, We can show [Problem 4.17] that the Laplace transform of the unit step function is, s>0, , and similarly we have
Page 111 :
LAPLACE TRANSFORMS, , CHAP. 4], , 101, , SOME SPECIAL THEOREMS ON LAPLACE TRANSFORMS, Because of the relationship between Laplace transforms and inverse Laplace transforms,, any theorem involving Laplace transforms will have a corresponding theorem involving, inverse Laplace transforms. In the following we shall consider some of the important, results involving Laplace transforms and corresponding inverse Laplace transforms. In, all cases we assume that f(t) satisfies the conditions of Theorem 4-1., then, , Theorem 4-5 [First translation theorem]., then, , Similarly if, , then, , Theorem 4-6 [Second translation theorem]., Similarly if, Theorem 4-7., , If, , then, then, then, , Similarly if, , then, , Theorem 4-8., , Similarly if, , then, , Theorem 4-9 [Periodic functions]. If f(t) has period, , Theorem 4-10 [Integration]., , Similarly if, , Theorem 4-11. If lim, , exists and, , then, , then, , then, , then
Page 112 :
102, , LAPLACE TRANSFORMS, , [CHAP. 4, , then, , Theorem 4-12 [Convolution theorem]., , Similarly if, , then, , We call the above integral the convolution of / and g and write, , We have the result f*g = g*f, i.e. the convolution is commutative. Similarly we can prove, that it is associative and distributive [see Problem 4.75]., PARTIAL FRACTIONS, , Although the above theorems are often useful in finding inverse Laplace transforms,, perhaps the most important single elementary method for our purposes is the method of, partial fractions. This is because in many problems which we shall encounter it will be, necessary to find the inverse of P(s)/Q(s) where P(s) and Q(s) are polynomials and the, degree of Q(s) is larger than that of P(s). For illustrations of the method see Problems, 4.39-4.41., SOLUTIONS OF DIFFERENTIAL EQUATIONS BY LAPLACE TRANSFORMS, , The method of Laplace transforms is particularly useful for solving linear differential, equations with constant coefficients and associated initial conditions. To accomplish this, we take the Laplace transform of the given differential equation [or equations in the case, of a system], making use of the initial conditions. This leads to an algebraic equation, [or system of algebraic equations] in the Laplace transform of the required solution. By, solving for this Laplace transform and then taking the inverse, the required solution is, obtained. For illustrations see Problems 4.42-4.44., APPLICATIONS TO PHYSICAL PROBLEMS, , Since formulation of many physical problems leads to linear differential equations, with initial conditions, the Laplace transform method is particularly suited for obtaining, their solutions. For applications to various fields see Problems 4.45-4.47., LAPLACE INVERSION FORMULAS, , There exists a direct method for finding inverse Laplace transforms, called the complex, inversion formula. This makes use of the theory of complex variables and is considered, in Chapter 14.
Page 113 :
CHAP. 4], , LAPLACE TRANSFORMS, , 103, , Solved Problems, LAPLACE TRANSFORMS OF ELEMENTARY FUNCTIONS, 4.1., , Prove that, , provided s — a > 0, i.e. s > a, , 4.2., , and p > —1., , (a) Prove that, , (6) Show that if p = n, a positive integer, then, , where s > 0., , Let st = u and note that in order for the integral to converge we, , («), , must have s > 0. Then the integral equals, , The restriction p > — 1 occurs because the integral defining the gamma function converges, if and only if p > —1., (6) Integrating by parts, we have, , i.e., , If p = n, then, , But, , 4.3., , Thus, , and so from, , Prove that =Ç is a linear operator., We must show that if c1( e2 are any constants and f i ( t ) , f 2 ( t ) any functions whose Laplace transforms exist, then, We have, , and so =Ç is a linear operator.
Page 114 :
104, , 4.4., , LAPLACE TRANSFORMS, , [CHAP. 4, , Prove that, Method 1., , We have if s > 0,, , Then taking real and imaginary parts, we have, , or, , Method 2., , By direct integration,, , 4.5., , Find the Laplace transforms of each of the following:, , 4.6., , Find the Laplace transforms of each of the following:, , Since this integral does not converge, the Laplace transform does not, exist., In parts (a), (b), (c) we have omitted the range of existence which can easily be supplied.
Page 115 :
CHAP. 4], , 4.7., , If, , LAPLACE TRANSFORMS, , 4.8., , find .Ç {/(Í)}., , /(Í), We have, , 105, , JL {/(*)}, , Find ^ {sin í cos £}., We have sin 2i = 2 sin í cos t so that sin t cos t = £ sin 2i., , Thus, , EXISTENCE OF LAPLACE TRANSFORMS, 4.9. Prove (a) Theorem 4-1, and (6) Theorem 4-2, page 99., (a) We have, Now since f(t) is piecewise continuous for Q S t S T so also is e~ st /(t), and thus the first, integral on the right exists., To show that the second integral on the right also exists, we use the fact that, |/(t)| S Me* so that, , for s > a and the required result is proved., (6) We have as in part (a),, Now since /(() is piecewise continuous for 0 = t = T, it is bounded, i.e. |/(t)| = K for some, constant K. Using this and the result (1), we have, , Taking the limit as s -> », it follows that, , 4.10. Prove that (a), , F(s) = 0 as required,, , exists,, , (a) In every finite interval ««/(* +4) is continuous fand thus certainly niecewise continuousl., Also for all tZO,, so that ezt/(t + 4) is of exponential order. Thus by Problem 4.9(a) the Laplace transform exists., (6) This follows at once from Problem 4.9(6) and the results of (a).
Page 116 :
106, , LAPLACE TRANSFORMS, , [CHAP. 4, , SOME ELEMENTARY INVERSE LAPLACE TRANSFORMS, 4.11. Prove that ^-1 is a linear operator., We have, since jf is a linear operator [Problem 4.31,, , Thus by definition, which shows that J^~1 is a linear operator., , 4.12. Find, , (d), , Since, , Then letting p = k — 1,, , 4.13., , Find, , LAPLACE TRANSFORMS OF DERIVATIVES, 4.14. Prove Theorem 4-3, page 100., Since f'(t) is piecewise continuous in 0 a t S T, there exists a finite number of subintervals,, say (0, TJ, (Tlt T2), ..., (Tn,T), in each of which f'(t) is continuous and where the limits of f'(t), as t approaches the endpoints of each subinterval are finite. Then
Page 117 :
CHAP. 4], , LAPLACE TRANSFORMS, , 107, , The right side becomes on integrating by parts,, , Since f(t) is continuous, we can thus write, (1), , Now since f(t) is of exDonential order, we have if T is large enoueh., (2), st, , Then taking the limit as T -» «> in (1), using the fact that lim e- f(T), , = 0, we have as required, , 4.15. Prove that if /'(£) is continuous and f"(t) is piecewise continuous in every finite, interval 0 a t a T and if f(t) and f'(t) are of exponential order for í > T, then, Let g(t) = f'(t)., , Then g(t) satisfies the conditions of Theorem 4-3 so that, , Thus, , 4.16. Let f(t) = teat. (a) Show that f(t) satisfies the equation f'(t) = af(t) + eat. (b) Use, part (a) to find ^{ieot}., , Thus using Problem 4.14, we have since, , i.e., , THE UNIT STEP FUNCTION, 4.17. Prove that, We have
Page 118 :
108, , LAPLACE TRANSFORMS, , in terms of the unit step function and, , 4.18. (a) Express the function f(t), thus (6) obtain its Laplace transform., (a) We have, , The result can also be obtained directly., , SPECIAL THEOREMS ON LAPLACE TRANSFORMS, 4.19. Prove Theorem 4-5, page 101., We have, Then, , 4.20. Prove Theorem 4-6, page 101., Method 1., , Since, , Method 2., , Since, , we have, , F(s), , 4.21. Prove Theorem 4-7, page 101., where we have used the transformation, , [CHAP. 4
Page 119 :
CHAP. 4], , LAPLACE TRANSFORMS, , 109, , 4.22. Prove Theorem 4-8, page 101, for (a) n - 1, (6) any positive integer n., we have on differentiating with respect to s and using Leibnitz's, , (a) Since F(s) =, rule,, , Thus, (b), , Thus, , 4.23. Prove Theorem 4-9, page 101., We have, , Then since /(«) has period P > 0, f(v + P) = f(v), f(v + 2P) = f(v), etc., replace the dummy variable v by t. Thus, , Furthermore, we can, , where, , 4.24. Prove Theorem 4-10, page 101., Let, , Then, , Thus, , and so, , Prom this we have, , 4.25. Prove Theorem 4-11, page 101., We shall assume that lim^ exists [as well as the conditions of Theorem 4-1. page 991,, t-»o t, ', otherwise its Laplace transform may not exist. Then we have if g(t) = /(<)/«, or /(i) = tg(t),, , Thus
Page 120 :
LAPLACE TRANSFORMS, , 110, , [CHAP. 4, , Now since g(t) satisfies the conditions of Theorem 4-1, page 99, it follows that lim G(s) = 0. Then, s, from (2) we see that c must be infinite and so, ~*°°, , 4.26. Prove Theorem 4-12, page 102., We have, Then, , where we have used the transformation t = u + v from the uv plane to the ut plane., , APPLICATIONS OF THE THEOREMS TO FINDING LAPLACE TRANSFORMS, 4.27. Find, (a) Since, , (b) Since, (e), , Since, , 4.28. Find ^ (F(t)} where, We have, , Then using Theorem 4-6,, , This can also be obtained directly without using Theorem 4-6.
Page 121 :
CHAP. 4], , LAPLACE TRANSFORMS, , 111, , 4.29. Given that, By Theorem 4-7, page 101,, , 4.30. Find, By Theorem 4-8, page 101, we have since, , 4.31. Find the Laplace transform of the function of period 2-n- which in the interval, O á í < 2w is given by f(t), The graph of this function, often called a rectified sine wave, is shown in Fig. 4-3., , By Theorem 4-9, page 101, the Laplace transform is given by, , 4.32. Find, Since, , by L'Hospital's rule and 1 — e~f is continuous and of expo-, , nential order, the conditions of Theorem 4-11 apply., follows that, , Then since, , APPLICATIONS OF THE THEOREMS TO FINDING, INVERSE LAPLACE TRANSFORMS, 4.33. Find
Page 122 :
112, , LAPLACE TRANSFORMS, , [CHAP. 4, , Now since, we have by Theorem 4-5,, , Thus, Then, , (6) Prom Problem 4.12(d) we have, by Theorem 4-5,, , (o), , 4.34. Find, (a) Since, , cos4t, we have by Theorem 4-6,, , (6) Since, , and so by Theorem 4-6,, , 4.35. Find, Let F(s), , so that, , Then by Theorem 4-8,, , or, , 4.36. Find, Since, , on letting u = v2., , we have by Theorem 4-10,
Page 123 :
CHAP. 4], , LAPLACE TRANSFORMS, , 113, , 4.37. Find, Let, [Theorem 4-12],, , Then, , Thus by the convolution theorem, , 4.38. Solve for y(t) the equation, , Taking the Laplace transform, we have by the convolution theorem,, , Then, and so, , which can be checked as a solution., The given equation is called an integral equation since the unknown function occurs under the, integral., , PARTIAL FRACTIONS, 4.39. Find, , To determine constants A, B, C, multiply by (a — 2)(s + l)(s — 3) so that, This must be an identity and thus must hold for all values of s. Then by letting s = 2, —1,3 in, succession we find A — -4/3, B = —1/6, C — 7/2. Thus, , 4.40. Find, , or, , Letting « = 1, we find A - 2. Letting s — 0, we find A — C = 1 so that C = 1. Then letting, 8 equal any other number, say —1, we find, —2 = 2A — 2(C — B) and B = —2. Thus, , Another method., Multiplying (1) by s after finding A and then letting s-» °°,, B = —A = —2. This method affords some simplification of procedure., , we find A+B = Q, , or
Page 124 :
114, , LAPLACE TRANSFORMS, , [CHAP. 4, , 4.41. Find, We have, or clearing of fractions, Letting s = -1, we find A = —1. Letting s = 2, we find B = -1. Letting a equal two other, numbers, say 0 and 1, we find —2C + 4D = 0 and —2C + 2D = -2 from which C = 2, D = 1., Thus, ., ., , Another method., , On multiplying (1) by a, after finding A, and letting s -* «°, we find that A+D = 0 or, D = 1, providing some simplification in the procedure., , SOLUTIONS OF DIFFERENTIAL EQUATIONS, 4.42. Solve y"(t) + y(t) = 1 given y(Q) = 1, yf(0) = 0., Take the Laplace transform of both sides of the given differential equation and let, Y = Y(s) = £ (y(t)}. Then, Since, , this becomes, , Thus, which is the required solution., , 4.43. Solve, Taking the Laplace transform of the given differential equation,, , Then using j/(0) = 2, j/'(0) = —1 and solving this algebraic equation for Y, we find using partial, fractions,, Thus taking the inverse Laplace transform, we obtain the required solution, , 4.44. Solve, Taking the Laplace transform of the given differential equation and using the initial conditions,, which can be written, , or
Page 125 :
CHAP. 4], , LAPLACE TRANSFORMS, , 115, , Now by using the special theorems on Laplace transforms [see Problem 4.78],, , and so we find the required solution, , APPLICATIONS TO PHYSICAL PROBLEMS, 4.45. Solve Problem 2.32, page 56, by using Laplace transforms., As in Problem 2.32, the differential equation is, (1), , (a) If E = 40 the Laplace transform of (1) is, , where / = =C {/}., , Then using 1(0) — 0 and solving for /, we find, , Thus, (6) If E = 20e-M then, , so that, Thus, (c) If E = 50 sin 5«, then, , so that, Then, , 4.46. A mass m [Fig. 4-4] is suspended from the end of a vertical spring of constant K, [force required to produce unit stretch]. An external force F(t) acts on the mass as, well as a resistive force proportional to the instantaneous velocity. Assuming that, x is the displacement of the mass at time í and that the mass starts from rest at, x = 0, (a) set up a differential equation for the motion and (&) find x at any time i., (a) The resistive force is given by, The restoring force, is given by — KX. Then by Newton's law,, , or, , where, , Fig. 4-4
Page 126 :
[CHAP. 4, , LAPLACE TRANSFORMS, , 116, , we obtain, , (b) Taking the Laplace transform of (1), using, so that on using (2),, , (S), where, Case 1, R > 0., , There are 3 cases to be considered., In this case let R = w2. We have, , Then using the convolution theorem, we find from (3), , te-pt/2m, , Case 2, R = 0., in (3) yields, , In this case, , Case 3, R < 0., , In this case let R = -a2. We have, , an<i, , the convolution theorem, , Then using the convolution theorem (3) yields, , 4.47. Solve Problem 3.48, page 93, by Laplace transforms., From equations (1) and (2) of Problem 3.48 we have on taking Laplace transforms, using, •C{/l}=/i, -Ç{/ 2 }=/2, / = /! + /», , -C {/}=A, , Using /i(0) = 0, 72(0) = 0, these become, , Solving,, , or, , Then, , +4
Page 127 :
CHAP. 4], , LAPLACE TRANSFORMS, , 117, , Supplementary Problems, LAPLACE TRANSFORMS OP ELEMENTARY FUNCTIONS, 4.48. Find the Laplace transforms of each of the following:, , 4.49., , Find ^ {/(i)} in each case:, , 4.50., , Prove that e*2 does not have a Laplace transform., , 4.51., , Determine whether sin i2 has a Laplace transform and justify your conclusions., , 4.52., , Find (a) £ {10 sin 3f cos5i}, (6) ^ {/(i)> if f(t) =, , ELEMENTARY INVERSE LAPLACE TRANSFORMS, 4.53., , Find, , 4.54., , Find, , LAPLACE TRANSFORMS OF DERIVATIVES, 4.55. Verify the result ^ {/'(«)} = s £ {f(s)} - f(0), (6) f(t) = cos 5t, (c) /(t) = «2 + 2t - 4., , for each of the following:, , (a) f(t) = 3e2t,, , 4.56., , Verify the result ^ {/"(*)} = s 2 ^ {/(i)> - s/(0) - f(0) for each of the functions in Problem 4.55., , 4.57., , If /(t) = t sin at, (a) show that /"(t) + a2/(t) = 2a cos at and thus (6) find ^ {/(t)}., , 4.58., , Does the result, , hold for, Explain,, , 4.59., , Prove that, , THE UNIT STEP FUNCTION, 4.60. Find, 4.61., , Discuss the significance of, , 4.62., , Find, , 4.63., , (a) Express the function, its Laplace transform., , giving conditions under which it holds., , and graph each of the given functions of t., where p is any positive integer., and graph., in terms of the unit step function and (6) find
Page 128 :
118, , LAPLACE TRANSFORMS, , [CHAP. 4, , MISCELLANEOUS LAPLACE TRANSFORMS, 4.64., , Find, , 4.65., , Find, , 4.66., , Find, , 4.67., , (a) Graph the function f(t) = t, 0 S í < 4 which is extended periodically with period 4 and, (6) find the Laplace transform of this function., , 4.68., , Find, , 4.69., , Verify that, , 4.70., , Show that, , 4.71., , Find, , 4.72., , Find, , 4.73., , Show that, , 4.74., , Evaluate («) t*e* and (6) £{t*e*}., , 4.75., , Prove that (a) f*g = g*f,, significance of the results., , and generalize., , where, , (6) f*(g + h) = f*g + f*h,, , (c) f*(g*h) = (f*g)*h, , MISCELLANEOUS INVERSE LAPLACE TRANSFORMS, 4.76., , Find, , 4.77., , Find, , 4.78., , Find, , 4.79., , Find, , 4.80., , Use the series for eu to show that, , 4.81., , Solve the integral equations, , 4.82., , Use Laplace transforms to evaluate, , [Hint. Consider, , and discuss the
Page 129 :
CHAP. 4], , LAPLACE TRANSFORMS, , 119, , SOLUTIONS OF DIFFERENTIAL EQUATIONS, 4.83., Solve each of the following:, (a) y"(t) - 8v'(t) + 2y(t) = 4; y(0) = 1, y'(0) = 0, (6) y"(t) + Uy(t) = 32i; y(0) = 3, y'(0) = -2, (c) y"(t) + 4»'(t) + 4»(t) = Be-»; 0(0) = -2, j/'(0) = 8, (d) y'"(t) + y'(t) = t + 1, 4.84., , Solve y'"(t) + 8y(t) - 32Í3 - 16i if j/(0) = i/'(0) = j/"(0) = 0., , 4.85., , Solve the simultaneous equations, , C2x(t) - y(t) - y'(t) = 4(1-e-«), , { 2x'(t) + y(t) - 2(l + 3e-2t), subject to the conditions *(0) = j/(0) = 0., 4.86., , Solve (a) Problem 3.76, (b) Problem 3.77, (c) Problem 3.78, (d) Problem 3.79 on page 95 by Laplace, transforms., , APPLICATIONS TO PHYSICAL PROBLEMS, 4.87., Solve (a) Problem 3.80, (b) Problem 3.81, (c) Problem 3.83, (d) Problem 3.84, page 95, by Laplace, transforms., 4.88., , Use Laplace transforms to find the charge and current at any time in a series circuit having an, inductance L, capacitance C, resistance R and e.m.f. E(t). Treat all cases assuming that the initial, charge and current are zero., , 4.89., , Solve Problem 3.85, page 95, by Laplace transforms., , 4.90., , (a) A particle of mass m moves along the x axis in such a way that the force acting is given by, F(t) =, , . Assuming that it starts from rest at the origin, find the position of the, , particle at any time and interpret physically, (b) Discuss the result in (a) if the limit is taken as, €^•0., , 4.91., , Let f(t) be any continuous function., , Suppose there exists a function S(t) such that, , Show that (a) £ {S(t)} = 1, (6) £ {8(í- Í0)} = e~^o,, Dirae delta function)., 4.92., , In Problem 4.90 replace the force by F(t) = F0 8(t), between &(t) and the function, , and solve., , We often call S(t) the, , Discuss a possible relationship, , [Hint. Examine the case where, , 4.93. Graph the function F(t), the relationship of this problem with ^, , (c) £{S'(t)} = s., , -1, , 2, , {s }., , Find (a) {F(t)} and (b) lim {F(t)}. Discuss, «-»o, , Answers to Supplementary Problems, 4.48., , 4.49.
Page 130 :
120, , LAPLACE TRANSFORMS, , [CHAP. 4, , 4.52., 4.53., , 4.54., 4.60., 4.62., 4.63., 4.64., , 4.65., , 4.67., , 4.71., 4.72., 4.74., 4.76., , 4.77., , 4.78., , 4.79., 4.83., , 4.84., 4.85., 4.90., , 4.82.
Page 131 :
Chapter 5, Vector Analysis, , VECTORS AND SCALARS, , There are quantities in physics characterized by both, magnitude and direction, such as displacement, velocity,, force and acceleration. To describe such quantities, we, introduce the concept of a vector as a directed line segment, PQ from one point P called the initial point to another point, Q called the terminal point. We denote vectors by bold, faced letters or letters-» with an arrow over them. Thus PQ, is denoted by A or A as in Fig. 5-1. The magnitude or, Fig. 5-1, length of the vector is then denoted by \PQ\, PQ, |A| or |A|., Other quantities in physics are characterized by magnitude only, such as mass, length, and temperature. Such quantities are often called scalars to distinguish them from, vectors, but it must be emphasized that apart from units such as feet, degrees, etc.,, they are nothing more than real numbers. We can thus denote them by ordinary letters, as usual., VECTOR ALGEBRA, , The operations of addition, subtraction and multiplication familiar in the algebra of numbers are, with suitable, definition, capable of extension to an algebra of vectors., The following definitions are fundamental., 1. Two vectors A and B are equal if they have the, same magnitude and direction regardless of their, initial points. Thus A = B in Fig. 5-1 above., 2. A vector having direction opposite to that of vector, A but with the same magnitude is denoted by —A, [see Fig. 5-2]., Fig.5_2, 3. The sum or resultant of vectors A and B of Fig. 5-3(a) below is a vector C formed, by placing the initial point of B on the terminal point of A and joining the, initial point of A to the terminal point of B [see Fig. 5-3(&) below]. The sum C, is written C = A + B. The definition here is equivalent to the parallelogram law, for vector addition as indicated in Fig. 5-3(c) below., , (a), , (b), , Fig. 5-3, , 121, , (c)
Page 132 :
122, , [CHAP. 5, , VECTOR ANALYSIS, , Extensions to sums of more than two vectors are immediate. For example,, Fig. 5-4 below shows how to obtain the sum or resultant E of the vectors, A, B, C and D., , Pig. 5-4, , 4. The difference of vectors A and B, represented by A —B, is that vector C which, added to B gives A. Equivalently, A — B may be defined as A + (—B). If A = B,, then A — B is defined as the null or zero vector and is represented by the symbol 0., This has a magnitude of zero but its direction is not defined., 5. Multiplication of a vector A by a scalar m produces a vector raA with magnitude m\ times the magnitude of A and direction the same as or opposite to that, of A according as m is positive or negative. If m = 0, mA = 0, the null vector., LAWS OF VECTOR ALGEBRA, If A, B and C are vectors, and m and n are scalars, then, Commutative Law for Addition, 1. A + B = B + A, Associative Law for Addition, 2. A + (B + C) = (A + B) + C, Associative Law for Multiplication, 3. m(nA.) = (mn)A — n(mA), 4. (ra + w)A = mA + «A, Distributive Law, Distributive Law, 5. m(A + B) = mA + mB, , Note that in these laws only multiplication of a vector by one or more scalars is, defined. On pages 123 and 124 we define products of vectors., UNIT VECTORS, Unit vectors are vectors having unit length. If A is any vector with length A > 0,, then A/A is a unit vector, denoted by a, having the same direction as A. Then A = Aa., RECTANGULAR UNIT VECTORS, The rectangular unit vectors i, j and k are unit vectors having the direction of the positive x, y and z axes, of a rectangular coordinate system [see Fig. 5-5]. We use, right-handed rectangular coordinate systems unless otherwise specified. Such systems derive their name from the, fact that a right threaded screw rotated through 90° from, Ox to Oy will advance in the positive z direction. In general,, , Fig. 5-5
Page 133 :
123, , VECTOR ANALYSIS, , CHAP. 5], , three vectors A, B and C which have coincident initial points and are not coplanar are said, to form a right-handed system or dextral system if a right threaded screw rotated through, an angle less than 180° from A to B will advance in the direction C [see Fig. 5-6 below]., , Pig. 5-6, , Fig. 5-7, , COMPONENTS OF A VECTOR, Any vector A in 3 dimensions can be represented with initial point at the origin O, of a rectangular coordinate system [see Fig. 5-7 above]. Let (Ai, A2, A3) be the rectangular, coordinates of the terminal point of vector A with initial point at O. The vectors Aii,, A 2 j and Ask are called the rectangular component vectors, or simply component vectors,, of A in the x, y and z directions respectively. A.\, A2 and A3 are called the rectangular, components, or simply components, of A in the x, y and z directions respectively., The sum or resultant of Aii, A 2 j and Ask is the vector A, so that we can write, A = Aii + A 2 j + Aak, , (1), , The magnitude of A is, (2), , In particular, the position vector or radius vector i from O to the point (x,y,z) is, written, T = xi + yj + zk, (3), and has magnitude, DOT OR SCALAR PRODUCT, The dot or scalar product of two vectors A and B, denoted by A • B (read A dot B), is denned as the product of the magnitudes of A and B and the cosine of the angle, between them. In symbols,, A- B = AB cos 0, O a 0 ^TT, (4), Note that A • B is a scalar and not a vector., The following laws are valid:, 1. A « B = B - A, Commutative Law for Dot Products, 2. A-(BH-C) = A - B + A - C, Distributive Law, 3. m(A'B) = (mA)-B = A-(wB) = (A'B)m,, where m is a scalar., 4., , i - i = j - j = k-k = 1,, , i-j = j-k = k-i = 0
Page 134 :
VECTOR ANALYSIS, , 124, , 5. If A = Aii + Azj + Ask and B =, A-B =, A-A =, B-B =, , [CHAP. 5, , Ba + B2j + #sk, then, AiBi + AzB2 + AsB», A2 = Al +Al +AÎ, Bz = B\ + Bl+Bl, , 6. If A • B = 0 and A and B are not null vectors, then A and B are perpendicular., , CROSS OR VECTOR PRODUCT, The cross or vector product of A and B is a vector C = A x B (read A cross B)., The magnitude of A x B is defined as the product of the magnitudes of A and B and the, sine of the angle between them. The direction of the vector C = A x B is perpendicular, to the plane of A and B and such that A, B and C form a right-handed system. In, Symbols, , ', A x B = AB sin e u,, 0 g 0ë *, (5), where u is a unit vector indicating the direction of A x B. If A = B or if A is parallel, to B, then sin 0 = 0 and we define A x B = 0., The following laws are valid:, 1. A x B = -B x A, (Commutative Law for Cross Products Fails), 2. A x ( B + C) = AxB + AxC, Distributive Law, 3. m(A x B) = (mA) x B = A x (mB) = (A x B)m,, where m is a scalar., 4., , ixi, , 5. If, , = jxj, , = kxk = 0,, , A = Aii + A2j + Ask, , i x j = k,, , and, , AxB, , jxk = i,, , B = B¿ + B2j + #sk,, , =, , then, , i, J, k, Ai A 2 As, ¡j\, , 6., 7., , kxi = j, , Jj2, , -O3, , |A x B| = the area of a parallelogram with sides A and B., If A x B = 0 and A and B are not null vectors, then A and B are parallel., , TRIPLE PRODUCTS, Dot and cross multiplication of three vectors A, B and C may produce meaningful, products of the form (A-B)C, A- (B x C) and A x (B x C). The following laws are valid:, 1. (A • B)C ¥= A(B'C) in general, 2. A - ( B x C ) = B ' ( C x A ) = C - ( A x B ) = volume of a parallelepiped having A, B,, and C as edges, or the negative of this volume according as A, B and C do or, do not form a right-handed system. If A = Aii + A2j + Ask, B = Bii + B2j + Bsk, and C = Cii + C2j + Csk, then, Ai A 2 As, A-(BXC) =, B! Bz B8, (6), , Ci C2 Cs, 3. A x (B x C) ¥> (A x B) x C, (Associative Law for Cross Products Fails), 4. A x ( B x C ) = (A-C)B- (A-B)C, ( A X B ) X C = (A-CJB-(B-C)A, The product A • (B x C) is sometimes called the scalar triple product or box product, and may be denoted by [ABC]. The product A x ( B x C ) is called the vector triple, product.
Page 135 :
CHAP. 5], , VECTOR ANALYSIS, , 125, , In A • (B x C) parentheses are sometimes omitted and we write A • B x C. However,, parentheses must be used in A x ( B x C ) (see Problem 5.25). Note that A - ( B x C ) =, (A x B) • C. This is often expressed by stating that in a scalar triple product the dot and, the cross can be interchanged without affecting the result [see Problem 5.22]., VECTOR FUNCTIONS, If corresponding to each value of a scalar u we associate a vector A, then A is called, a function of u denoted by A(u). In three dimensions we can write A(u) = Ai(u)i +, Aa(u)j + As(u)k., The function concept is easily extended. Thus if to each point (x,y,z) there corresponds a vector A, then A is a function of (x, y, z), indicated by A(x, y, z) = A\(x, y, z)i +, A2(x, y, z)j + Aa(x, y, z)k., We sometimes say that a vector function A(x, y, z) defines a vector field since it associates a vector with each point of a region. Similarly $(x, y, z) defines a scalar field since it, associates a scalar with each point of a region., LIMITS, CONTINUITY AND DERIVATIVES OF VECTOR FUNCTIONS, Limits, continuity and derivatives of vector functions follow rules similar to those, for scalar functions already considered. The following statements show the analogy, which exists., 1. The vector function A(u) is said to be continuous at MO if given any positive number e, we can find some positive number 8 such that \A(u) — A(uo)\ < £ whenever, \u-uo < S. This is equivalent to the statement lim A(u) — A(uo)., 2. The derivative of A(u) is defined as, (7), , provided this limit exists. In case A(u) = Ai(u)i + A2(u)j + A3(w)k; then, , Higher derivatives such as d2A/du2, etc., can be similarly defined., 3. If A(x, y, z) = Ai(x, y, z)i + A2(x, y, z)j + A3(x, y, z)k, then, is the differential of A., , (*), , 4. Derivatives of products obey rules similar to those for scalar functions. However,, when cross products are involved the order may be important. Some examples are:
Page 136 :
[CHAP. 5, , VECTOR ANALYSIS, , 126, , GEOMETRIC INTERPRETATION OF A VECTOR DERIVATIVE, , If r is the vector joining the origin O of a, coordinate system and the point (x,y,z), then, specification of the vector function r(u) defines, x, y and z as functions of u. As u changes, the, terminal point of r describes a space curve, (see e Fig. 5-8) having parametric equations, x - x(u), y = y(u), z = z(u). If the parameter u, is the arc length s measured from some fixed, point on the curve, then, (9), , is a unit vector in the direction ot the tangent, to the curve and is called the unit tangent, vector. If u is the time t, then, =, , Fig. 5-8, , (10), , v, , is the velocity with which the terminal point of r describes the curve. We have, =, , vT, , from which we see that the magnitude of v, often called the speed, is v = ds/dt., =, , a, , («), , Similarly,, (12), , is the acceleration with which the terminal point of r describes the curve. These concepts, have important applications in mechanics., GRADIENT, DIVERGENCE AND CURL, (del) defined by, Consider the vector operator, (13), , Then if <t>(x, y, z) and A(x, y, z) have continuous first partial derivatives in a region (a condition which is in many cases stronger than necessary), we can define the following., 1. Gradient. The gradient of ^ is defined by, grad, , An interesting interpretation is that if <¡>(x, y, z) — c is the equation of a surface,, is a normal to this surface (see Problem 5.36), then, 2. Divergence. The divergence of A is defined by, div A, , (15)
Page 137 :
CHAP. 5], , VECTOR ANALYSIS, , 127, , 3. Curl. The curl of A is defined by, curl A, , (16), , Note that in the expansion of the determinant, the operators d/dx, d/dy, d/dz must, precede Ai, At, A3., FORMULAS INVOLVING, , If the partial derivatives of A, B, U and V are assumed to exist, then, or grad (U + V) = grad u + grad V, 1., or div (A + B) = div A + div B, 2., or curl (A + B) = curl A + curl B, 3., 4., 5., 6., 7., 8., is called the Laplacian of U, , 9., , is called the Laplacian operator., 10., 11., 12., , The curl of the gradient of U is zero., . The divergence of the curl of A is zero., , ORTHOGONAL CURVILINEAR COORDINATES., The transformation equations, * = f(ui,U2,Us),, , y = g(ui, w2, u3),, , JACOBIANS, z = h(ui,U2,ua), , (17), , [where we assume that f , g , h are continuous, have continuous partial derivatives and, have a single-valued inverse] establish a one to one correspondence between points in an, xyz and uiu¿ua rectangular coordinate system. In vector notation the transformation (17), can be written, r = xi + yj + zk = f(u\, «2, Us)i + g(ui, u-¡, u3)j + h(ui, u2, u3)k, (18)
Page 139 :
129, , VECTOR ANALYSIS, , CHAP. 5], , —, , 3., , =, , 4., , curl A, , —, , Laplacian of $ =, , These reduce to the usual expressions in rectangular coordinates if we replace (ui, u2, Us), by (x,y,z), in which case ei, e2 and e3 are replaced by i, j and k and hi-h2 = hs = 1., SPECIAL CURVILINEAR COORDINATES, 1. Cylindrical Coordinates (p, <j>, z). See Fig. 5-10., Transformation equations:, x — p cos <t>, y = p sin <£, z = z, where p è 0, 0 ë <j> < 2*, -°° < z < «>., Scaie factors: hi = 1, hz = p, ha = 1, Element of arc length: ds2 = dp2 + p2 cZ<£2 + dz2, /aco&wm:, Element of volume: dV — pdpd<j> dz, Laplacian:, , Fig.5-10, , Note that corresponding results can be obtained for polar coordinates in the, plane by omitting z dependence. In such case for example, ds2 = dp2 + p2 d<j>2, while, the element of volume is replaced by the element of area. dA = p dp d<j>., 2. Spherical Coordinates (r,0,<f>). See Fig. 5-11., Transformation equations:, x = r sin 0 cos ó, y = r sin 6 sin A, z — r cos 6, where, Scale factors:, Element of arc length:, Jacobian:, Element of volume:, Laplacian, Other types of coordinate systems are possible., , Fig.5-11
Page 140 :
[CHAP. 5, , VECTOR ANALYSIS, , 130, , Solved Problems, VECTOR ALGEBRA, , 5.1., , Show that addition of vectors is commutative, i.e. A + B = B + A. See Fig. (a) below., and, , OP + PQ = OQ, , or, , A+ B = C, , OR + RQ = OQ, , or, , B+ A = C, , Then A + B = B + A., , _Q, , Fig. (b), , Fig. (a), , 5.2., , Show that the addition of vectors is associative, i.e. A + (B + C) = (A + B) + C., See Fig. (b) above., OP + PQ = OQ = (A + B), , Since, , and, , OP + PR = OR = D,, OQ + QR = OR = D,, , PQ + QR = PR = (B + C), , i.e. A + (B + C) = D, i.e. (A + B) + C = D, , we have A + (B + C) = (A + B) + C., Extensions of the results of Problems 5.1 and 5.2 show that the order of addition of any, number of vectors is immaterial., , 5.3., , Prove that the line joining the midpoint of two sides, of a triangle is parallel to the third side and has, half its length., From Fig. 5-12, AC + CB = AB or b + a = c., Let DE = d be the line joining the midpoints of sides AC, and CB. Then, d = DC + CE = £b + £a = £(b + a) = £c, Thus d is parallel to c and has half its length., , 5.4., , Fig. 5-12, , Prove that the magnitude A of the vector A =, Aii + Azj + Ask is A —, Fig. 5-13., , See, , By the Pythagorean theorem,, (OP)* = (OQ)2 + (QP>, where OP denotes the magnitude of vector OP, etc., Similarly, (OQ)2 = (OR)* + (RQ)*., Then (OP)2 = (OB)2 + (RQ)2 + (QP> or, , A 2 = A2 +A% +Al, i.e. A =, , Fig. 5-13
Page 141 :
5.5., , 131, , VECTOR ANALYSIS, , CHAP. 5], , Determine the vector having initial point, P(xi,yi,zi) and terminal point Q(xz,yz,Zi) and, find its magnitude. See Fig. 5-14., The position, The position, r! + PQ = r2, PQ = r2 - T!, , vector of P is rt = ¡»1i + j/iJ + «jk., vector of Q is r2 = œ2i + j/2j + «2^or, = (a;2i + j/aj + 22k) - («ji + j/jj + ztk), =, , (*2 - *!)i + (j/2 - Vl)î + (*2 ~ *l)k, , Magnitude of PO = P0, Fig. 5-14, , Note that this is the distance between points P and Q., , THE DOT OR SCALAR PRODUCT, 5.6. Prove that the projection of A on B is equal to, A-b, where b is a unit vector in the direction, of B., Through the initial and terminal points of A pass, planes perpendicular to B at G and H respectively as in, the adjacent Fig. 5-15; then, Projection of A on B = GH = EF = A cos e — A • b, , 5.7., , H, , Fig. 5-15, , Prove A- (B + C) = A - B + A-C., Let a be a unit vector in the direction of A; then [see Fig. 5-16], Projection of (B + C) on A = projection of B on A + projection of C on A, (B + C ) - a = B - a + C - a, Multiplying by A,, (B + C ) ' A a = B - A a + C ' A a, and, , (B + C) • A = B • A + C • A, , Then by the commutative law for dot products,, A « (B + C) = A - B + A - C, and the distributive law is valid., , 5.8., , Fig. 5-16, , Prove that (A + B ) - ( C + D) = A - C + A - D + B-C + B'D., By Problem 5.7, (A + B) • (C + D) = A • (C + D) + B • (C + D) = A • C + A • D + B • C + B • D., The ordinary laws of algebra are valid for dot products., , 5.9., , Evaluate each of the following., (a), (6), (c), (d), (e), , 5.10., , i - i = |i||i| cosO° = (1)(1)(1) = 1, i - k = |i||k|cos90° = (1)(1)(0) = 0, k - j = |k|¡j|cos90° = (1)(1)(0) = 0, j-(2i-3j + k) = 2 j - i - 3 j - j + j - k = 0-3 + 0 = -3, (2i-j)-(3i + k) = 2i-(3i + k ) - j-(3i + k), = 6i • i + 2i • k - 3j • i - j • k = 6 + 0 - 0 - 0, , = 6, , If A = Aii + A 2 j + Aak and B = Bii + B2j + B&, prove A - B = AiBi + A&i + A&t, A • B = (Aii + A 2 j + A3k) • (BJ + B2j + B3k), = A t i • (B¿ + BZJ + £3k) + A 2 j • (Bji + B2j + Bsk) + A3k • (Bji + B2j + Bsk), = A^ji • i + At£2i • j + Ai#3i • k + A2BJ • i + A2B2j • j + A2£3j • k, + A-^k • i + A3B2k • j + A3B3k • k, = A,B! + A2J52 + A3B3, , since i - i = j ' j = k « k = l and all other dot products are zero.
Page 142 :
132, , VECTOR ANALYSIS, , [CHAP. 5, , 5.11. If A = Aii + Azj+Ask, show that AA • A = (A)(A) cos 0° = A2. Then A = •, Also,, , A • A = (Aji + A 2 j + A3k) • (Axi + A 2 j + A3k), = (AiXA,) + (A2)(A2) + (A3)(A3) - A\ + £ + A¡, by Problem 5.10, taking B = A., Then A =, , ¡s the magnitude Of, , A>, , Sometimes A • A is written A2., , THE CROSS OR VECTOR PRODUCT, 5.12. Prove A x B = -B X A., , (a), , (&), , Fig. 5-17, , A*B+C has magnitued AB sin and direction such that A,B and C form a right a right handed, system [Fig. 5-17(a) above]., system^ 5-17h(6) %££** ** ^ ' ^ ^^ "* ** B'A ^ ° to, system (Fig.5.17(b) above]., Then D has the same magnitude as C but is opposite in direction, i.e. C = -D, ~I> X A., , The commutative law for cross products is not valid., , 5.13. Prove that A x ( B + C) = AxB +, A x C for the case where A is perpendicular to B and also to C., Since A is perpendicular to B, A X B is a, vector perpendicular to the plane of A and B, and having magnitude AB sin 90° = AB or, magnitude of AB. This is equivalent to multiplying vector B by A and rotating the, resultant vector through 90° to the position, shown in Fig. 5-18., Similarly, A X C is the vector obtained, by multiplying C by A and rotating the, resultant vector through 90° to the position, shown., In like manner, A X (B + C) is the vector, obtained by multiplying B + C by A and, rotating the resultant vector through 90° to, the position shown., Since A X (B + C) is the diagonal of the, parallelogram with A X B and A X C as sides, we have AX (B + C) = AX B + Ax C., ', , „. -, , 10, * lg. 5-18, , &, , "^-handed, or A X B =
Page 143 :
133, , VECTOR ANALYSIS, , CHAP. 5], , 5.14. Prove that A x ( B + C) = AxB + AxC, in the general case where A,B and C are, non-coplanar. See Fig. 5-19., Resolve B into two component vectors, one, perpendicular to A and the other parallel to A,, and denote them by B± and BM respectively., Then B = Bj_ +B| ( ., If 9 is the angle between A and B, then, Bj_ = B sin e. Thus the magnitude of A X B x, is AB sin e, the same as the magnitude of A X B., Also, the direction A X B is the same as the, direction of A X B. Hence A X Bj_ = A X B ., Similarly if C is resolved into two component vectors C N and Cj_, parallel and perpendicular respectively to A, then A X C ± = A X C ., , Fig. 5-19, , Also, since B + C = B± + B,, + Cj_ + C M = (Bj_ + C x ) + (BM + C,,) it follows that, A X (Bj_ + Cj_) = A X (B + C), , Now B X and Cj_ are vectors perpendicular to A and so by Problem 5.13,, A X ( B ± + CX) = A X B x + A X C j _, Then, , AX(B+C), , = A X B + AXC, , and the distributive law holds. Multiplying by —1, using Problem 5.12, this becomes (B + C) X A —, B X A + C X A. Note that the order of factors in cross products is important. The usual laws of, algebra apply only if proper order is maintained., , i j k, 5.15. If A = Aii + A2j + Ask and B = B¿ + B2j + Bak, prove that A x B = Ai A2 A3 ., BI, , BZ, , A X B = (Ati + A 2 j + A3k) X (J5,i + B£ + B3k), = Aji X (B,i + B2j + B3k) + AHJ X (Bji + Eg + B3k) + A3k X (Bji + B2j + B3k), - AjBji X i + A^i X j + A^gi x k + A2B¡j X i + A2B^ x j + A2B3j x k, + A3B,k X i + A3£2k X j + A3B3k X k, i, j k, = (A2B3-A3£2)i + (A 3 B 1 -A 1 B 8 )j + (AjB 2 -A 2 B,)k =, A^ A 2 A 3, Bi, , 5.16., , Bn, , If A = 3i - j + 2k and B = 2i + 3j - k, find A X B., , 5.17. Prove that the area of a parallelogram with, sides A and B is |A x B . See Fig. 5-20., Area of parallelogram = h [B|, = |A| sin e |B|, , = IAXBJ, Note that the area of the triangle with sides A, and B = -||AXB|., , Fig. 5-20, , B», , By
Page 144 :
[CHAP. 5, , VECTOR ANALYSIS, , 134, , 5.18. Find the area of the triangle with vertices at P(2,3,5), Q(4,2,-l), ¿2(3,6,4)., PQ = (4-2)i + (2-3)j + (-l-5)k = 2i - j - 6k, PB = (3-2)i + (6-3)j + (4-6)k = i + 3j - k, Area of triangle = £ | P Q X P R | = £ | (2i-j-6k) X (i +3j-k) |, , TRIPLE PRODUCTS, 5.19. Show that A • (B x C) is in absolute, value equal to the volume of a parallelepiped with sides A, B and C. See, Fig. 5-21., Let n be a unit normal to parallelogram, I, having the direction of B X C, and let h be, the height of the terminal point of A above, the parallelogram I., , Fig. 5-21, , Volume of parallelepiped = (height A)(area of parallelogram 7), = (A-n)(|BXC|), = A-{|BXC|n} = A - ( B X C ), If A, B and C do not form a right-handed system, A - n < 0 and the volume = |A • (B X C)|., , 5.20. If A = Aii + A2J + A3k, B = B¿ + B2j + Bsk, C = Cii + C2j + Cak show that, AI, , A-(BxC), , =, , Cl, , A-(BXC), , =, , AZ, , A3, , Bi B2 B3, C/2, , C/3, , i J k, A- B! BZ B3, Ci GZ C3, , = (Ati + A-j + A3k) • [(B2C3 - B3C2)i + (B3C, - B^s)] + (BjC2 - B^k], , AI AZ A3, = Aí(BíCs-BaCz)+Az(B3Cl-BlCí), , + AA(BlCí-B.iCl), , =, , Bl B2 B3, Ci C% Cg, , 5.21. Find the volume of a parallelepiped with sides A = 3i - j, B = j + 2k, C = i + 5j + 4k., 3-1 0, By Problems 5.19 and 5.20, volume of parallelepiped = |A • (B X C)| = | 0 1 2 |, 1 5 4, = |-20| = 20, , 5.22. Prove that A- (B x C) = (A x B) • C, i.e. the dot and cross can be interchanged., ¿1 ¿2 ¿3, , By Problem 5.20:, , A-(BXC) =, , B^ B2 B3 ,, , Cl, , (A X B) • C = C ' ( A X B ) =, , C*i C2 C3, , Since the two determinants are equal, the required result follows., , C2 C3, , A x Az A 3, BI B2 B3
Page 145 :
135, , VECTOR ANALYSIS, , CHAP. 5], , 5.23. Let ri = Xii + yij + Zik, r2 = xà + j/2j + z2k, and rs = xd + 2/sj + Zsk be the position, vectors of points PI(XI, y\, Zi), P2(£C2, y2, z2), and Ps(xs,ys,zs). Find an equation for, the plane passing through Pi,Pz and Pa., See Fig. 5-22., We assume that P1( P2 and P3 do not lie in, the same straight line; hence they determine, a plane., Let r = xi + yj + zk denote the position, vector of any point P(x, y, z) in the plane. Consider vectors PiP2 — r2 ~ r i> PI PS = r3 ~~ ri, and PjP = r — rx which all lie in the plane,, Thel1, , or, , (r - r x ) • (ra - r t ) X (i3 - r x ) = 0, , Fig. 5-22, , In terms of rectangular coordinates this becomes, [(* - *i)i + (V-, , Vl)j, , + (Z- 2j)k] • [(*2 - Xjl + (j/2 - 2/j)j + (Z2 - 2j)k], , X [(xa - xji + (2/3 - y j j + (z3 - «!)k] = 0, X — %!, , or, using Problem 5.20,, , *2-*l, , «3 — Xl, , y - 3/1, Vz-Vi, , 2/3-2/1, , Z-Z!, , =, , *2-«l, , 0, , Z3-Z,, , 5.24. Find an equation for the plane passing through the points Pi (3,1, — 2), P2(-l,2,4),, P,(2,-l,l)., The position vectors of P1( P2, P3 and any point P(x, y, z) on the plane are respectively, rj = 3i + j - 2k, r2 = -i + 2j + 4k, r3 = 2i - j + k, r = xi + y} + zk, Then P P j ^ r - r j , P2P1 = r2 —r,, P3P, =r3 — rlt all lie in the required plane and so the, required equation is (r — rj) • (r2 —1¡) X (rs — rt) = 0, i.e.,, {(*-3)i+ (2/-l)j + (z + 2)k}-{-4i + j + 6k> X {-i-2j + 3k> = 0, {(* - 3)i + (y- l)j + (z + 2)k> • {15i + 6j + 9k} = 0, , 1B(* - 3) + 6(j/ - 1) + 9(z + 2) = 0, Another method., , or, , 5x + 2y + 3z = 11, , By Problem 5.23, the required equation is, x-B y-1 2 + 2, -1-3 2-1, 4+ 2, 2 - 3 - 1 - 1 1+2, , =00, , or, , 5x + 2y + 3z = 11, , 5.25. If A = i + j, B = 2i-3j + k, C = 4j-3k, find (a) (AxB)xC, (6) A x ( B x C ) ., (a) A X B =, , i J k, 1 1 0, 2 - 3 1, , i, (6) B X C =, , J, , i J, k, = i - j - 5k. Then (A X B) X C = 1 -1 -5 = 23i + 3j + 4k., 0 4-3, , k, , 2 - 3 1, 0 4 - 3, , = 5i + 6j + 8k. Then A X (B X C) =, , It follows that, in general, (A X B) X C ¥- A X (B X C)., , i J k, 1 1 0, 5 6 8, , = 8i-8j + k.
Page 147 :
CHAP. 5], , VECTOR ANALYSIS, , 137, , Method 2., dA, , 5.30. A particle moves along a space curve r = r(f), where t is the time measured from, some initial time. If v = \dr/dt\ = ds/dt is the magnitude of the velocity of the, particle (s is the arc length along the space curve measured from the initial position),, prove that the acceleration a of the particle is given by, where T and N are unit tangent and normal vectors to the space curve and, , The velocity of the particle is given by v = vT. Then the acceleration is given by, , Since T has unit magnitude, we have T • T = 1. Then differentiating with respect to s,, , from which it follows that dT/ds is perpendicular to T. Denoting by N the unit vector in the direction of dT/ds, and called the principal normal to the space curve, we have, , where « is the magnitude of dT/ds., dT/ds = eZV<fe2. Hence, , Now since X = dr/ds, , [see equation (9), page 126], we have, , Defining p = 1/rc, (2) becomes dT/ds = N/p. Thus from (Í) we have, as required,, , The components d-v/dt and v2/p in the direction of T and N are called the tangential and normal, components of the acceleration, the latter being1 sometimes called the centripetal acceleration. The, quantities p and K are respectively the radius of curvature and curvature of the space curve., , GRADIENT, DIVERGENCE AND CURL, 5.31. If A = x2yz3 and A = xzi - y*j + 2x2yk, find, , (d) div
Page 148 :
iQQ, , VECTOR ANALYSIS, , [CHAP. 5, , (d) div(0A), , (e) curl (0A), , 5.32., , Prove, , 5.33., , Prove that, is a vector perpendicular to the surface ¿>(x, y, z) = c, where c is a, constant., Let r = xi + yj + zk be the position vector to any point P(x, y, z) on the surface., Then dr = dx i + dy j + dz k lies in the plane tangent to the surface at P. But, is perpendicular to ar and therefore to the surface., , 5.34. Find a unit normal to the surface 2x2 + 4yz - 5z* = -10, By Problem 5.33, a vector normal to the surface is, Then a unit normal to the surface at P is, Another unit normal to the surface at P is, , 5.35., , find, Laplacian of <f>, , and, , at the point P(B, —1,2).
Page 149 :
CHAP. 5], , VECTOR ANALYSIS, , 139, , Another method., , 5.36. Prove div curl A = 0., div curl A, , assuming that A has continuous second partial derivatives so that the order of differentiation is, immaterial., , 5.37. Find equations for (a) the tangent plane and (6) the normal line to the surface, F(x,y,z) = 0 at the point P(xo,y0,Zo). See Fig. 5-23., (a) A vector normal to the surface at P is N0 = VF\P. Then if r0 and r are the vectors drawn, respectively from O to P(x0, y0, z0) and Q(x,y,z) on the plane, the equation of the plane is, since r — r0 is perpendicular to N0. In rectangular form this is, (6) If r is the vector drawn from O in Pig. 5-23 to any point (x,y,e) on the normal line, then, r — r0 is collinear with N0 and so, , which in rectangular form is, , Pig. 5-23, , Fig. 5-24, , 5.38. Find equations for (a) the tangent line, (6) the normal plane to a space curve, * = f(u), y = g(u), z = h(u) at the point where u = u0. See Fig. 5-24.
Page 150 :
140, , VECTOR ANALYSIS, , [CHAP. 5, , (a) If R = f(u)i + g(u)j + h(u)k, then a vector tangent to the curve C at point P is given by, T0 = -T- . Then if r0 and r are the vectors drawn respectively from O to P and Q on the, tangent line, we have, ,, (r-r0)XT0 = (r-r0)xg p = 0, since r — r0 is collinear with T0. In rectangular form this becomes, , (6) If r is the vector from O to any point (x, y, z) on the normal plane, it follows that r — r0 is, perpendicular to T0. Then the required equation is, (r-r 0 )'T 0 = ( r - r, , or in rectangular form, , f'(uo)(x - x0) +ff'(u0)(y-, , Vo), , 0, , ) ' = 0, , + h'(u0)(z - z0) = 0, , 5.39. If F(x, y, z) is defined at any point on a space curve C and s is the arclength to (x, y, z), from some given point on C, show that, , V^'T, , =, , where T = dr/ds is a unit tangent vector to C at (x, y, z)., We have, , =, , VF'^- = VF'T, ds, The fact that T = dr/ds is a unit vector follows from, dF/ds the directional derivative of F at (x, y, z) along C., , dr2 = dx2 + dy2 + dz2., , We often call, , CURVILINEAR COORDINATES AND JACOBIANS, 5.40. Find ds2 in (a) cylindrical and (6) spherical coordinates and determine the scale, factors., (a) Method 1., dx, Then, ds2, , x = p cos <f>,, =, , — p sin 0 d0 + cos 0 dp,, , = dx2 + dy2 + dz2, 2, , 2, , =, , 2, , dy, , =, , z = z, , p cos 0 d<f> + sin 0 dp,, , dz = dz, , (-p sin 0 d<f> + cos <t> dp)2 + (p cos 0 d0 + sin 0 dp)2 + (dz)2, , =, 2, , + p (d0) + (dz), , (dp), , y — p sin <j>,, , = hi(dp)2 + ¿I(d0)2 + ¿|(dz)2, , and hi = hp = 1, h2 — h<l> = p, h3 = hz — 1 are the scale factors., Method 2., , The position vector is r = p cos 0 i + p sin 0 j + zk., dr, , Thus, , (b), , ds2 =, , Then, , =, =, , (cos 0 i + sin 0 j) dp + (— p sin 0 i + p cos 0 j) d0 + k dz, , =, , (cos 0 dp — p sin 0 d0)i + (sin 0 dp + p cos 0 d0)j + k dz, , dr • dr, , =, , (cos 0 dp — p sin 0 d0)2 + (sin 0 dp + p cos 0 d0)2 + (dz)2, , =, , (dp)2 + p2(d0)2 + (dz)2, , x — r sin 9 cos 0,, , y — r sin 9 sin 0,, , z = r cos 9
Page 151 : CHAP. 5], , VECTOR ANALYSIS, Then, , and, , 141, , dx = —r sin e sin <f> d<j> + r cos e cos <f> de + sin e cos 0 dr, dy = r sin e cos <j> d<f> + r cos « sin <j> de + sin 9 sin 0 dr, <fe = — r sin 9 dfl + cos 9 dr, (<fe)2 =, , (da;)2 + (<*2/)2 + (<fe)2 =, , (<fr)2 + rz(de)2 + r2 sin2 e (d<p)*, , The scale factors are hl = hr = 1, ¿2 = ¿e = r, ha = h^, = r sin #., , 5.41. Find the volume element dV in (a) cylindrical and (&) spherical coordinates and, sketch., The volume element in orthogonal curvilinear coordinates MI( uz, ua is, dV, , = h^hg dut du2 dua, , =, , d^ du2 dua, , (a) In cylindrical coordinates, Ui — p, M2 = 0» % = «, AI = 1, hz = p, h3 = 1 [see Problem 40(a)]., Then, , ciF = (l)(p)(l)dpd^d« =, pdpdfdz, This can also be observed directly from Fig. 5-25(a) below., , (a) Volume element in cylindrical coordinates. (6) Volume element in spherical coordinates., Fig. 5-25, , (6) In spherical coordinates, MI = r, u2 = e, u3 = <f>, h^ = 1, hz = r, h3 = r sin e, 5.40(6)]. Then, dV = (l)(r)(r sin e) dr de d<f> = r2 sin e dr de d<t>, , [see Problem, , This can also be observed directly from Fig. 5-25(6) above., , 5.42., , Express in cylindrical coordinates: (a) grad *, (&) divA, (c), Let M! = p, M2 = 0, w3 = z, fi! = 1, fe2 = p, A3 = 1 [see Problem 5.40(a)] in the results 1, 2,, page 128, and 4, page 129. Then, (a), , grad * =, where e1( e2, e3 are the unit vectors in the directions of increasing p, <f>, z respectively., , (6), , divA =, , =, , where A = A1e1 + A2e2 + A3es.
Page 153 :
CHAP. 5], , VECTOR ANALYSIS, , 143, , 5.47., , If A, B and C are non-coplanar vectors (vectors which do not all lie in the same plane) and, a^A + 2/jB + «tC = x2A + j/2B + Z2C, prove that necessarily Xi = x2, yt = yz, «i = ^2-, , 5.48., , Let ABCD be any quadrilateral and points P, Q, R and S the midpoints of successive sides. Prove, (a) that PQRS is a parallelogram and (6) that the perimeter of PQRS is equal to the sum of the, lengths of the diagonals of ABCD., , 5.49., , Prove that the medians of a triangle intersect at a point which is a trisection point of each median., , 5.50., , Find a unit vector in the direction of the resultant of vectors, C = 3i - 2j + 4k., , A = 2i — j + k, B = i + j + 2k,, , THE DOT OR SCALAR PRODUCT, 5.51. Evaluate |(A + B) • (A - B)| if A = 2i - 3j + 5k and B = 3i + j - 2k., 5.52., , Prove the law of cosines for a triangle., use C - C = (A-B)-(A-B).], , [Hint. Take the sides as A, B, C where C = A — B. Then, , 5.53., , Find a so that 2i — 3j + 5k and 3i + oj — 2k are perpendicular., , 5.54., , If A = 2i + j + k, B = i - 2j + 2k and C = 3i - 4j + 2k, find the projection of A + C in the, direction of B., , 5.55., , A triangle has vertices at A(Z, 3,1), B(—l, 1,2), C(l, —2,3). Find (a) the length of the median drawn, from B to side AC and (6) the acute angle which this median makes with side BC., , 5.56., , Prove that the diagonals of a rhombus are perpendicular to each other., , 5.57., , Prove that the vector (AB + BA)/(A + B) represents the bisector of the angle between A and B., , THE CROSS OR VECTOR PRODUCT, 5.58. If A = 2i-j + k and B - i + 2j-3k, find |(2A + B) X (A- 2B)|., 5.59., , Find a unit vector perpendicular to the plane of the vectors A = 3i — 2j 4- 4k and B = i + j — 2k., , 5.60., , If A X B = A X C, does B = C necessarily?, , 5.61., , Find the area of the triangle with vertices (2, -3,1), (1, -1,2), (-1, 2, 3)., , 5.62., , Find the shortest distance from the point (3,2,1) to the plane determined by (1,1,0), (3, —1,1),, (-1,0,2)., , TRIPLE PRODUCTS, 5.63. If A = 2i + j-3k, B = i-2j + k, C = -i + j-4k, find (a) A • (B X C), (6) C • (A X B), (c) A X, (B X C), (d) (A X B) X C., 5.64., , Prove that, , (a) A • (B X C) = B • (C X A) = C • {A X B), (6) A X (B X C) = B(A • C) - C(A • B)., , 5.65., , Find an equation for the plane passing through (2, —1, —2), (—1,2, —3), (4,1,0)., , 5.66., , Find the volume of the tetrahedron with vertices at (2,1,1), (1, -1, 2), (0,1, -1), (1, -2,1)., , 5.67., , Prove that (A X B) • (C X D) + (B X C) • (A X D) + (C X A) • (B X D) = 0., , DERIVATIVES, 5.68. A particle moves along the space curve r = e~* cos t i + e~* sin t j + e-*k. Find the magnitude, of the (a) velocity and (6) acceleration at any time t., 5.69., , Prove that, , where A and B are differentiable functions of u.
Page 154 :
144, , VECTOR ANALYSIS, , [CHAP. 5, , 5.70., , Find a unit vector tangent to the space curve * = t, y = t2, z = i3 at the point where í = 1., , 5.71., , If r = a cos at + b sin at, where a and b are any constant noncollinear vectors and « is a constant, scalar, prove that (a) r X, , 5.72., , If A = X2i -yj + xzk, B = yi + xj- xyzk, (6) d[A • (B X C)] at the point (1, -1,2)., , and C = i - yj + xszk, find (, , 5.73., , If R, , 5.74., , If A is a differentiate function of u and |A(tt)| = 1, prove that dA/du is perpendicular to A., , 5.75., , Let T and N denote respectively the unit tangent vector and unit principal normal vector to a space, curve r = i(u), where r(u) is assumed differentiate. Define a vector B = T X N called the unit, binormal vector to the space curve. Prove that, , a, , ), , a, , n, , d, , at the point, , These are called the Frenet-Serret formulas. In these formulas K is called the curvature, T is called, the torsion; and the reciprocals of these, p — l//c and a = I/T, are called the radius of curvature, and radius of torsion respectively., GRADIENT, DIVERGENCE AND CURL, 5.76. If U, V, A, B have continuous partial derivatives prove that:, (a), 5.77., , If <t> = xy + yz + zx and A = x*yi + y2zj + z2xk, find, at the point (3, —1,2)., , 5.78., , Show that, , 5.79., , Prove:, , 5.80., , Prove that curl grad U = 0, stating appropriate conditions on U., , 5.81., , Find a unit normal to the surface xzy — 2xz + 2?/2z4 = 10 at the point (2,1, —1)., , 5.82., , If, , 5.83., , (a) Prove that ', Problem 5.82., , 5.84., , Find the equations of the (a) tangent plane and (b) normal line to the surface x2 + y2 = 4z at, (2,-4, 5)., , 5.85., , Find the equations of the (a) tangent line and (6) normal plane to the space curve x = 6 sin t,, y = 4 cos 3t, z = 2 sin 5t at the point where t = ir/4., , 5.86., , (a) Find the directional derivative of U = 2xy — z2 at (2, —1,1) in a direction toward (3,1, —1)., (6) In what direction is the directional derivative a maximum? (c) What is the value of this, maximum?, , 5.87., , Prove that the acute angle y between the z axis and the normal to the surface F(x, y, z) = 0 at any, point is given by sec y =, , 5.88., , (a) Develop a formula for the shortest distance from a point (x0, y0, z0) to a surface. (6) Illustrate, the result in (a) by finding the shortest distance from the point (1,1, —2) to the surface z = xz + yz., , = 0 where r = xi + yj + zk and r = |r|., , A = 3xz*i - yzj + (x + 2z)k, find curl curl A., (6) Verify the result in (a) if A is given as in
Page 155 :
CHAP. 5], 5.89., , VECTOR ANALYSIS, , 145, , Let E and H be two vectors assumed to have continuous partial derivatives (of second order at least), with respect to position and time. Suppose further that E and H satisfy the equations, U), , prove that E and H satisfy the equation, , (2), , [The vectors E and H are called electric and magnetic field vectors in electromagnetic theory. Equations (1) are a special case of Maxwell's equations. The result (2) led Maxwell to the conclusion, that light was an electromagnetic phenomena. The constant c is the velocity of light.], 5.90., , Use the relations in Problem 5.89 to show that, , JACOBIANS AND CURVILINEAR COORDINATES, 5.91., , Prove that, , 5.92., , Express (a) grad *, (6) div A in spherical coordinates., , 5.93., , The transformation from rectangular to parabolic cylindrical coordinates is defined by the equations, x = £(w.2 — vz), y = uv, z = z. (a) Prove that the system is orthogonal. (6) Find ds2 and the scale, factors, (c) Find the Jacobian of the transformation and the volume element., , 5.94., , Write (a) V2* and (6) div A in parabolic cylindrical coordinates., , 5.95., , Prove that for orthogonal curvilinear coordinates,, V* =, , [Hint. Let V* = ct1e1 + <t2e2 + a3es and use the fact that d<I> = V* • dr must be the same in both, rectangular and the curvilinear coordinates.], 5.96., , Prove that the acceleration of a particle along a space curve is given respectively in (a) cylindrical,, (6) spherical coordinates by, (p- P^)ep + (p* + 2¿¿fe» + '¿9., ( V — rez — rj>2 sin2 e)er + (r 'ê + 2re — rj>2 sin e cos 0)ee + (2f 0 sin e + 2rèj> cos e + r 0 sin e)e^,, where dots denote time derivatives and ep, e^, ez, er, ee, e,,, are unit vectors in the directions of, increasing p, 0, z, r, 9, 0 respectively., , 5.97., , If F = x + 3y2 - zs, G - 2x2yz, and H = 2z2 — xy, evaluate, , 5.98., , If F = xy + yz + zx, G = a;2 + j/2 + z2, and H = x + y + z, determine whether there is a functional, relationship connecting F, G, and H, and if so find it., , 5.99., , If F(P, V, T) = 0, prove that (a), , at (1, -1,0)., , = -1, , where a, , subscript indicates the variable which is to be held constant. These results are useful in thermodynamics where P, V, T correspond to pressure, volume and temperature of a physical system., 5.10.0, d(u'v'w) * °' ^ Give an interPretation of the result of (a) in terms of transformations.
Page 156 :
146, , VECTOR ANALYSIS, , Answers to Supplementary Problems, 5.45., , 5.50., 5.51., 5.53., 5.54., 5.55., 5.58., 5.59., 5.61., 5,62., 5.63., 5.65., 5.66., 5.68., 5.70., 5.72., 5.73., 5.77., 5.81., 5.82., 5.84., 5.85., 5.86., 5.92., , 5.93., , 5.94., , 5.97., 5.98., , 33.6 miles, 13.2° north of east, , [CHAP. 5
Page 157 :
Chapter 6, , Multiple, Line and Surface, Integrals and Integral Theorems, , DOUBLE INTEGRALS, Let F(x, y) be defined in a closed region ^ of the, xy plane (see Fig. 6-1). Subdivide <3?. into n subregions A"^ of area &Ak, k = 1,2,...,n. Let (¿k,%), be some point of A<R fc . Form the sum, (1), Consider, (*), , where the limit is taken so that the number n of subdivisions increases without limit and such that the, largest linear dimension of each A^fc approaches, zero. If this limit exists it is denoted by, , Fig. 6-1, , (3), , and is called the double integral of F(x, y) over the region "5?.., It can be proved that the limit does exist if F(x, y) is continuous (or piecewise continuous) in <%., ITERATED INTEGRALS, If "^ is such that any lines parallel to the y axis meet the boundary of ^ in at most, two points (as is true in Fig. 6-1), then we can write the equations of the curves ACB, and ADB bounding ^ as y = fi(x) and y = f-¿(x) respectively, where fi(x) and fz(x) are, single-valued and continuous in a ë x à b. In this case we can evaluate the double, integral (3) by choosing the regions A^fc as rectangles formed by constructing a grid of, lines parallel to the x and y axes and AA k as the corresponding areas. Then (3) can be, written, (4), , where the integral in braces is to be evaluated first (keeping x constant) and finally, integrating with respect to x from a to b. The result (4) indicates how a double integral, can be evaluated by expressing it in terms of two single integrals called iterated integrals., , 147
Page 158 :
148, , MULTIPLE, LINE AND SURFACE INTEGRALS. INTEGRAL THEOREMS, , [CHAP. 6, , If ^ is such that any lines parallel to the x axis meet the boundary of <% in at most, two points (as in Fig. 6-1), then the equations of curves CAD and CBD can be written, x = gi(y) and x = g2(y) respectively and we find similarly, (5), , If the double integral exists, (4) and (5) will in general yield the same value. In writing a, double integral, either of the forms (4) or (5), whichever is appropriate, may be used. We, call one form an interchange of the order of integration with respect to the other form., In case <^ is not of the type shown in the above figure, it can generally be subdivided, into regions "3^, <R,2, ... which are of this type. Then the double integral over ^ is found, by taking the sum of the double integrals over "3^, ^^ ...., , TRIPLE INTEGRALS, , The above results are easily generalized to closed regions in three dimensions. For, example, consider a function F(x,y,z) defined in a closed three dimensional region "tR.., Subdivide the region into n subrogions of volume AFfc, k = 1,2,..., n. Letting (ffc, i)k, £fc), be some point in each subregion, we form, («), , where the number n of subdivisions approaches infinity in such a way that the largest, linear dimension of each subregion approaches zero. If this limit exists we denote it by, (7), , called the triple integral of F(x, y, z) over <R. The limit does exist if F(x, y, z) is continuous (or piecewise continuous) in <]{., If we construct a grid consisting of planes parallel to the xy, yz and xz planes, the, region ^ is subdivided into subrogions which are rectangular parallelepipeds. In such, case we can express the triple integral over ^ given by (7) as an iterated integral of the, form, (*), , (where the innermost integral is to be evaluated first) or the sum of such integrals. The, integration can also be performed in any other order to give an equivalent result., Extensions to higher dimensions are also possible., , TRANSFORMATIONS OF MULTIPLE INTEGRALS, In evaluating a multiple integral over a region <?(, it is often convenient to use, coordinates other than rectangular, such as the curvilinear coordinates considered in, Chapter 5.
Page 159 :
CHAP. 6], , MULTIPLE, LINE AND SURFACE INTEGRALS. INTEGRAL THEOREMS, , 149, , If we let (u,v) be curvilinear coordinates of points in a plane, there will be a set of, transformation equations x = f(u, v), y = g(u, v) mapping points (x, y) of the xy plane, into points (u,v) of the uv plane. In such case the region <3( of the xy plane is mapped, into a region <^' of the uv plane. We then have, , (9), where G(u,v) = F{f(u,v), g(u,v)}, , and, , (10), is the Jacobian of x and y with respect to u and v (see Chapter 5)., Similarly if (u, v,w) are curvilinear coordinates in three dimensions, there will be, a set of transformation equations x = f(u, v, w), y = g(u, v, w), z = h(u, v, w) and we can, write, , du dv dw, , (11), , where, , (12), , is the Jacobian of x, y and z with respect to u, v and w., The results (9) and (11) correspond to change of variables for double and triple integrals., Generalizations to higher dimensions are easily made., LINE INTEGRALS, Let C be a curve in the xy plane which, connects points A(ai,61) and B(a2,b2), (see, Fig. 6-2). Let P(x,y) and Q(x,y) be singlevalued functions defined at all points of C., Subdivide C into n parts by choosing (n — 1), points on it given by (xi,yi), (x2,y2), .. .,(*»-!,, yn-i). Call AXfc = Xk — xk-i and Ayk = yk-yk-\,, k = 1,2, ..., n where (oi, bi) = (x0, y0), (a2, b2) =, (xn,yn) and suppose that points (lfc,vk) are, chosen so that they are situated on C between, points (Xk-i, 2/fc-i) and (XH, yk). Form the sum, n, , 2) W* vj **fc + Q(ft, Vt) Ai/fc}, , B(a,, b,), "(#,,+,,J4+i), , Pig. 6-2, , (15), , The limit of this sum as n -» « in such a way that all the quantities Axk, A^/fc approach, zero, if such limit exists, is called a line integral along C and is denoted by, or, , (14)
Page 160 :
150, , MULTIPLE, LINE AND SURFACE INTEGRALS. INTEGRAL THEOREMS, , [CHAP. 6, , The limit does exist if P and Q are continuous (or piecewise continuous) at all points of C., The value of the integral depends in general on P, Q, the particular curve C, and on the, limits (ai, &i) and (a2, ba)., In an exactly analogous manner one may define a line integral along a curve C in, three dimensional space as, (15), , where A\, A2 and A3 are functions of x, y and z., Other types of line integrals, depending on particular curves, can be defined. For, example, if Ask denotes the arc length along curve C in the above figure between points, (xk,yk) and (xk+i,yk+i), then, (16), is called the line integral of U(x,y) along curve C., dimensions are possible., , Extensions to three (or higher), , VECTOR NOTATION FOR LINE INTEGRALS, It is often convenient to express a line integral in vector form as an aid in physical, or geometric understanding as well as for brevity of notation. For example, we can, express the line integral (15) in the form, (17), , where A = Aii + A2j + Ask and dr = dxi + dyj + dzk., special case of this with z — 0., , The line integral (14) is a, , If at each point (x, y, z) we associate a force F acting on an object (i.e. if a force field, is defined), then, (18), represents physically the total work done in moving the object along the curve C., EVALUATION OF LINE INTEGRALS, If the equation of a curve C in the plane z — 0 is given as y = f(x), the line integral, (14) is evaluated by placing y - f(x), dy = f'(x) dx in the integrand to obtain the definite, integral, , (19), which is then evaluated in the usual manner., Similarly if C is given as x = g(y), then dx = g'(y) dy and the line integral becomes, (20)
Page 161 :
CHAP. 6], , MULTIPLE, LINE AND SURFACE INTEGRALS. INTEGRAL THEOREMS, , 151, , If C is given in parametric form x = $(t), y - <¡,(t), the line integral becomes, , (21), where U and ¿2 denote the values of t corresponding to points A and B respectively., Combinations of the above methods may be used in the evaluation., Similar methods are used for evaluating line integrals along space curves., , PROPERTIES OF LINE INTEGRALS, Line integrals have properties which are analogous to those of ordinary integrals., For example:, , Thus reversal of the path of integration changes the sign of the line integral., 3., , where (a3, b3) is another point on C., Similar properties hold for line integrals in space., , SIMPLE CLOSED CURVES. SIMPLY AND MULTIPLY-CONNECTED REGIONS, A simple closed curve is a closed curve which does not intersect itself anywhere., Mathematically, a curve in the xy plane is defined by the parametric equations x = <£(i),, y = $(t) where <£ and <l> are single-valued and continuous in an interval íi á t a Í2. If, 4>(ti) = <t>(ta) and ^(ii) = ^(£2), the curve is said to be closed. If <¡>(u) = <¡>(v) and <¡/(u) = *¡>(v), only when u = v (except in the special case where u = íi and v = ¿2), the curve is closed, and does not intersect itself and so is a simple closed curve. We shall also assume, unless, otherwise stated, that <j> and ^ are piecewise differentiate in íi a t á Í2., If a plane region has the property that any closed curve in it can be continuously, shrunk to a point without leaving the region, then the region is called simply-connected,, otherwise it is called multiply-connected [see Problem 6.19]., As the parameter t varies from ti to tz, the plane curve is described in a certain sense, or direction. For curves in the xy plane, we arbitrarily describe this direction as positive, or negative according as a person traversing the curve in this direction with his head, pointing in the positive z direction has the region enclosed by the curve always toward his, left or right respectively. If we look down upon a simple closed curve in the xy plane, this, amounts to saying that traversal of the curve in the counterclockwise direction is taken, as positive while traversal in the clockwise direction is taken as negative., , GREEN'S THEOREM IN THE PLANE, Let P,Q,dP/dy,dQ/dx be single-valued and continuous in a simply-connected region, ^ bounded by a simple closed curve C. Then, , (22)
Page 162 :
152, , MULTIPLE, LINE AND SURFACE INTEGRALS. INTEGRAL THEOREMS, , [CHAP. 6, , where, is used to emphasize that C is closed and that it is described in the positive, direction"., This theorem is also true for regions bounded by two or more closed curves (i.e., multiply-connected regions). See Problem 6.19., CONDITIONS FOR A LINE INTEGRAL TO BE INDEPENDENT OF THE PATH, Theorem 6-1., , A necessary and sufficient condition for, to be independent, of the path C joining any two given points in a region ^ is that in ^, dP/dy = dQ/dx, , (23), , where it is supposed that these partial derivatives are continuous in ^., The condition (23) is also the condition that Pdx + Qdy is an exact differential,, i.e. that there exists a function <f>(x,y) such that Pdx + Qdy = d^>. In such case if the, end points of curve C are (x\, yi) and (xz, y¿), the value of the line integral is given by, (24), , In particular if (23) holds and C is closed, we have Xi = x-2, y\ — yz and, , (25), For proofs and related theorems, see Problems 6.22 and 6.23., The results in Theorem 6-1 can be extended to line integrals in space. Thus we have, Theorem 6-2. A necessary a n d sufficient condition f, , *., , o, , r, , t, , o be, , independent of the path C joining any two given points in a region ^ is, that in ^, (26), where it is supposed that these partial derivatives are continuous in "•£., The results can be expressed concisely in terms of vectors., the line integral can be written, , If A = Aii + A2j + Ask,, , A-dr and condition (26) is equivalent to the condition, , If A represents a force field F which acts on an object, the result is equivalent to the statement that the work done in moving the object from one point to another, is independent of the path joining the two points if and only if, Such a force, field is often called conservative., The condition (26) [or the equivalent condition, is also the condition that, Aidx + Azdy + Asdz [or A'dr] is an exact differential, i.e. that there exists a function, <j>(x,y,z) such that Aidx + A2dy + A3dz = d<j>. In such case if the endpoints of curve C, are (xi,yi,zi) and (#2,2/2,22), the value of the line integral is given by, , (27), In particular if C is closed and, , we have, , (28)
Page 163 :
CHAP. 6], , MULTIPLE, LINE AND SURFACE INTEGRALS. INTEGRAL THEOREMS, , SURFACE INTEGRALS, Let S be a two-sided surface having, projection 'R. on the xy plane as in the, adjoining Fig. 6-3. Assume that an, equation for S is z = f(x, y), where / is, single-valued and continuous for all x, and y in ••£. Divide <K into n subregions, of area AAP, p = 1,2,...,«, and erect, a vertical column on each of these subregions to intersect S in an area ASP., Let <j)(x,y,z) be single-valued and, continuous at all points of S. Form, the sum, (29), , k, , 153, , nn, , Fig. 6-3, , where (£ ,vp,£p) is some point of ASp. If the limit of this sum as »-» « in such a way, that each ASp -» 0 exists, the resulting limit is called the surface integral of $(x, y, z) over, S and is designated by, (30), Since ASp = |sec yp AA p approximately, where yp is the angle between the normal line, to S and the positive z axis, the limit of the sum (29) can be written, (31), , The quantity |sec y| is given by, , |sec y|, , =, , (32), , Then assuming that z = f(x, y) has continuous (or sectionally continuous) derivatives in, •3^, (31) can be written in rectangular form as, , In case the equation for S is given as F(x, y, z) = 0, (33) can also be written, (34), , The results (33) or (34) can be used to evaluate (30)., In the above we have assumed that S is such that any line parallel to the z axis, intersects S in only one point. In case S is not of this type, we can usually subdivide S, into surfaces Si,S2, :.. which are of this type. Then the surface integral over S is defined, as the sum of the surface integrals over Si, S 2 ,... ., The results stated hold when S is projected on to a region ^ of the xy plane. In some, cases it is better to project S on to the yz or xz planes. For such cases (30) can be, evaluated by appropriately modifying (33) and (34).
Page 164 :
154, , MULTIPLE, LINE AND SURFACE INTEGRALS. INTEGRAL THEOREMS, , [CHAP. 6, , THE DIVERGENCE THEOREM, , Let S be a closed surface bounding a region of volume V. Choose the outward drawn, normal to the surface as the -positive normal and assume that a, p, y are the angles which, this normal makes with the positive x, y and z axes respectively. Then if At, A2 and A3 are, continuous and have continuous partial derivatives in the region, (35), , which can also be written, (36), , In vector form with A = Aa +A 2 j + Ask and n = cos ai + cos /Jj + cos yk, these can be, simply written as, (37), , In words this theorem, called the divergence theorem or Green's theorem in space, states, that the surface integral of the normal component of a vector A taken over a closed, surface is equal to the integral of the divergence of A taken over the volume enclosed, by the surface., STOKES' THEOREM, , Let S be an open, two-sided surface bounded by a closed non-intersecting curve C, (simple closed curve). Consider a directed line normal to S as positive if it is on one side, of S, and negative if it is on the other side of S. The choice of which side is positive is, arbitrary but should be decided upon in advance. Call the direction or sense of C positive if an observer, walking on the boundary of <S with his head pointing in the direction, of the positive normal, has the surface on his left. Then if Ai,A2, A3 are single-valued,, continuous, and have continuous first partial derivatives in a region of space including S,, we have, (38), , In vector form with A = Aii + A 2 j + A3k and n = cos ai + cos/îj + cos-yk, this is simply, expressed as, (39), In words this theorem, called Stokes' theorem, states, tangential component of a vector A taken around a simple, surface integral of the normal component of the curl of, having C as a boundary. Note that if, as a special case, the result (28)., , that the line integral of the, closed curve C is equal to the, A taken over any surface S, V x A = 0 in (39), we obtain
Page 165 :
CHAP. 6], , MULTIPLE, LINE AND SURFACE INTEGRALS. INTEGRAL THEOREMS, , 155, , Solved Problems, DOUBLE INTEGRALS, , 6.1., , (a) Sketch the region 9?. in the xy plane bounded by y = x2, x = 2, y = 1., (&) Give a physical interpretation to, (c) Evaluate the double integral in (&)., (a) The required region ^ is shown shaded in Fig. 6-4 below., (b) Since x2 + y2 is the square of the distance from any point (x, y) to (0,0), we can consider the, double integral as representing the polar moment of inertia (i.e. moment of inertia with respect, to the origin) of the region ^. (assuming unit density)., We can also consider the double integral as representing the mass of the region "3{ assuming, a density varying as x2 + y2., , Fig. 6-4, (c) Method 1., , Fig. 6-5, , The double integral can be expressed as the iterated integral, , The integration with respect to y (keeping x constant) from y — 1 to y — x2 corresponds, formally to summing in a vertical column (see Fig. 6-4). The subsequent integration with, respect to x from x = 1 to x = 2 corresponds to addition of contributions from all such, vertical columns between x — 1 and x = 2., Method 2., , The double integral can also be expressed as the iterated integral, , In this case the vertical column of region ^ in Fig. 6-4 above is replaced by a horizontal, column as in Fig. 6-5 above. Then the integration with respect to x (keeping y constant) from, x = i/y to x = 2 corresponds to summing in this horizontal column. Subsequent integration, with respect to y from y - Í to y = 4 corresponds to addition of contributions for all such, horizontal columns between y = 1 and y = 4.
Page 166 :
156, 6.2., , MULTIPLE, LINE AND SURFACE INTEGRALS. INTEGRAL THEOREMS, , [CHAP. 6, , Find, the volume, of the region common to the intersecting cylinders xz + y2 = a2 and, 2, 2, 2, , x +z = a ., , Required volume = 8 times volume of region shown in Fig. 6-6, , As an aid in setting up this integral note that z dy áx corresponds to the volume of a column, such as shown darkly shaded in the figure. Keeping x constant and integrating with respect to y, from y = 0 to y = ^az — x^ corresponds to adding the volumes of all such columns in a slab, parallel to the yz plane, thus giving the volume of this slab. Finally, integrating with respect to, * from x = 0 to x = a, corresponds to adding the volumes of all such slabs in the region, thus, giving the required volume., , Fig. 6-6, , Fig. 6_7, , TRIPLE INTEGRALS, 6.3., , (a) Sketch the 3 dimensional region ^ bounded by x + y + z = a(a,>0), x = 0 y = Q, 2 = 0., , ', , (&) Give a physical interpretation to, , (c) Evaluate the triple integral in (6)., (a) The required region ^ is shown in Fig. 6-7., (6) Since x2 + y* + 32 is the square of the distance from any point (x, y, z) to (0, 0,0), we can consider, the triple integral as representing the polar moment of inertia (i.e. moment of inertia with, respect to the origin) of the region ^ (assuming unit density)., We can also consider the triple integral as representing the mass of the region if the, density varies as a2 + j/2 + z2., (c) The triple integral can be expressed as the iterated integral
Page 167 :
CHAP. 6], , MULTIPLE, LINE AND SURFACE INTEGRALS. INTEGRAL THEOREMS, , 157, , The integration with respect to z (keeping * and y constant) from z = 0 to z = a-x-y, corresponds to summing the polar moments of inertia (or masses) corresponding to each cube, in a vertical column. The subsequent integration with respect to y from y = 0 to y = a — x, (keeping * constant) corresponds to addition of contributions from all vertical columns contained, in a slab parallel to the yz plane. Finally, integration with respect to x from * = 0 to x = a, adds up contributions from all slabs parallel to the yz plane., Although the above integration has been accomplished in the order z, y, x, any other order, is clearly possible and the final answer should be the same., , 6.4., , Find the (a) volume and (b) centroid of the region <5? bounded bv the narahnlip, cylinder z = 4-x ¿ and the planes x = Q, y = 0, y = Q,z^Q assuming the density, to be a constant a., The region ^ is shown in Fig. 6-8., (a) Required volume, , Fig. 6-8, (6) Total mass, , by part (a), since a is constant. Then, , Total moment about yz plane, Total mass, Total moment about xz plane, Total mass, Total moment about xy plane, Total mass, Aims me centróla ñas coordinates (3/4,3,8/5)., Note that the value for y could have been predicted because of symmetry.
Page 168 :
158, , MULTIPLE, LINE AND SURFACE INTEGRALS. INTEGRAL THEOREMS, , [CHAP. 6, , TRANSFORMATION OF DOUBLE INTEGRALS, 6.5. Justify equation (21), page 151, for changing variables in a double integral., In rectangular coordinates, the double integral, oí' F(x,y) over the region ^ (shaded in Fig. 6-9) is, We can also evaluate this double, integral by considering a grid formed by a family, of u and v curvilinear coordinate curves constructed, on the region ••£ as shown in the figure., Let P be any point with coordinates (x, y) or, (u,v), where x = f(u,v) and y = g(u,v). Then, the vector r from O to P is given by r = xi + yj =, f(u, v)i + g(u, v)j. The tangent vectors to the coordinate curves u — ct and v = cz, where Cj and C2 are, constants, are dr/dv and dr/du respectively. Then, the area of the region A^ of Fig. 6-9 is given approximately by, AwAt>., , Fig. 6-9, , But, , so that, The double integral is the limit of the sum, taken over the entire region "3Ç. An investigation reveals that this limit is, , •\., , where <^' is the region in the uv plane into which the region "^ is mapped under the transformation, x = f(u, v), y = g(u, v)., , 6.6., , Evaluate, , £2 + 3,2 = 4 and x2 + y2 = 9., , where 'K. is the region in the xy plane bounded by, , The presence of xz + y2 suggests the use of polar coordinates (p, 0), where » = p cos <f>, y = p sin <f>., Under this transformation the region "^ [Fig. 6-10(a)] is mapped into the region <3(' [Fig. 6-10(6)].
Page 169 :
CHAP. 6], , MULTIPLE, LINE AND SURFACE INTEGRALS. INTEGRAL THEOREMS, , Since, , 159, , = p, it follows that, , We can also write the integration limits for ^' immediately on observing the region % since, for fixed <j>, p varies from p = 2 to p = 3 within the sector shown dashed in Fig. 6-10(a). An, integration with respect to <j> from <j> = 0 to <f> = 2ir then gives the contribution from all sectors., Geometrically pdpd<f> represents the area dA as shown in Fig. 6-10(a)., , TRANSFORMATION OF TRIPLE INTEGRALS, , 6.7., , Justify equation (11), page 149, for changing variables in a triple integral., , Fig. 6-11, By analogy with Problem 6.5 we construct a grid of curvilinear coordinate surfaces which subdivide the region "^ into subregions, a typical one of which is A^ (see Fig. 6-11)., The vector r from the origin O to point P is, , r = xi + yj + zk = f(u, v, w)\ + g(u, v, w)j + h(u, v, w)k, assuming that the transformation equations are x = f(u, v, w), y = g(u, v, w) and z = h(u, v, w)., Tangent vectors to the coordinate curves corresponding to the intersection of pairs of coordinate, surfaces are given by dr/du, dr/dv, dr/dw. Then the volume of the region A"^ of Fig. 6-11 is given, approximately by, , The triple integral of F(x, y, z) over the region is the limit of the sum, , An investigation reveals that this limit is, , where 9?' is the region in the uvw space into which the region <% is mapped under the transformation.
Page 170 : 160, , 6.8., , MULTIPLE, LINE AND SURFACE INTEGRALS. INTEGRAL THEOREMS, , [CHAP. 6, , E x p r «e s s \ d x d y d z in cylindrical coordinates., The transformation equations in cylindrical coordinates are, The Jacobian of the transformation is, , x = p cos <f>, y = p sin <f>, z — z., , Then by Problem 6.7 the triple integral becomes, , where ^' is the region in the p, <p, z space corresponding to "•£ and where, , G(p, </>, «) =, , 6.9., , F(p cos 0, p sin <f>, z), , Find the volume of the region above the, xy plane bounded by the paraboloid, z = x2 + y2 and the cylinder xz + y2 - a2., The volume is most easily found by using, cylindrical coordinates. In these coordinates the, equations for the paraboloid and cylinder are respectively z = p2 and p = a. Then, Required volume, =, , 4 times volume shown in Pig. 6-12, , Fig. 6-12, The integration with respect to z (keeping p and <f> constant) from z = 0 to z = p2 corresponds, to summing the cubical volumes (indicated by dV) in a vertical column extending from the xy plane, to the paraboloid. The subsequent integration with respect to p (keeping <t> constant) from p = 0, to p = a corresponds to addition of volumes of all columns in the wedge shaped region. Finally,, integration with respect to <f> corresponds to adding volumes of all such wedge shaped regions., The integration can also be performed in other orders to yield the same result., We can also set up the integral by determining the region ^' in p, $, z space into which ^ is, mapped by the cylindrical coordinate transformation., , LINE INTEGRALS, 5.10. Evaluate, , along (a) a straight line from (0,1) to (1,2),, , (6) straight lines from (0,1) to (1,1) and then from (1,1) to (1,2), (c) the parabola, x = t, y = t2 + l., (a) An equation for the line joining (0,1) and (1,2) in the xy plane is y = x + l. Then dy = dx, and the line integral equals
Page 171 :
CHAP. 6], , MULTIPLE, LINE AND SURFACE INTEGRALS. INTEGRAL THEOREMS, , 161, , (b) Along the straight line from (0,1) to (1,1), y = 1, dy = 0 and the line integral equals, , Along the straight line from (1,1) to (1,2), x = 1, dx = 0 and the line integral equals, , Then the required value = -2/3 +10/3 = 8/3., (c) Since í = 0 at (0,1) and f = 1 at (1, 2), the line integral equals, , 6.11., , If A = (3a; 2 -6ye)i + (2y + 3xz)j + (1 -4xyz 2 )k, evaluate, k-dr from (0,0,0) to, (1,1,1) along the following paths C:, (a) x-t, y = t2, z = t3., (b) the straight lines from (0,0,0) to (0,0,1), then to (0,1,1), and then to (1,1,1)., (c) the straight line joining (0,0,0) and (1,1,1)., , (a) If x = t, y = t2, z = t3, points (0, 0,0) and (1,1,1) correspond to t = 0 and í = 1 respectively., Then, , Another method., Along C, A = (3 «2 - 6i«)i + (2«2 + 3t*)j + (1 - 4i")k and r = x\ + yj + zk = tí. + t*j + t%,, dr = (i + 2ij + 3f%) dt. Then, , (b) Along the straight line from (0,0,0) to (0,0,1), x = O, y = 0, dx - 0, dy = 0 while z varies, from 0 to 1. Then the integral over this part of the path is, , Along the straight line from (0, 0,1) to (0,1,1), x = 0, z - 1, dx = 0, dz = 0 while y, varies from 0 to 1. Then the integral over this part of the path is, , Along the straight line from (0,1,1) to (1,1,1), y = 1, z = 1, dy = 0, dz = 0, x varies from 0 to 1. Then the integral over this part of the path is, , Adding,, , while
Page 172 :
162, , MULTIPLE, LINE AND SURFACE INTEGRALS. INTEGRAL THEOREMS, , [CHAP. 6, , (c) The straight line joining (0,0,0) and (1,1,1) is given in parametric form by * = t, y = t, z = t., Then, , 6.12. Find the work done in moving a particle once around an ellipse C in the xy plane,, if the ellipse has center at the origin with semi-major and semi-minor axes 4' and 3, respectively, as indicated in Fig. 6-13, and if the force field is given by, F = (3x-4y + 2z)i + (4x + 2y- 3z2) j + (2xz - 4y* + «3)k, In the plane 2 = 0, F = (3« - 4j/)i + (4* + 2y)j - 4î/2k and, dr = dxi + dyj so that the work done is, , Choose the parametric equations of the ellipse as x = 4 cos t,, y = 3 sin í where t varies from 0 to 2ir (see Fig. 6-13). Then the, line integral equals, , Fig. 6-13, , In traversing C we have chosen the counterclockwise direction indicated in Fig. 6-13. We call, this the positive direction, or say that C has been traversed in the positive sense. If C were traversed, in the clockwise (negative) direction the value of the integral would be — 96ir., , 6.13. Evaluate, , y ds along the curve C given by, , Since, , GREEN'S THEOREM IN THE PLANE, 6.14. Prove Green's theorem in the plane if C, is a closed curve which has the property, that any straight line parallel to the coordinate axes cuts C in at most two points., Let the equations of the curves AEB and AFB, (see adjoining Fig. 6-14) be y - Fj (a;) and, y = Y2(x) respectively. If <% is the region bounded, by C, we have, , from x = 3 to x = 24., we have
Page 173 :
CHAP. 6], , MULTIPLE, LINE AND SURFACE INTEGRALS. INTEGRAL THEOREMS, , 163, (1), , Then, Similarly let the equations of curves EAF and EBF be x - X^y), Then, , and x = X2(y), , Then, , respectively., , <«>, , Adding (1) and (2),, , dx dy., , 6.15. Verify Green's theorem in the plane for, where C is the closed curve of the region, bounded by y = x2 and y2 = x., The plane curves y = x2 and y2 = * intersect at (0,0) and (1,1). The positive direction in, traversing C is as shown in Pig. 6-15., Along y = x2, the line integral equals, , Fig. 6-15, , Along yz = x the line integral equals, , Then the required line integral = 7/6 - 17/15 = 1/30., , Hence Green's theorem is verified., , 6.16. Extend the proof of Green's theorem in the, plane given in Problem 6.14 to the curves, C for which lines parallel to the coordinate, axes may cut C in more than two points., Consider a closed curve C such as shown in, the adjoining Fig. 6-16, in which lines parallel to, the axes may meet C in more than two points. By, constructing line ST the region is divided into two, regions "^ and ^2 which are of the type considered in Problem 6.14 and for which Green's, theorem applies, i.e.,, , p¡g. 6-16
Page 174 :
164, , MULTIPLE, LINE AND SURFACE INTEGRALS. INTEGRAL THEOREMS, , [CHAP. 6, , (1), S', , Adding the left hand sides of (Í) and (2), we have, omitting the integrand P dx + Q dy in each, case,, , using the fact that, Adding the right hand sides of (1) and (2), omitting the integrand,, where "^ consists of regions "^ and ^2Then, , d x d y, , a n d t h e t h e o r e m i s p r o v e d ., , A region "3J such as considered here for which any closed curve lying in "3^ can be continuously, shrunk to a point without leaving <ï{, is called a simply-connected region. A region which is not, simply-connected is called multiply-connected. We have shown here that Green's theorem in the, plane applies to simply-connected regions bounded by closed curves. In Problem 6.19 the theorem, is extended to multiply-connected regions., For more complicated simply-connected regions it may be necessary to construct more lines,, such as ST, to establish the theorem., , 6.17. Show that the area bounded by a simple closed curve C is given by, In Green's theorem, put P = —y, Q = x., , Then, , where A is the required area. Thus ¡, , 6.18. Find the area of the ellipse x — a cos 0, y = b sin 0., Area, , [(a cos e)(b cos e) de — (b sin »)(—a sin e) de], , 6.19. Show that Green's theorem in the plane is, also valid for a multiply-connected region "3Ç., such as shown in Fig. 6-17., The shaded region *R_, shown in the figure, is, multiply-connected since not every closed curve lying, in ^ can be shrunk to a point without leaving <£, as, is observed by considering a curve surrounding, DEFGD for example. The boundary of ^, which, consists of the exterior boundary AHJKLA and the, interior boundary DEFGD, is to be traversed in the, positive direction, so that a person traveling in this, direction always has the region on his left. It is, seen that the positive directions are those indicated, in the adjoining figure., figure., , Fig. 6-17, , In order to establish the theorem, construct a line, such as AD, called a cross-cut, connecting, the exterior and interior boundaries. The region bounded by ADEFGDALKJHA is simply-connected, and so Green's theorem is valid. Then
Page 175 :
CHAP. 6], , MULTIPLE, LINE AND SURFACE INTEGRALS. INTEGRAL THEOREMS, , 165, , But the integral on the left, leaving out the integrand, is equal to, , since, , . Thus if Ct is the curve ALKJHA, C2 is the curve DEFGD and C is the, , boundary of "3? consisting of C1 and C2 (traversed in the positive directions), then, and so, , -, , INDEPENDENCE OF THE PATH, 6.20. Let P(x, y) and Q(x, y) be continuous and have continuous first partial derivatives at, each point of a simply connected region ^. Prove that a necessary and sufficient, condition that, , around every closed path C in ^ is that, , dP/dy = dQ/dx identically in ^., Sufficiency., , Suppose dP/dy = dQ/dx., , Then by Green's theorem,, , "A., , where ^ is the region bounded by C., Necessity., Suppose, , around every closed path C in ^ and that dP/dy ¥= dQ/dx, , at, , some point of % In particular suppose dP/dy — dQ/dx > 0 at the point (o;0, j/0)., By hypothesis dP/dy and dQ/dx are continuous in "3J, so that there must be some region r containing (x0, y0) as an interior point for which dP/dy — dQ/dx > 0. If r is the boundary of T,, then by Green's theorem, , contradicting t h e hypothesis that h, a, t, f, o, r a l l closed curves i n 9{.. Thus, dQ/dx — dP/dy cannot be positive., Similarly we can show that dQ/dx — dP/dy cannot be negative, and it follows that it must be, identically zero, i.e. dP/dy = dQ/dx identically in <%., , 6.21. Let P and Q be defined as in Problem 6.20. Prove that a necessary and sufficient condition that, , [Pdx + Qdy] be independent of the path in <% joining points A and B, , is that dP/dy - dQ/dx identically in <•£., Sufficiency., , If dP/dy = dQ/dx, then by Problem 6.20,, , (see Fig. 6-18). From this, omitting for brevity the integrand, Pdx + Qdy, we have, and so, , i.e. the integral is independent of the path.
Page 176 :
166, , MULTIPLE, LINE AND SURFACE INTEGRALS. INTEGRAL THEOREMS, , 6.22. (a) Prove that, , [CHAP. 6, , is independent of the path, , joining (1,2) and (3,4). (&) Evaluate the integral in (a)., (a) P = 6xy2 -y3, Q = 6x2y - 3xy2. Then dP/dy = 12xy - 3y2 = dQ/dx, line integral is independent of the path., , and by Problem 6.21 the, , (6) Method 1., , Since the line integral is independent of the path, choose any path joining (1,2) and (3,4),, for example that consisting of lines from (1,2) to (3,2) [along which y = 2, dy = 0] and then, (3,2) to (3,4) [along which * = 3, dx = 0]. Then the required integral equals, , Method 2., , Since ^ = ||, we must have (1) |, | = Qxy2 - y3, (2) |^ = 6x*y - 3xy2., From (1), <t> = 3x2y2 — xy3 + f(y). From (2), <f> = 3x2y2 — xy3 + g(x). The only way in, which these two expressions for <f> are equal is if f(y) = g(x) = c, a constant. Hence $ =, SxV - xya + c. Then, , Note that in this evaluation the arbitrary constant c can be omitted., We could also have noted by inspection that, (6xy2 - y3) dx + (6x2y - 3xy2) dy = (6xy2 dx + Qx2y dy) - (y3 dx + 3xy2 dy), = d(3x2y2) - d(xy3) = d(3x2y2 - xy3), from which it is clear that <t> = 3x2y2 — xy3 + c., , [(x2y cos x + 2xy sin x — y2ex) dx + (x2 sin x — 2yex) dy] around the, , 6.23. Evaluate, z/s, , pocycloid x, , 23, , hy-, , 23, , +y ' = a ' ., , 2, , P = x y cos x + 2xy sin x — yzex, Q — x2 sin x — 2yex., Then dP/dy = x2 cos x + 2x sin x — 2yex — dQ/dx, so that by Problem 6.20 the line integral, around any closed path, in particular x2/s + y2'3 = a2'3, is zero., , SURFACE INTEGRALS, 6.24. If y is the angle between the normal line to any point (x, y, z) of a surface S and the, positive z axis, prove that, |secy|, , =, , according as the equation for S is z = f(x, y) or F(x, y, z) = 0., If the equation of S is F(x,y,z) - 0, a normal to S at (x,y,z) is ' F = Fxi + Fvj + Fzk., Then, F-k = |VF| |k| cosy, or, Fz —, cosy, from which, , |secy|. =, , as required., , In case the equation is « = f(x, y), we can write F(x, y, z) = z — f(x, y) = 0, from which, Fx = —zx, Fy = — zy, Fz= 1 and we find |see y| =
Page 177 :
CHAP. 6], , MULTIPLE, LINE AND SURFACE INTEGRALS. INTEGRAL THEOREMS, , 6.25. Evaluate, , 167, , U(x, y, z) dS where S is the surface of the paraboloid z = 2 - (x2 + y2), , above the xy plane and U(x, y, z) is equal to (a) 1, (b) x2 + y2, (c) 3z. Give a physical, interpretation in each case., The required integral is equal to, U), , where ^ is the projection of S on the xy plane, given by «2 + y2 = 2, z = 0., Since zx = -2x, zy = ~2y, (1) can be, written, , (2), (a) If U(x, y, z) = 1, (2) becomes, , To evaluate this, transform to polar coordinates (p, <f>). Then the integral becomes, , Fig. 6-19, , Physically this could represent the surface area of S, or the mass of 5 assuming unit, density., (6) If, , U(x, y, z)=x* + 2/2, (2) becomes, , dy or in polar coordinates, , where the integration with respect to p is accomplished by the substitution, Physically this could represent the moment of inertia of S about the z axis assuming unit, density, or the mass of S assuming a density = x2 + y2., (c) If U(x, y, z) = 82, (2) becomes, , or in polar coordinates,, , Physically this could represent the mass of S assuming a density = 3z, or three times the, first moment of S about the xy plane., , 6.26. Find the surface area of a hemisphere of, radius a cut off by a cylinder having this, radius as diameter., Equations for the hemisphere and cylinder (see, Fig. 6-20) are given respectively by x2 + yz +, z2 = a2 (or z =, and (* - a/2)2 +, 2, 2, z, y - a /4 (or x + y* = ax)., Since, , zx =, wehave, , and, Fig. 6-20
Page 178 :
168, , MULTIPLE, LINE AND SURFACE INTEGRALS. INTEGRAL THEOREMS, , Required surface area, , [CHAP. 6, , =, , Two methods of evaluation are possible., Method 1. Using polar coordinates., Since x2 + y2 = ax in polar coordinates is p — a cos <f>, the integral becomes, , Method 2., , The integral is equal to, , Letting x = a tan2 e, this integral becomes, , Note that the above integrals are improper and should actually be treated by appropriate, limiting procedures., , 6.27. Find the centroid of the surface in Problem 6.25., By symmetry, x = y — 0 and, , The numerator and denominator can be obtained from the results of Problem 6.25(c) and, „. ., „ ,, ., ,, ., 377T/10 111, 6.25(a) respectively, and we Athus have z = .„ .„ = rñ¿-, , 6.28., , Evaluate, , where A - xyi- x2j + (x + z)k, S is that portion of the, , plane 2x + 2y + z = 6 included in the first octant, and n is a unit normal to S., A normal, and so, , The required surface integral is therefore, , Fig. 6-21
Page 179 :
CHAP. 6], , 6.29., , MULTIPLE, LINE AND SURFACE INTEGRALS. INTEGRAL THEOREMS, , 169, , In dealing with surface integrals we have restricted ourselves to surfaces which are, two-sided. Give an example of a surface which is not two-sided., Take a strip of paper such as ABCD as, shown in the adjoining Fig. 6-22. Twist the, strip so that points A and B fall on D and C, respectively, as in the adjoining figure. If n, is the positive normal at point P of the surface, we find that as n moves around the, surface it reverses its original direction when, it reaches P again. If we tried to color only, one side of the surface we would find the, whole thing colored. This surface, called a, Moebius strip, is an example of a one-sided, surface. This is sometimes called a nonorientable surface. A two-sided surface is, orientable., , Fig. 6-22, , THE DIVERGENCE THEOREM, , 6.30., , Prove the divergence theorem., , Fig. 6-23, Let S be a closed surface which is such that any line parallel to the coordinate axes cuts S in at, most two points. Assume the equations of the lower and upper portions, Sj and S2, to be z = f1 (x, y), and « = fz(x, y) respectively. Denote the projection of the surface on the xy plane by ^., Consider, , ///£" - ///£*** - if[Cl£*>*, *, -, , JJ As(x,y,z), dydx =, JJ[Aa(x,y,f2)-Aa(x,y,fi)]dydx, V., ', ', 91, For the upper portion S2, dy dx = cos y2 dS2 = k • n2 dS2 since the normal n2 to S2 makes an, acute angle j2 with k.
Page 180 :
170, , MULTIPLE, LINE AND SURFACE INTEGRALS. INTEGRAL THEOREMS, For the lower portion S1( dy dx = -cos y1 dSt = —k • n¡ dS1, an obtuse angle y t with k., , [CHAP. 6, , since the normal HJ to Sj makes, , Then, , and, , so that, , /¿\, , Similarly, by projecting S on the other coordinate planes,, (2), , (3), Adding (Í), (3) and (a),, , or, , The theorem can be extended to surfaces which are such that lines parallel to the coordinate, axes meet them in more than two points. To establish this extension, subdivide the region bounded, by S into subregions whose surfaces do satisfy this condition. The procedure is analogous to that, used in Green's theorem for the plane., , 6.31. Verify the divergence theorem for A = (2x- z)i + x2yj - xz2k taken over the region, bounded by x = 0, x = 1, y = O, y = 1, z = 0, z - 1., We first e v a l ui a t e d S where S is the, surface of the cube in Fig. 6-24., Face DEFG: n = i, x = 1., , FaceABCO: n =-i, * = 0., , Then, , Then, , Fig. 6.24
Page 181 :
CHAP. 6], , MULTIPLE, LINE AND SURFACE INTEGRALS., , Face ABEF: n = j, y = 1., , Face OGDC: n = -j, y = 0., , Face BCDE: n = k, z = 1., , INTEGRAL THEOREMS, , 171, , Then, , Then, , Then, , Face AFGO: n = -k, z = 0., , Then, , Adding,, , the divergence theorem is verified in this case., , 6.32. Evaluate, , • n dS, where S is a closed surface., , By the divergence theorem,, , where V is the volume enclosed by S., , 6.33. vevaluate «, where S is the entire, s, surface of the hemispherical region bounded by z —, and z — 0 (a) by, the divergence theorem (Green's theorem in space), (b) directly., (a) Since, , dy dz = dS cos a, dz dx = dS cos ¡3, dx dy = dS cos y,, , where A - xz*i + (xty — zs)j + (2xy + yzz)k, ward drawn unit normal., , the integral can be written, , and n = (cos a) i + (cos /J)j + (cosy)k, the out-, , Then by the divergence theorem the integral equals, , where V is the region bounded by the hemisphere and the xy plane.
Page 182 :
172, , MULTIPLE, LINE AND SURFACE INTEGRALS. INTEGRAL THEOREMS, , [CHAP. 6, , By use of spherical coordinates, this integral is equal to, , (6) If Si is the convex surface of the hemispherical region and S2 is the base (z = 0), then, , By addition of the above, we obtain, , i, , Since by symmetry all these integrals are equal, the result is, on using polar coordinates,, , STOKES' THEOREM, , 6.34. Prove Stokes' theorem., Let S be a surface which is such that its, projections on the xy, yz and <KZ planes are, regions bounded by simple closed curves, as indicated in Fig. 6-25. Assume S to have representation z = f(x, y) or x — g(y, z) or y = h(x, z),, where /, g, h are single-valued, continuous and, differentiable functions. We must show that, , where C is the boundary of S., Consider first, , Fig. 6-25
Page 183 :
CHAP. 6], , MULTIPLE, LINE AND SURFACE INTEGRALS. INTEGRAL THEOREMS, , 173, , Since, , (Í), , If z = f(x, y) is taken as the equation of S, then the position vector to any point of S is r = xi +, yj + 2k = xi + yj + f(x, y)k so that J^ = j + |^k = J + f ^ k - B«t f^ is a vector tangent to S, and thus perpendicular to n, so that, , Substitute in (1) to obtain, , or, , (2), , Now on S, A j (x, y,z) = A^x, y,f(x,y}} = F(x,y)\ hence, , and (2) becomes, , Then, where 9( is the projection of S on the xy plane. By Green's theorem for the plane the last integral, equals, , F dx where T is the boundary of i(. Since at each point (x, y) of F the value of F is the, , same as the value of A í at each point (x, y, z) of C, and since dx is the same for both curves, we, must have, , or, , Similarly, by projections on the other coordinate planes,, , Thus by addition,, , The theorem is also valid for surfaces S which may not satisfy the restrictions imposed above., For assume that S can be subdivided into surfaces Slt S2, ..., Sk with boundaries Clt C2, ..., Ck, which do satisfy the restrictions. Then Stokes' theorem holds for each such surface. Adding, these surface integrals, the total surface integral over S is obtained. Adding the corresponding line, integrals over C1( C2, ..., Ck, the line integral over C is obtained.
Page 184 :
174, , MULTIPLE, LINE AND SURFACE INTEGRALS. INTEGRAL THEOREMS, , [CHAP. 6, , 6.35. Verify Stokes' theorem for A = 3yi - xzj + yz*k, where S is the surface of the, paraboloid 2z = x2 + y2 bounded by 2 = 2 and C is, its boundary., The boundary C of S is a circle with equations *2 + j/2 — 4,, z = 2 and parametric equations x = 2 cos t, y — 2 sin t, z = 2,, where 0 S t < 2v. Then, , Also,, , Fig. 6-26, , and, Then, , In polar coordinates this becomes, , 6.36. Prove that a necessary and sufficient condition t, curve C is that, identically., Sufficiency., , h, , a, , t, , f, , o, , r every closed, , Then by Stokes1 theorem, , Suppose, , Necessity., Suppose, , aroun, , d ^ery closed path C, and assume, , at some point P., , Then assuming, , is continuous there will be a region with P as an interior point, where, be, a surface contained, this region, *«»*•, • a C ine contained, in this «**<» who*e normal n at each point has the same, direction as, where, is, a, postive, constant. Let C be the boundary of S., Then by Stokes' theorem, , which contradicts the hypothesis that, It follows that, , and shows that, , is also a necessary and sufficient condition for a line integral, , dr to be independent of the path joining points Pj and P2.
Page 185 :
CHAP. 6], 6.37., , MULTIPLE, LINE AND SURFACE INTEGRALS. INTEGRAL THEOREMS, , Prove that a necessary and sufficient condition that, Sufficiency., , , then, , Necessity., If, , 175, , is that, , by Problem 5.80, page 144., , then by Problem 6.36,, , around every closed path and ^, , A-dr, , is independent of the path joining two points which we take as (a, b, c) and (x, y, z). Let us define, , Then, , Since the last integral is independent of the path joining (x, y, z) and (x + AÍK, y, z), we can, choose the path to be a straight line joining these points so that dy and dz are zero. Then, , where we have applied the law of the mean for integrals., Taking the limit of both sides as Aa; -» 0 gives d<j>/dx — A t., Similarly we can show that d<j>ldy — A2, d<j>ldz = A3., Thus A = A t i + A 2 j + A3k =, , = V0., , 6.38. (a) Prove that a necessary and sufficient condition that Aidx + Azdy + Asdz = d$,, an exact differential, is that, where A = Aii + Azj + Ajk., (b) Show thatun such case,, , (a) Necessity., , If, , AI dx + A2dy + A3dz = d<f> •, , then, , Then by differentiating we have, assuming continuity of the partial derivatives,, , which is precisely the condition V X A = 0., Another method., , If Ajda; + A2dy + Asdz = d<j>, then, A = Ají + A 2 j + A3k =, , =, , V0, , from which, Sufficiency., , If, , A1dx + A2dy + A3dz, (b) Prom part (a), <j>(x, y, z) -, , then by Problem 6.37,, = A «cur =, , V0 • dr =, , [A1 dx + A 2 dy + A3 dz]., , and, = ¿4,
Page 186 :
176, , MULTIPLE, LINE AND SURFACE INTEGRALS. INTEGRAL THEOREMS, , [CHAP. 6, , Then omitting the integrand A1dx + A2dy + A3dz, we have, , 6.39. (a) Prove that F = (2xz5 + 6y)i + (Qx - 2yz)j + (3x2z2 - y2)k is a conservative force, field. (&) Evaluate, , F • dr where C is any path from (1, -1,1) to (2,1, -1)., , (c) Give, , a physical interpretation of the results., (a) A force field F is conservative if the line integral JI F • dr is independent of the path C joining, c, any two points. A necessary and sufficient condition that F be conservative is that V X F = 0., , Since here, , =0,, , F is conservative., , (6) Method 1., By Problem 6.38, F • dr = (2xz3 + 6y) dx + (Gx - 2yz) dy + (Bx2z2 - y2) dz is an exact differential d<j>, where <f> is such that, , From these we obtain respectively, <6 = xW + Gxy + f^y.z), , 0 = 6xy - y*z + f2(x, z), z, , <j> = x*z* - y*z + f s ( x , y ), , 2 3, , These are consistent if f1 (y, z) = —y z + c, /2 (x, z) = x z + c, fs (x, y) = 6xy + c in which case, <f> = X2z3 + Gxy - y2z + c. Thus by Problem 6.38,, , Alternatively we may notice by inspection that, F • dr, , =, , (2xz3 dx + 3x*z2 dz) + (6y dx + 6x dy) - (2yz dy + yz dz), , =, , d(x2zs) + d(6xy) - d(y2z), , = d(x2zs + 6xy - y2z + c), , from which 0 is determined., Method 2., Since the integral is independent of the path, we can choose any path to evaluate it; in, particular we can choose the path consisting of straight lines from (1, —1,1) to (2, —1,1), then, to (2,1,1) and then to (2,1, -1). The result is, , where the first integral is obtained from the line integral by placing y = —1, z = 1, dy = 0,, dz — 0; the second integral by placing x = 2, z = 1, dx = 0, dz — 0; and the third integral by, placing x — 2, y — I, dx — 0, dy — 0., (c) Physically, , F • dr represents the work done in moving an object from (1, —1,1) to (2,1, —1), , along C. In a conservative force field the work done is independent of the path C joining these, points.
Page 187 :
CHAP. 6], , MULTIPLE, LINE AND SURFACE INTEGRALS. INTEGRAL THEOREMS, , 177, , Supplementary Problems, DOUBLE INTEGRALS, 6.40. («) Sketch the region ^ in the xy plane bounded by y2 = 2x and y = x., (c) Find the polar moment of inertia of "^ assuming constant density a., 6.41., , Find the centroid of the region in the preceding problem., , 6.42., , Given, , dxdy., , (6) Find the area of 9£., , (a) Sketch the region and give a possible physical interpretation, , of the double integral. (6) Interchange the order of integration, (c) Evaluate the double integral., 6.43., , Show that, , 6.44., , Find the volume of the tetrahedron bounded by x/a + y/b + z/c — 1 and the coordinate planes., , 6.45., , Find the volume of the region bounded by z = x2 + y2, z = 0, x = —a, x — a, y = —a, y = a., , 6.46., , Find (a) the moment of inertia about the z axis and (6) the centroid of the region in Problem 6.45, assuming a constant density a., , TRIPLE INTEGRALS, 6.47., , (a) E v, in (a)., , a, , l, , u, , a, , t, , e, , 6.48., , Find the (a) volume and (6) centroid of the region in the first octant bounded by x/a + y/b + z/c = 1,, where a, b, c are positive., , 6.49., , Find the (a) moment of inertia and, Problem 6.48., , 6.50., , Find the mass of the region corresponding to x1 + y2 + z2 =i 4, x a O, y = 0, z = 0, if the density, is equal to xyz., , 6.51., , Find the volume of the region bounded by z = x2 + y2 and z = 2x., , x, , y, , z, , dz dy dx. (V) Give a physical interpretation to the integral, , (6) radius of gyration about the z axis of the region in, , TRANSFORMATION OF DOUBLE INTEGRALS, dx dy, where ^ is the region x2 + y2 S a2., , 6.52., , Evaluate, , 6.53., , If "^ is the region of Problem 6.52, evaluate, , 6.54., , By using the transformation x + y = u, y = uv, show that, , 6.55., , Find the area of the region bounded by xy = 4, xy = 8, xy3 — 5, xy3 = 15. [Hint. Let xy — u,, xy3 = v.], , 6.56., , Show that the volume generated by revolving the region in the first quadrant bounded by the, parabolas y2 = x, y2 = 8x, x2 = y, x2 = 8y about the * axis is 279W2. [Hint. Let y2 = ux,, x2 = vy.], , 6.57., , Find the area of the region in the first quadrant bounded by y = x3, y = 4a;3, * = y3, x — íy3., , 6.58., , Let ^ be the region bounded by x + y = 1, x = 0, y = 0. Show that, t, h, [Hint. Let x — y = u, x + y = v.], , dx dy., , a, , t, , TRANSFORMATION OF TRIPLE INTEGRALS, 6.59. Find the volume of the region bounded by z = 4 — x2 — y2 and the xy plane., 6.60., , Find the centroid of the region in Problem 6.59, assuming constant density a., , d, , x dy — sin 1.
Page 188 :
178, , 6.61., , MULTIPLE, LINE AND SURFACE INTEGRALS. INTEGRAL THEOREMS, , (a) Evaluate, , [CHAP. 6, , dx dy dz, where ^ is the region bounded by the plane z = 3 and, , the cone z =, , (6) Give a physical interpretation of the integral in (a). [Hint. Perform the, integration in cylindrical coordinates in the order p, z, $.], 6.62., , Show that the volume of the region bounded by the cone, z - x2 + y2 is jT/6., , z=, , and the paraboloid, , 6.63., , Find the moment of inertia of a right circular cylinder of radius a and height b, about its axis if, the density is proportional to the distance from the axis., , 6.64., , (a) Evaluate, , , where "^ is the region bounded by the spheres *2 + y2 + z2 = a2, , and x2 + y2 + z2 — b2 where a > b > 0. (6) Give a physical interpretation of the integral in (a)., 6.65., , (a) Find the volume of the region bounded above by the sphere r = 2a cos 0, and below by the cone, 0 = a where 0 < a < ir/2. (b) Discuss the case a — ir/2., , 6.66., , Find the centroid of a hemispherical shell having outer radius a and inner radius b if the density, (a) is constant, (6) varies as the square of the distance from the base. Discuss the case a — b., , LINE INTEGRALS, 6.67., , Evaluate, , (a) the parabola y2 = x,, , [(a; + y) dx + (y — x) dy] along, , (6) a straight line,, , (c) straight lines from (1,1) to (1,2) and then to (4,2), (d) the curve * = 2*2 + t + 1, y = í2 + 1., 6.68., , Evaluate, , [(2x — y + 4)dx + (5y + 3* — 6) dy] around a triangle in the xy plane with vertices at, , (0,0), (3,0), (3,2) traversed in a counterclockwise direction., 6.69., , Evaluate the line integral in the preceding problem around a circle of radius 4 with center at (0,0)., , 6.70., , (a) If, , F = (x2 — y2)i + 2xyj,, , evaluate, , • dr along the curve C in the xy plane given by, , 2, , y = x — x from the point (1,0) to (2,2). (6) Interpret physically the result obtained., 6.71., , Evaluate, , (2x + y) ds, where C is the curve in the xy plane given by x2 + y2 = 25 and s is the, , arc length parameter, from the point (3, 4) to (4,3) along the shortest path., 6.72., , If F = (3o; - 2y)i + (y + 2z)j - x2k, evaluate, 2, , F • dr from (0,0,0) to (1,1,1), where C is a path, 3, , consisting of (a) the curve x = t, y = t , z = t , (b) a straight line joining these points, (c) the, straight lines from (0,0,0) to (0,1,0), then to (0,1,1) and then to (1,1,1), (d) the curve x = z2,, z = 2/2., 6.73., , If T is the unit tangent vector to a curve C (plane or space curve) and F is a given force field, prove, that under appropriate conditions, , F "Ids, , where s is the arc length parameter., , Interpret the result physically and geometrically., , GREEN'S THEOREM IN THE PLANE. INDEPENDENCE OF THE PATH, 6.74., , Verify Green's theorem i n t h e plane f, vertices at (0,0), (2,0), (2,2), (0,2)., , o, , r, , w, , h, , e, , r, , e C i s a square with, , 6.75., , Evaluate the line integrals of (a) Problem 6.68 and (b) Problem 6.69 by Green's theorem.
Page 189 :
CHAP. 6], 6.76., , MULTIPLE, LINE AND SURFACE INTEGRALS. INTEGRAL THEOREMS, , (a) Let C be any simple closed curve bounding a region having area A., &i> &2> &s are constants,, , 179, , Prove that if a,, a2, a3(, , (5) Under what conditions will the line integral around any path C be zero?, 6.77., , Find the area bounded by the hypocycloid x2/a + y2'3 = o2/3., [Hint. Parametric equations are x = a cos3 t, y = a sin3 i, 0 a t S 2ir.], , 6.78., , If * = p cos <j>, y = p sin <t>, prove that, , 6.79., , Verify Green's theorem in the plane f, 2, , 2, , p2 <Z0 and interpret., where C is the boundary of the, 2, , 2, , region enclosed by the circles x + y = 4 and x + y — 16., 6.80., , (a) Prove that, , is independent of the path joining (1,0), •^(1,0), , and (2,1). (6) Evaluate the integral in (a)., 6.81., , Evaluate, , Zx = iry2, , along the parabola, , from, , (0,0)to(»/2,l)., 6.82., , Evaluate the line integral in the preceding problem around a parallelogram with vertices at (0,0),, (3,0), (5,2), (2,2)., , 6.83., , Prove that if x — f ( u , v), y — g(u, v) defines a transformation which maps a region "^ of the, xy plane into a region <3f of the uv plane then, , by using Green's theorem on the integral, , and interpret geometrically., , SURFACE INTEGRALS, 6.84., , (a) Evaluate, , where S is the surface of the cone z2 = 3(x2 + y2) bounded by z = 0, , and z = 3. (6) Interpret physically the result in (a)., 6.85., , Determine the surface area of the plane 2x + y + 2z = 16 cut oif by (a) x - 0, y - 0, x - 2, y — 3,, (6) * = 0, y = 0 and x2 + y2 = 64., , 6.86., , Find the surface area of the paraboloid 2z = x2 + y2 which is outside the cone z =, , 6.87., , Find the area of the surface of the cone z2 - S(x2 + y2) cut out by the paraboloid z = x2 + y2., , 6.88., , Find the surface area of the region common to the intersecting cylinders, 2, , 2, , 2, , x +z = a ., , 6.89., , x2 + y2 = a2 and, , (a) Show that in general the equation r = r(u, v) geometrically represents a surface. (6) Discuss, the geometric significance of u = clt v = cz where Cj and c2 are constants, (c) Prove that the, element of arc length on this surface is given by, ds2 - Edu2 + 2Fdudv + Gdv2, where E =, , 6.90., , (a) Referring to Problem 6.89, show that the element of surface area is given by dS =, (b) Deduce from (a.) that the area of a surface r = T(U, v) is, [Hint., , Use the fact that, , (A X B) • (C X D) = (A • C)(B • D) - (A • D)(B • C)., , du dv., , lu dv., and then use the identity
Page 190 :
180, 6.91., , MULTIPLE, LINE AND SURFACE INTEGRALS. INTEGRAL THEOREMS, , [CHAP. 6, , (a) Prove that r = (a sin u cos v)i + ct(sin u sin v)j + (a cos u)k, 0 S uë v, 0 ^ v < 2w represents a, sphere of radius o. (6) Use Problem 6.90 to show that the surface area of this sphere is 4ira2., , THE DIVERGENCE THEOREM, 6.92., Verify the divergence theorem for A = (2xy + z)i + y2j — (x + 3j/)k taken over the region bounded, by 2x + Zy + z - 6, * = O, y - 0, z = 0., 6.93., , F • n dS, where F = (z2 - x)i - xyj + Szk, , Evaluate, , and S is the surface of the region, , 2, , bounded by z = 4 — y , x = 0, x = 3 and the xy plane., 6.94., , A • n dS, where A = (2* + 3z)i - (xz + y)j + (y2 + 2z)k and S is the surface of the, , Evaluate, , sphere having center at (3,—1,2) and radius 3., 6.95., , Determine the value of, , [xdydz + ydzdx + zdxdy], where S is the surface of the region, , bounded by the cylinder x2 + y2 = 9 and the planes z = 0 and z = 3, (a) by using the divergence, theorem, (6) directly., 6.96., , [4a;z dy dz — y2 dz dx + yz dx dy], where S is the surface of the cube bounded by, , Evaluate, , * = O, y = 0, z = 0, * = 1, y = 1, z = 1, (a) directly, (6) by Green's theorem in space (divergence, theorem)., 6.97., , Prove that, , for any closed surface S., s, , 6.98., , Prove that, , where n is the outward drawn normal to any closed surface S., , 6.99., , If n is the unit outward drawn normal to any closed surface S bounding the region V, prove that, , STOKES' THEOREM, 6.100. Verify Stokes' theorem for A = 2yi + Sxj — z2k, where S is the upper half surface of the sphere, x2 + y2 + z2 = 9 and C is its boundary., 6.101. Verify Stokes' theorem for A = (y + z)i — xzj + y2k, where S is the surface of the region in the, first octant bounded by 2x + z = 6 and y = 2 which is not included in the (a) xy plane, (6) plane, y = 2, (e) plane 2x + » = 6 and C is the corresponding boundary., • n dS, where A = (x-z)i + (x3 + yz)j - 3*2/2k and S is the surface of the, , 6.102. Evaluate, o, , cone 2 = 2 —, , above the xy plane., , 6.103. If y is a region bounded by a closed surface S and, , prove t h a t B • n dS = 0., , 6.104. (a) Prove that F = (2xy + 3)i + (a;2 - 4z)j - 4j/k is a conservative force field. (b) Find <f> such, that F = V0. (c) Evaluate, , F • di, where C is any path from (3,-1,2) to (2,1,-1)., , 6.105. Let C be any path joining any point on the sphere x2 + y2 + z2 = a2 to any point on the sphere, x2 + y2 + z2 = b2. Show that if F = 5r3 r, where r = xi + yj + zk, then, 6.106. In Problem 6.105 evaluate, , F • dr = 65 - a5., , F • dr if F = /(r)r, where /(r) is assumed to be continuous.
Page 191 :
CHAP. 6], , MULTIPLE, LINE AND SURFACE INTEGRALS. INTEGRAL THEOREMS, , 181, , 6.107. Determine whether there is a function <f> such that F = V0, where:, (a) F = (xz- y)i + (x2y + z3)j + (3xz* - xy)k., (b) F = 2xe—»i + (cos z — x2e~y)j — (y sin z)k. If so, find it., 6.108. Solve the differential equation (z« - 4xy) dx + (6y - 2x2) dy + (3xzz + 1) dz = 0., , Answers to Supplementary Problems, 6.40., , (6) 2/3;, , (c) 48<r/35 = 72M/35, where M, is the mass of ^., , 6.41., , * = 4/5, y = l, , 6.42., , (b], , 6.44., , abc/Q, , 6.45., , 8aV3, , 6.46., , (a) ^ae<r = ff-Ma2, where M = mass;, , (x + y) dy dx,, , (c) 241/60, , (6) « = y = 0, Z = ¿a2, , 6.70., , (a) 124/15, , 6.71., , 15, , 6.72., , (a) 23/15, (6) 5/3, (c) 0, (d) 13/30, , 6.74., , Common value = 8, , 6.76., , (6) o2 = 6t, , 6.77., , 3jro2/8, , 6.79., , Common value = 120n-, , 6.80., , (6)5, , 6.47., , (a) 3/8, , 6.81., , ÎT2/4, , 6.48., , (a) abc/6; (b) x - a/4, y - 6/4, z = c/4, , 6.82., , 0, , 6.49., , (a) M(a2 + &2)/10, (6) V(«2 + &2)/10, , 6.84., , (a) Sir, , 6.50., , 4/3, , 6.85., , (a) 9, (6) 24ir, , 6.51., , ÎT/2, , 6.86., , §a-(5VB-l), , 6.52., , frt», , 6.87., , 6>r, , 6.53., , 7r(l - e-°2), , 6.88., , 16a2, , 6.55., , 2 In 3, , 6.92., , Common value = 27, , 6.57., , 1/8, , 6.93., , 16, , 6.59., , %TT, , 6.94., , 10877-, , 6.60., , x = y = 0, ¿ = |, , 6.95., , 81)7, , 6.61., , 27^(2\/2-l)/2, , 6.96., , 3/2, , 6.63., , fMa2, , 6.100. Common value = 9ir, , 6.64., , (a) 4s- In (a/6), , 6.101. The common value is, (a) -6, (6) -9, (c) -18, , 6.65., , |jra3(l - COS4 a), , 6.66., , Taking the z axis as axis of symmetry:, (a) x - y = 0, z = f (a4 - 64)/(a3 - 6s);, , 6.104. (6) $ = x2y — 4yz + Sx + constant,, , (6) « = y = 0, z - f (a6 - &6)/(a5 - &5), , 6.106., , 6.102. 12;r, , r f(r) dr, , 6.67., , (a) 34/3, (b) 11, (c) 14, (d) 32/3, , 6.68., , 12, , 6.107. (a) 0 does not exist., (b) <f> — x2e~i + y cos z + constant, , 6.69., , 64)r, , 6.108. xzs - 2x2y + 83/2 + 2 = constant, , (c) 6
Page 192 :
Chapter 7, Fourier Series, , PERIODIC FUNCTIONS, A function f(x) is said to have a period T or to be periodic with period T if for all x,, f(x + T)- f(x), where T is a positive constant. The least value of T > 0 is called the, least period or simply the period of /(#)., Example 1. The function sin x has periods 2ir, 4v, &¡r, ..., since sin (x + 2v), sin (x + 4s-), sin (* + 6n-),, ... all equal sin x. However, 2ir is the least period or the period of sin x., Example 2. The period of sin nx or cos nx, where M is a positive integer, is 2ir/n., Example 3. The period of tan x is ir., Example 4. A constant has any positive number as period., , Other examples of periodic functions are shown in the graphs of Figures 7-l(a), (b), and (c) below., , (a), , (b), , (c), , Fig. 7-1, , FOURIER SERIES, Let f(x) be defined in the interval (—L, L) and outside of this interval by f(x + 2L) = f(x),, i.e. assume that f(x) has the period 2L. The Fourier series or Fourier expansion corresponding to f(x) is given by, (1), , where the Fourier coefficients, , an and bn are, (2), , If f(x) has the period 2L, the coefficients an and bn can be determined equivalently, from, (3), , where c is any real number. In the special case c = —L, (3) becomes (2)., , 182
Page 193 :
CHAP. 7], , FOURIER SERIES, , 183, , To determine a0 in (1), we use (2) or (3) with n - 0. For example, from (2) we, see t h a t f ( x ) d x ., , Note that the constant term in (1) is equal to, , =, , /(a;) àx, which is the mean of f(x) over a period., If L = r, the series (1) and the coefficients (2) or (3) are particularly simple., function in this case has the period 2-*., , The, , DIRICHLET CONDITIONS, Theorem 7-1. Suppose that, (1) f(x) is defined and single-valued except possibly at a finite number of, points in (—L,L), (2), , f(x) is periodic outside (—L,L) with period 2L, , (3) f(x) and f'(x) are piecewise continuous in (—L,L)., Then the series (1) with coefficients (2) or (3) converges to, (a) f(x) if a; is a point of continuity, (6), , if z is a point of discontinuity, , In this theorem f(x + 0) and f(x — 0) are the right and left hand limits of f(x) at x and, represent lim f(x + £) and lim f(x — e) respectively where e > 0. These are often written, £-*0, e-»0, lim f(x + e) and lim f i x — <•) to emphasize that « is approaching zero through positive, e-»0+, e-»0 +, values. For a proof see Problems 7.18-7.23., The conditions (1), (2) and (3) imposed on f(x) are sufficient but not necessary, and, are generally satisfied in practice. There are at present no known necessary and sufficient, conditions for convergence of Fourier series. It is of interest that continuity of f(x) does, not alone insure convergence of a Fourier series., ODD AND EVEN FUNCTIONS, A function f(x) is called odd if /(—*) = —/(«)• Thus x3, x5 — Sx3 + 2x, sin x, tan 3a;, are odd functions., A function f(x) is called even if f(—x) = f(x). Thus x4, 2x6 — 4a;2+ 5, cosa;, ex + e~x, are even functions., The functions portrayed graphically in Figures 7-l(a) and 7-1(6) are odd and even, respectively, but that of Fig. 7-1 (c) is neither odd nor even., In the Fourier series corresponding to an odd function, only sine terms can be, present. In the Fourier series corresponding to an even function, only cosine terms (and, possibly a constant which we shall consider a cosine term) can be present., HALF RANGE FOURIER SINE OR COSINE SERIES, A half range Fourier sine or cosine series is a series in which only sine terms or, only cosine terms are present respectively. When a half range series corresponding to, a given function is desired, the function is generally defined in the interval (0, L) [which, is half of the interval (-L,L), thus accounting for the name half range] and then the
Page 194 :
134, , FOURIER SERIES, , [CHAP. 7, , function is specified as odd or even, so that it is clearly defined in the other half of the, interval, namely (-L, 0). In such case, we have, for half range sine series, (4), , for half range cosine series, , PARSEVAL'S IDENTITY states that, (5), , if an and bn are the Fourier coefficients corresponding to f(x) and if f(x) satisfies the Dirichlet, conditions., DIFFERENTIATION AND INTEGRATION OF FOURIER SERIES, Differentiation and integration of Fourier series can be justified by using the theorems, on page 7 which hold for series in general. It must be emphasized, however, that those, theorems provide sufficient conditions and are not necessary. The following theorem for, integration is especially useful., Theorem 7-2. The Fourier series corresponding to f(x) may be integrated term by term, from a to x, and the resulting series will converge uniformly to, , f(u) du, , provided that f(x) is piecewise continuous in — L ê x a L and both a and x, are in this interval., COMPLEX NOTATION FOR FOURIER SERIES, Using Euler's identities,, e*» = cos 0 + i sin O,, , where i =, , e~u = cos 9 - i sin e, , (6), , [see Problem 1.61, page 30], the Fourier series for f(x) can be written as, ( 7 ), , where, , c», , =, , (8), , In writing the equality (7), we are supposing that the Dirichlet conditions are satisfied, and further that f(x) is continuous at x. If f(x) is discontinuous at x, the left side of (7), should be replaced by, , ORTHOGONAL FUNCTIONS, Two vectors A and B are called orthogonal (perpendicular) if A • B = 0 or AiBi +, AzBz + A3B3 = 0, where A = AJ + A2j + Aak and B = BA + B2j + #sk. Although not geometrically or physically evident, these ideas can be generalized to include vectors with
Page 195 :
CHAP. 7], , FOURIER SERIES, , 185, , more than three components. In particular we can think of a function, say A(x), as being a, vector with an infinity of components (i.e. an infinite dimensional vector), the value of each, component being specified by substituting a particular value of x in some interval (a, b)., It is natural in such case to define two functions, A(x) and B(x), as orthogonal in (a, b) if, (9), , A vector A is called a unit vector or normalized vector if its magnitude is unity, i.e., if A - A = A2 = 1. Extending the concept, we say that the function A(x) is normal or, normalized in (a, b) if, (10), From the above it is clear that we can consider a set of functions {</>fc(*)}, k = 1,2,3,...,, having the properties, (11), (12), , In such case, each member of the set is orthogonal to every other member of the set and, is also normalized. We call such a set of functions an orthonormal set in (a, b)., The equations (11) and (12) can be summarized by writing, (13), , where 8mn, called Kronecker's symbol, is defined as 0 if m ¥* n and 1 if m = n., Just as any vector r in 3 dimensions can be expanded in a set of mutually orthogonal, unit vectors i, j, k in the form r = di + Caj + c3k, so we consider the possibility of expanding, a function f(x) in a set of orthonormal functions, i.e.,, (U), , Such series, called orthonormal series, are generalizations of Fourier series and are of great, interest and utility both from theoretical and applied viewpoints., If, , (15), , where w(x) ê 0, we often say that $m(x) and qn(x) are orthonormal with respect to the, density function or weight function w(x). In such case the set of functions, is, an orthonormal set in (a, 6).
Page 196 :
186, , FOURIER SERIES, , [CHAP. 7, , Solved Problems, FOURIER SERIES, 7.1. Graph each of the following functions., (a) f(x), , =, , Period = 10, , Fig. 7-2, , Since the period is 10, that portion of the graph in — 5 < « < 5 (indicated heavy in Fig. 7-2, above) is extended periodically outside this range (indicated dashed). Note that /(*) is not, defined at x = 0,5, —5,10, —10,15, —15, etc. These values are the discontinuities of f(x)., , (6) f(x), , =, , Period = 2*, , Fig. 7-3, Refer to Fig. 7-3 above. Note that /(«) is defined for all x and is continuous everywhere., , (c) f(x), , =, , Period = 6, , Fig. 7-4, , Refer to Fig. 7-4 above. Note that /(*) is defined for all x and is discontinuous at x = ±2,, ±4, ±8, ±10, ±14,, , 7.2., , Prove
Page 197 :
CHAP. 7], , 7.3., , FOURIER SERIES, , 187, , L, *LL, m-vx cos —fn-n-x , = C, *x sin-7-da, .• n-nx, _., , . C, mirx, n-Trx, • m, wivx, n-^x , = •{J O m ¥° n, \ sin—=^, Prove (a), I cos^f^cos—f-dx, cos -^^dx, r, J-L, L, J-L, L, LI, -L, L, LI, \Lm-n, , (b) c si, L, , . m-irX, n-irX ,, sm^f—cos—j^dx, — 0f ., LI, LI, where m and n can assume any of the values 1,2,3,... ., J-L, , (a) From trigonometry: cos A cos B = ^{cos (A—B) + cos (A + B)}, sin A sin B = ^{cos (A — B) —, cos(A+B)}., Then, if m ¥* n, we have by Problem 7.2,, , J, , L, , ,, — Lt, , mirx, mrx ,, Í CL }, (m — ri)vx ., (m + n)vx\,, cos—^cos-^iZa; = 7; I, •{ cos- =-*, h cos^ ~—> dx, Li, Li, ¿ J lj. \.\, Li, L, I, s, , = n0, , Similarlyy if m, m ¥* n,, 1, J'*ç*ç* .!/'{.cos-(m —LIj-n)vx, — LJ, , Ll, , \,, , (m + rilirx ,, cos^ LIf-1— Y dx, , =, , ., 0, , If m — n, we have, L, , mirx, , nvx ,, , cos —-f—, cos -7— dx, L, / -L, Li, , 1 CL /., ., , 2nvx dx, L, , = 2-^ I, 1 + cos J-L V, , = L, , L, . —f—, rmrx sin, . —=—, nvx,, IC sin, dx, L, L, , L, 2nirx\ ,, ,, = 2-¿1 IC IL1 — cos —=—, I dx — L, L, I, -'-L, J-L V, Note that if m = n = 0 these integrals are equal to 2L and 0 respectively., , (6) We have sin A cos B = £{sin (A - B) + sin (A +B)}. Then by Problem 7.2, if m¥*n,, , If m — n,, , The results of parts (a) and (6) remain valid even when the limits of integration —L,L are, replaced by c, c + 2L respectively., , 7.4., , °o, , f, , Vitr'Y, , 'Mir'ï*\, , Li, , Lt /, , If the series A + 2J ( a» cos —j- + bn sin -=r- ) converges uniformly to f(x) in (-L, L),, n=l \, , show that for n = 1,2,3, ... ,, , (a) Multiplying, , /(«) = A + 2 fa, , (1), , by cosm?x and integrating from — L to L, using Problem 7.3, we have, LI, (2)), , Thuss
Page 198 :
Igg, , FOURIER SERIES, (6) Multiplying (Í) by sin, , [CHAP. 7, , and integrating from -L to L, using Problem 7.3, we have, , Thus, (c) Integration of (1) from — L to L, using Problem 7.2, gives, , Putting m = 0 in the result of part (a), we find a0 =, , f(x) dx and so, , The above results also hold when the integration limits —L, L are replaced by c,c + 2L., Note that in all parts above, interchange of summation and integration is valid because the, series is assumed to converge uniformly to f(x) in (—L,L). Even when this assumption is not, warranted, the coefficients am and 6m as obtained above are called Fourier coefficients corresponding, to /(«), and the corresponding series with these values of am and bm is called the Fourier series, corresponding to /(a;). An important problem in this case is to investigate conditions under which, this series actually converges to f(x). Sufficient conditions for this convergence are the Dirichlet, conditions established below., , 7.5., , (a) Find the Fourier coefficients corresponding to the function, f(x), , Period = 10, , =, , (&) Write the corresponding Fourier series., (c) How should f(x) be defined at x — —5, x = 0 and x = 5 in order that the Fourier, series will converge to f(x) for —5 a x a 5?, The graph of f(x) is shown in Fig. 7-5 below., , Fig. 7-5, (a) Period = 2L = 10 and L = 5. Choose the interval c to c + 2L as -5 to 5, so that c = -5., Then, , If
Page 199 :
CHAP. 7], , FOURIER SERIES, , 189, , (b) The corresponding Fourier series is, , (c) Since f(x) satisfies the Dirichlet conditions, we can say that the series converges to f(x) at all, points of continuity and to —, ., at points of discontinuity. At x = —5, 0 and 5,, which are points of discontinuity, the series converges to (3 + 0)/2 = 3/2 as seen from the, graph. If we redefine f(x) as follows,, , f(x), , Period = 10, , =, , then the series will converge to f(x) for —5 S a; S 5., , 7.6., , Expand f(x) = y?, 0 < x < 2* in a Fourier series if (a) the period is 2*, (b) the, period is not specified., (a) The graph of f(x) with period 2;r is shown in Fig. 7-6 below., , Fig. 7-6, Period = 2L = 2ir and L = ir., , If, , Then, , Choosing c = 0, we have
Page 200 :
!90, , FOURIER SERIES, , [CHAP. 7, , This is valid for 0 < x < 2ir. At * = 0 and * = 2v the series converges to 27r2., (5) If the period is not specified, the Fourier series cannot be determined uniquely in general., , 7.7., , Using the results of Problem 7.6, prove that, At x = 0 the Fourier series of Problem 7.6 reduces to, By the Dirichlet conditions, the series converges at x = 0 to, Then, , and so, , ODD AND EVEN FUNCTIONS. HALF RANGE FOURIER SERIES, 7.8. Classify each of the following functions according as they are even, odd, or neither, even nor odd., (a) /(»), , =, , Period = 6, , From Fig. 7-7 below it is seen that /(—x) = —/(«), so that the function is odd., , Fig. 7-7, , (6) /(»), , =, , Period = 2*, , From Fig. 7-8 below it is seen that the function is neither even nor odd., , Fig. 7-8, , (c) f(x) = «(10-a;), 0 < x < 10, Period = 10., From Fig. 7-9 below the function is seen to be even.
Page 201 :
CHAP. 7], , 7.9., , FOURIER SERIES, , 191, , Show that an even function can have no sine terms in its Fourier expansion., Method 1., , No sine terms appear if bn = 0, n = 1,2, 3,, , To show this, let us write, , a), If we make the transformation x = — u in the first integral on the right of (Í), we obtain, (2), , where we have used the fact that for an even function /(—u) = /(«) and in the last step that the, dummy variable of integration u can be replaced by any other symbol, in particular x. Thus from, (1), using (2), we have, , Method 2., , Assume, Then, If f(x) is even, f(—x) = f(x). Hence, , and so, , and no sine terms appear., In a similar manner we can show that an odd function has no cosine terms (or constant term), in its Fourier expansion., , 7.10. If f(x) is even, show that (a) a« =, (a), , Letting x = —u,, , since by definition of an even function /(—u) — f(u)., , Then, , (6) This follows by Method 1 of Problem 7.9., , 7.11. Expand f(x) = sin x, 0 < x < -a, in a Fourier cosine series., A Fourier series consisting of cosine terms alone is obtained only for an even function. Hence, we extend the definition of f(x) so that it becomes even (dashed part of Fig. 7-10 below). With this, extension, f(x) is then defined in an interval of length 2ir. Taking the period as 2jr, we have, 2L = 2n- so that L = v.
Page 202 :
192, , FOURIER SERIES, , [CHAP. 7, , Fig. 7-10, By Problem 7.10, bn = 0 and, , For n - 1,, For n = 0,, , Then, , 7.12. Expand f(x) = x, 0 < x < 2, in a half range (a) sine series, (b) cosine series., (a) Extend the definition of the given function to that of the odd function of period 4 shown in, Fig. 7-11 below. This is sometimes called the odd extension of /(«). Then 2L = 4, L = 2., , Fig. 7-11, Thus an — 0 and, , Then
Page 203 :
CHAP. 7], , FOURIER SERIES, , 193, , (&) Extend the definition of f(x) to that of the even function of period 4 shown in Fig. 7-12 below., This is the even extension of f(x). Then 2L = 4, L = 2., , It should be noted that the given function f(x) = x, 0 < x < 2, is represented equally well, by the two different series in (a) and (6)., , PARSEVAL'S IDENTITY, 7.13. Assuming that the Fourier series corresponding to f(x) converges uniformly to f(x), in (—L,L), prove Parseval's identity, , where the integral is assumed to exist., then multiplying by /(se) and integrating term, by term from — L to L (which is justified since the series is uniformly convergent) we obtain, , (1), where we have used the results, (*), , obtained from the Fourier coefficients., The required result follows on dividing both sides of (J) by L., under less restrictive conditions than that imposed here., , Parseval's identity is valid
Page 204 :
194, , FOURIER SERIES, , [CHAP. 7, , 7.14. (a) Write Parseval's identity corresponding to the Fourier series of Problem 7.12(6)., (&) Determine from (a) the sum S of the series jï.+ 2* + 34 + ' ' ' + ^* + " ' •, , 4, (a) Here L = 2, ac = 2, an = -g-g (cos «"" ~ 1)>, , w, , ^ °> 6» = °-, , Then Parseval's identity becomes, , 7.15. Prove that for all positive integers M,, , where o« and bn are the Fourier coefficients corresponding to f(x), and f(x) is assumed, piecewise continuous in (—L, L)., Let, , U), , For M = 1,2,3,... this is the sequence of partial sums of the Fourier series corresponding to /(«)., We have, , (2), , since the integrand is non-negative. Expanding the integrand, we obtain, (3), , Multiplying both sides of (1) by 2f(x) and integrating from —L to L, using equations (a) of, Problem 7.13, gives, (4), Also, squaring (1) and integrating from —L to L, using Problem 7.3, we find, (5), , Substitution of (4) and (5) into (3) and dividing by L yields the required result., Taking the limit as M -» », we obtain Bessel's inequality, (6), , If the equality holds, we have Parseval's identity (Problem 7.13)., We can think of SM(a;) as representing an approximation to /(*), while the left hand side of, (2), divided by 2L, represents the mean square error of the approximation. Parseval's identity, indicates that as M -» °° the mean square error approaches zero, while Bessel's inequality indicates, the possibility that this mean square error does not approach zero., The results are connected with the idea of completeness of an orthonormal set. If, for example,, we were to leave out one or more terms in a Fourier series (say cos 4vx/L, for example) we could, never get the mean square error to approach zero no matter how many terms we took. For an, analogy with 3 dimensional vectors, see Problem 7.46.
Page 205 :
CHAP. 7], , FOURIER SERIES, , 195, , DIFFERENTIATION AND INTEGRATION OF FOURIER SERIES, 7.16. (a) Find a Fourier series for f(x) = xz, 0 < x < 2, by integrating the series of, Problem 7.12(a). (b) Use (a) to evaluate the series, (a) From Problem 7.12(a),, (1), , Integrating both sides from 0 to a; (applying Theorem 7-2, page 184) and multiplying by 2,, we find, (*), where, (6) To determine C in another way, note that (2) represents the Fourier cosine series for x2 in, 0 < x < 2. Then since L = 2 in this case,, , Then from the value of C in (a), we, , 7.17. Show that term by term differentiation of the series in Problem 7.12(a) is not valid., Term by term differentiation yields 2, Since the nth term of this series does not approach 0, the series does not converge for any, value of x., , CONVERGENCE OF FOURIER SERIES, 7.18. Prove that, , (a) We have cos nt sin, Then summing from n = 1 to M,, sin ^t{cos í + cos 2< + • • • + cos Mt}, , =, , On dividing by sin -Jt and adding ^, the required result follows., (6) Integrate the result in (a) from —v to 0 and 0 to w respectively., results, since the integrals of all the cosine terms are zero., , 7.19. Prove that, continuous., This, , followsfollows, , This gives the required, , is piecewise, at once from Problem 7.15, since if the series, , The result is sometimes called Riemann's theorem., , is convergent,, is conver
Page 206 :
196, , FOURIER SERIES, , 7.20. Prove that, , [CHAP. 7, , is piecewise continuous., , We have, sin MX dx, , Then the required result follows at once by using the result of Problem 7.19, with /(») replaced by, /(*) sin %x and /(«) cos £x respectively which are piecewise continuous if /(«) is., The result can also be proved when the integration limits are a and b instead of —r and a-,, , 7.21. Assuming that L = *, i.e. that the Fourier series corresponding to f(x) has period, 2L = 2ir, show that, , Using the formulas for the Fourier coefficients with L = JT, we have, an cos nx + bn sin nx =, , sin nx, , =, , f(u) (cos MM cos nx + sin nu sin nx) du, , =, , /(M) cos n(u — x) du, , Also,, Then, , using Problem 7.18. Letting u — x = t, we have, , Since the integrand has period 2ir, we can replace the interval —ir — x,ir — x by any other, interval of length 2ir, in particular — -IT, v. Thus we obtain the required result., , 7.22. Prove that, , From Problem 7.21,, U), , Multiplying the integrals of Problem 7.18(6) by f(x — 0) and f(x + 0) respectively,, (2), Subtracting (S) from (Í) yields the required result.
Page 207 :
CHAP. 7], , FOURIER SERIES, , 197, , 7.23. If f(x) and /'(«) are piecewise continuous in (—IT, *), prove that, , The f u n c t i o n i s piecewise continuous in, continuous., Also., , because f(x) is piecewise, , exists, since by hypothesis /'(*) is piecewise continuous so that the right hand derivative of f(x) at, each x exists., Thus, , is piecewise continuous in, A, , Similarly,, , is piecewise continuous in, , Then from Problems 7.20 and 7.22, we have, , ORTHOGONAL FUNCTIONS, 7.24. (a) Show that the set of functions, , forms an orthogonal set in the interval (—L, L)., (b) Determine the corresponding normalizing constants for the set in (a) so that the, set is orthonormal in (—L, L)., (a) This follows at once from the results of Problems 7.2 and 7.3., (6) By Problem 7.3,, , Then, Also,, Thus the required orthonormal set is given by, , 7.25. Let, , be a set of functions which are mutually orthonormal in (a,b)., , that, , CO, , Prove, , converges uniformly to f(x) in (a, b), then, , Multiplying both sides of, (1), , by <i>m(x) and integrating from a to 6, we have
Page 208 :
198, , FOURIER SERIES, , [CHAP. 7, , (2), , where the interchange of integration and summation is justified by using the fact that the series, converges uniformly to /(»). Now since the functions {<t>n(x)} are mutually orthonormal in (a, 6),, we have, so that (2) becomes, (*), as required., We call the coefficients cm given by (S) the generalized Fourier coefficients corresponding to, /(») even though nothing may be known about the convergence of the series in (i). As in the case, , o f Fourier series, convergence o, , i, , f, , i i s then investigated using t h e coefficients (S). T h e, , conditions of convergence depend of course on the types of orthonormal functions used., , Supplementary Problems, FOURIER SERIES, , 7.26., , Graph each of the following functions and find their corresponding Fourier series using properties, of even and odd functions wherever applicable., Period 4, , P, , e, , r, , i, , Period 10, , o, , d8, Period 6, , 7.27., , In each part of Problem 7.26, tell where the discontinuities of /(*) are located and to what value, the series converges at these discontinuities., , 7.28., , Expand, , 7.29., , (a) Expand . . ., ,, in a Fourier sine series., (6) How should f(x) be defined at * = 0 and x = v so that the series will converge to f(x) for, OSxSvl, , 7.30., , (a) Expand in a Fourier series f(x) = cos x, 0 < * < v if the period is v, and (6) compare with, the result of Problem 7.29, explaining the similarities and differences if any., , 7.31., , Expand, , 7.32., , Prove that for, , 7.33., , Use Problem 7.32 to show that, , in a Fourier series of period 8., , in a series of (a) sines, (6) cosines.
Page 209 :
CHAP. 7], , 7.34., , FOURIER SERIES, , 199, , Show that, , DIFFERENTIATION AND INTEGRATION OF FOURIER SERIES, 7.35. (a) Show that for, , (6) By integrating the result of (a), show that for —v'&x'&v,, , ), (e), , 7.36., , By integrating the result of (b), show that for —tr S x S v,, , (a) Show that for, , (6) Use (a) to show that for, , 7.37., , By differentiating the result of Problem 7.32(6), prove that for 0 S x S v,, , PARSEVAL'S IDENTITY, 7.38. By using Problem 7.32 and Parseval's identity, show that, , 7.39., , Show that, , 7.40., , Show that, , 7.41., , Show that, , [Hint. Use Problem 7.11.], , ORTHOGONAL FUNCTIONS, 7.42. Given the functions a0, «j + a¿e, a3 + atx + asx2 where a0, 0*5 are constants. Determine the, constants so that these functions are mutually orthonormal in (—1,1) and thus obtain the functions., 7.43., , Generalize Problem 7.42., , 7.44., , (a) Show that the functions, , are mutually orthonormal in (—w, v). (b) Show, , how to expand a function /(») in a series of these functions and explain the connection with Fourier, series., 7.45., , Let f(x) be approximated by the sum of the first M terms of an orthonormal series, , where the functions <j>n(x) are orthonormal in (a, 6). (a) Show that
Page 210 :
200, , FOURIER SERIES, , [CHAP. 7, , (6) By interpreting, as the mean square error of SM(x) from /(») [and the square root as the root mean square or, r.m.s. error], show that Parseval's identity is equivalent to the statement that the root mean, square error approaches zero as M -» »., (c) Show that if the root mean square error may not approach zero as M -» », then we still have, Bessel's inequality, (d) Discuss the relevance of these results to Fourier series., 7.46., , Let r be any three dimensional vector., , Show that, , and discuss these with reference to Bessel's inequality and Parseval's identity., Problem 7.15., , Compare with, , 7.47., , Suppose that one term in any orthonormal series [such as a Fourier series] is omitted, (a) Can we, expand a function /(«) into the series? (6) Can Parseval's identity be satisfied? (c) Can Bessel's, inequality be satisfied? Justify your answers., , 7.48., , Let, , be orthonormal in (a, b). Prove that, , is a minimum when, Discuss the connection of this to (a) Fourier series and (6) Problem 7.45., 7.49., , (a) Show that the functions 1,1 — x, 2 — 4» + x2 are mutually orthogonal in (0, ») with respect to, the density function e~x. (b) Obtain a mutually orthonormal set., , 7.50., , Give a vector interpretation to functions which are orthonormal with respect to a density or, weight function., , Answers to Supplementary Problems, 7.26., , (a), <«), , 7.27., , (a), (c), , 7.27. ( a ) b ) n o discontinuities, , 7.28., 7.29., , (a), , 7.30., , Same answer as in Problem 7.29., , 7.31., , (a)
Page 211 :
Chapter 8, Fourier Integrals, , THE FOURIER INTEGRAL, , Let us assume the following conditions on f(x):, 1. /(*) satisfies the Dirichlet conditions (page 183) in every finite interval (-L,L)., 2., , dx converges, i.e. f(x) is absolutely integrable in, , Then Fourier's integral theorem states that, (1), where, , (2), , The result (1) holds if a; is a point of continuity of f(x)., , If a; is a point of discontinuity,, , we must replace f(x) by, as in the case of Fourier series. Note that, the above conditions are sufficient but not necessary., The similarity of (1) and (2) with corresponding results for Fourier series is apparent., The right hand side of (1) is sometimes called a Fourier integral expansion of f(x)., EQUIVALENT FORMS OF FOURIER'S INTEGRAL THEOREM, , Fourier's integral theorem can also be written in the forms, (3), , W, , where it is understood that if f(x) is not continuous at x the left side must be replaced, by, These results can be simplified somewhat if f(x) is either an odd or an even function,, and we have, , 201, , is even, , (5), , is odd, , (6)
Page 212 :
202, , FOURIER INTEGRALS, , [CHAP. 8, , FOURIER TRANSFORMS, From (4) it follows that if, (7), , then, , (*), , The function F(a) is called the Fourier transform of f(x) and is sometimes written, The function f(x) is the inverse Fourier transform of F(a) and is written, Note: The constants preceding the integral signs in (7) and (8) were here taken as, equal to l/\/2^. However, they can be any constants different from zero so long as their, product is 1/27T. The above is called the symmetric form., If f(x) is an even function, equation (5) yields, , (9), , and we call Fc (a) and f(x) Fourier cosine transforms of each other., If f(x) is an odd function, equation (6) yields, (10), , and we call Fs(a) and f(x) Fourier sine transforms of each other., PARSEVAL'S IDENTITIES FOR FOURIER INTEGRALS, If Fs(a) and Gs(a) are Fourier sine transforms of f(x) and g(x) respectively, then, , (11), Similarly if Fc(a) and Gc(a) are Fourier cosine transforms of f(x) and g(x), then, , (12), In the special case where, , ) and (12) become respectively, (18), (14), , The above relations are known as Parseval's identities for integrals. Similar relations hold for general Fourier transforms. Thus if F(a) and G(a) are Fourier transforms, of f(x) and 0(0;) respectively, we can prove that
Page 213 :
203, , FOURIER INTEGRALS, , CHAP. 8], , (15), , where the bar signifies the complex conjugate obtained by replacing i by -i. fc>e<, Problem 8.24., THE CONVOLUTION THEOREM, If F(«) and G(«) are the Fourier transforms of f(x) and g(x) respectively, then, (16), , If we define the convolution, denoted by / * g, of the functions / and g to be, (17), , then (16) can be written, , (18), , or in words, the Fourier transform of the convolution of two functions is equal to the, product of their Fourier transforms. This is called the convolution theorem for Fourier, transforms., , Solved Problems, THE FOURIER INTEGRAL AND FOURIER TRANSFORMS, 8.1., , (a) Find the Fourier transform of, (b) Graph f(x) and its Fourier transform for a = 3., (a) The Fourier transform of f(x) is, , For a = 0, we obtain, (6) The graphs of /(*) and F(a) for a = 3 are shown in Figs. 8-1 and 8-2 respectively., , Fig. 8-1, , Fig. 8-2
Page 214 :
204, , 8.2., , FOURIER INTEGRALS, , [CHAP. 8, , (a) Use the result of Problem 8.1 to evaluate, (6) Deduce the value of, (a) From Fourier's integral theorem, if, then, Then from Problem 8.1,, (1), , The left side of (1) is equal to, (*), , The integrand in the second integral of (2) is odd and so the integral is zero. Then from, (1) and (2), we have, (3), , (b) If x = 0 and a — 1 in the result of (a), we have, , since the integrand is even., , 8.3., , If f(x) is an even function show that:, (a), We have, (1), , (a) If f(u) is even, f(u) cos XM is even and f(u) sin Xw is odd. Then the second integral on the, right of (1) is zero and the result can be written, , (6) From (a), F(—a) = F(a) so that F(a) is an even function. Then by using a proof exactly, analogous to that in (a), the required result follows., A similar result holds for odd functions and can be obtained by replacing the cosine by, the sine., , 8.4., , Solve the integral equation, L e t, Problem 8.3,, , a n d c, , tHEN BY
Page 215 :
CHAP. 8], , 8.5., , FOURIER INTEGRALS, , 205, , Use Problem 8.4 to show that, As obtained in Problem 8.4,, , Taking the limit as a -» 0+, we find, , But this integral can be written as, , which becomes, , on letting, , x = 2u, so that the required result follows., , 8.6., , Show that, Let, , in the Fourier integral theorem, , Then, But from the result 16 on page 6 we have, or, , PARSEVAL'S IDENTITY, , 8.7., , Verify Parseval's identity for Fourier integrals for the Fourier transforms of, Problem 8.1., We must show that, , where, , and, , This is equivalent to, , or, , i.e.,, By letting aa, = u and using Problem 8.5, it is seen that this is correct. The method can also, be used to find, directly.
Page 216 :
206, , FOURIER INTEGRALS, , [CHAP. 8, , CONVOLUTION THEOREM, , 8.8., , Solve t h e integral e, are given., , q, , u, , a, , t, , i, , o, , n, , w, , h, , e, , r, , e g(x) a n d r(#), , Suppose that the Fourier transforms of y(x), g(x) and r(x) exist, and denote them by Y(a), G(a), and R(a) respectively. Then taking the Fourier transform of both sides of the given integral, equation, we have by the convolution theorem, , Then, assuming this integral exists., , 8.9., , Solve for y(x) the integral, , We have, , where we have used the transformation u = bv and the result of Problem 8.6. Then taking the, Fourier transform of both sides of the integral equation, we find, , i.e., , or, , Thus, , PROOF OF THE FOURIER INTEGRAL THEOREM, , 8.10. Present a heuristic demonstration of Fourier's integral theorem by use of a limiting, form of Fourier series., Let, , (i), , where, Then by substitution of these coefficients into (1) we find, (2), , If we assume t h a t c o n v e r g e s , the first term on the right of (2) approaches zero as, L -» «, while the remaining part appears to approach, (3), This last step is not rigorous and makes the demonstration heuristic.
Page 217 :
CHAP. 8], , FOURIER INTEGRALS, Calling, , 207, , can be written, U), , where we have written, , (5), , But the limit (4) is equal to, , which is Fourier's integral formula., This demonstration serves only to provide a possible result. To be rigorous, we start with the, integral, and examine the convergence. This method is considered in Problems 8.11-8.14., , 8.11. Prove that:, , (a), , (6), can be, , (a), shown by using Problem 8.27., (6), , 8.12. Riemann's theorem states that if F(x) is piecewise continuous in (a, 6), then, , with a similar result for the cosine (see Problem 8.28)., , Use this to prove that, , (a), (6), where /(#) and f'(x) are assumed piecewise continuous in (0, L) and (—L, 0) respectively., (a), , Using Problem 8.11(a), it is seen that a proof of the given result amounts to proving that, , This follows a t once from Riemann's theorem, because, b, e, , is piece-, , wise continuous in (0,L) since lim F(v) exists and f(x) is piecewise, D-»0 +, , conti, contin, , (6) A proof of this is analogous to that in part (a) if we make use of Problem 8.11(6)., , 8.13., , If f ( x ) satisfies the additional condition that, , converges, prove that, , (a), We have, , w, (2)
Page 218 :
208, , FOURIER INTEGRALS, , [CHAP. 8, , Subtracting,, (*), , Denoting the integrals in (3) by I, /1; /2 and /3 respectively, we have / = /t + /2 +13 so that, (4), Now, , Also, Since, , and, , both converge, we can choose L so large that |/2| S «/3,, , |/8| S «/3. Also, we can choose a so large that |/jj S e/3. Then from (4) we have |/| < e for a and, L sufficiently large, so that the required result follows., This result follows by reasoning exactly analogous to that in part (a)., , 8.14. Prove Fourier's integral formula where f(x) satisfies the conditions stated on page 201., We must prove that, Since, , which converges, it follows by the Weier-, , strass test for integrals [see Problem 1.123, page 33] that, , converges abso-, , lutely and uniformly for all a. We can show from this that the order of integration can be, reversed to obtain, , where we have let u = x + v., Letting L -» «, we see by Problem 8.13 that the given integral converges to, as required., , Supplementary Problems, THE FOURIER INTEGRAL AND FOURIER TRANSFORMS, 8.15., , (a) Find the Fourier transform of, (6) Determine the limit of this transform as e -» 0+ and discuss the result., , 8.16., , (a) Find the Fourier transform of, (6) Evaluate, , 8.17., , If, f(x)., , fi, , the (a) Fourier sine transform, (6) Fourier cosine transform of, , In each case obtain the graph of f(x) and its transform., , of
Page 219 :
CHAP. 8], 8.18., , FOURIER INTEGRALS, , 209, , (a) Find the Fourier sine transform of, (6) Show that, , by using the result in (a)., , 8.19., , (c) Explain from the viewpoint of Fourier's integral theorem why the result in (6) does not hold, for m — 0., Solve for Y(x) the integral equation, , 8.20, , Establish equation (4), page 201, from equation (3), page 201., , and verify the solution by direct substitution., , PARSEVAL'S IDENTITY, 8.21., , Evaluate, , (a), , (6), , by use of Parseval's identity., , [Hint. Use the Fourier sine and cosine transforms of, 8.22., , Use Problem 8.17 to show that, , 8.23., , Show that, , 8.24., , (a) If F(a) and <?(<*) are the Fourier transforms of f(x) and g(x) respectively, prove that, , (a), , (b), , where the bar signifies the complex conjugate., (6) From (a) obtain the results (11)-(U), page 202., CONVOLUTION THEOREM, 8.25. Verify the convolution theorem for the functions, 8.26., , Prove the result (18), page 203., [Hint. If, , and, , then, , Now make the transformation, PROOF OF FOURIER INTEGRAL THEOREM, 8.27. By interchanging the order of integration in, , prove that, , and thus complete the proof in Problem 8.11., 8.28., , Prove Riemann's theorem [see Problem 8.12]., , Answers to Supplementary Problems, 8.15., , (a), , 8-17., , (a), , 8.19., , 8.16., , (b), (6), , (a), , (6), , 8.18., , (a), , 8.21., , (a), , (6)
Page 220 :
Chapter 9, Gamma, Beta and, , Other Special Functions, , THE GAMMA FUNCTION, The gamma function denoted by T(ri) is defined by, , (1), which is convergent for n > 0., A recursion or recurrence formula for the gamma function is, (*), where r(l) = 1 (see Problem 9.1). From (2), T(n) can be determined for all n > 0 when, the values for 1 g n < 2 (or any other interval of unit length) are known (see table below)., In particular if n is a positive integer, then, , (3), , For this reason T(ri) is sometimes called the factorial function., Examples., , It can be shown (Problem 9.4) that, (-4), , The recurrence relation (2} is a difference equation which has (1) as a solution. By, taking (1) as the definition of r(n) for n > 0, we can generalize the gamma function to, n < 0 by use of (2) in the form, , (5), See Problem 9.7, for example. The process is called analytic continuation., TABLE OF VALUES AND GRAPH OF THE GAMMA FUNCTION, n, , T(n), , 1.00, 1.10, 1.20, 1.30, 1.40, 1.50, 1.60, 1.70, 1.80, 1.90, 2.00, , 1.0000, 0.9514, 0.9182, 0.8975, 0.8873, 0.8862, 0.8935, 0.9086, 0.9314, 0.9618, 1.0000, , Fig 9 1, , ' ", , 210
Page 221 :
CHAP. 9], , GAMMA, BETA AND OTHER SPECIAL FUNCTIONS, , 211, , ASYMPTOTIC FORMULA FOR r(»), If n is large, the computational difficulties inherent in a calculation of r(«) are, apparent. A useful result in such case is supplied by the relation, («), For most practical purposes the last factor, which is very close to 1 for large n, can be, omitted. If n is an integer, we can write, (7), , where ~ means "is approximately equal to for large n". This is sometimes called Stirling's, factorial approximation or asymptotic formula for nl, MISCELLANEOUS RESULTS INVOLVING THE GAMMA FUNCTION, 1., , In particular if, , 2., This is called the duplication formula for the gamma function., 3., , The result 2 is a special case of this with m — 2., 4., , This is called Stirling's asymptotic series for the gamma function., braces is an asymptotic series (see pages 212 and 219)., , The series in, , 5., , where y is Euler's constant and is defined as, , 6., , THE BETA FUNCTION, The beta function, denoted by B(m,ri) is defined by, («), , which is convergent for m > 0, n > 0., The beta function is connected with the gamma function according to the relation, See Problem 9.11., Many integrals can be evaluated in terms of beta or gamma functions., results are, , (»), , Two useful, (W)
Page 222 :
212, , GAMMA, BETA AND OTHER SPECIAL FUNCTIONS, , [CHAP. 9, , valid for TO > 0 and n > 0 [see Problems 9.11 and 9.14] and, , (11), See Problem 9.18., DIRICHLET INTEGRALS, , If V denotes the closed region in the first octant bounded by the surface, and the coordinate planes, then if all the constants are positive,, (12), , Integrals of this type are called Dirichlet integrals and are often useful in evaluating, multiple integrals (see Problems 9.24 and 9.25)., OTHER SPECIAL FUNCTIONS, , Many special functions are of importance in science and engineering. Some of these, are given in the following list. Others will be considered in later chapters., 1. Error function., 2. Exponential Integral., 3. Sine Integral., 4. Cosine Integral., 5. Fresnel Sine Integral., 6. Fresnel Cosine Integral., ASYMPTOTIC SERIES OR EXPANSIONS, , Consider the series, (18}, , and suppose that, , (14), , are the partial sums of the series., If, , where f(x) is given, is such that for every n, , (15), then S(x) is called an asymptotic series or expansion of f(x) and we denote this by writing
Page 223 :
CHAP. 9], , GAMMA, BETA AND OTHER SPECIAL FUNCTIONS, , 213, , In practice the series (13) diverges. However, by taking the sum of successive terms, of the series, stopping just before the terms begin to increase, we may obtain a useful, approximation for /(#)• The approximation becomes better the larger the value of x., Various operations with asymptotic series are permissible. For example, asymptotic, series may be multiplied or integrated term by term to yield another asymptotic series., , Solved Problems, THE GAMMA FUNCTION, , 9.1., , Prove: (a) T(n +1) = nr(n), n>0; (b) r(n +1) = n!, n=l,2,8,, (a), , W, , Put M = 1,2,3, ..., , in T(n + 1) = n T(n)., , Then, , r(2) = lr(l) = 1, r(3) = 2r(2) = 2 - 1 = 2!,, In general, T(n+ 1) = n\ if n is a positive integer., , 9.2., , Evaluate each of the following., (a), (6), , (o), (d), , 9.3., , Evaluate each integral., (a), , (&), , Then the integral becomes, , r(4) = 3r(3) = 3 - 2 ! = 3!
Page 224 :
214, , 9.4., , GAMMA, BETA AND OTHER SPECIAL FUNCTIONS, , [CHAP. 9, , Prove that, on letting x = u2. It follows that, , We have, , Changing to polar coordinates (p, <f>) where u = p cos <f>, v — p sin 0, the last integral becomes, , and so, , 9.5., , Evaluate each integral., ( a ) L e t t i n g ys = x , t h e integral becomes, - o, , Let (4 In S)z2 = x and the integral, , (6), becomes, , Let —In x = u. Then x = e~u. When x = 1, u = 0; when * = 0, u = «>. The, , (c), , integral becomes, , 9.6., , Evaluate, , where m,n,a are positive constants., , Letting axn — y, the integral becomes, , 9.7., , Evaluate (a) r(-l/2), (6) r(-5/2)., We use the generalization to negative values defined by r(n) = 'm, , '., , (a) Letting, (6) Letting, , using (a)., , Then, , 9.8., , Prove that, Letting, integral becomes, , where n is a positive integer and m > —1., t h e integral becomes, b, , this last
Page 225 :
CHAP. 9], , 9.9., , GAMMA, BETA AND OTHER SPECIAL FUNCTIONS, , 215, , Prove that, Let, , Then, sin /?X d\, , Thus, , (1), , Integration with respect to /J yields, , or, , (2), on letting x = «X2, so that, , But, from (2), C = -\fw/2^fa., , Thus as required,, , 9.10. A particle is attracted toward a fixed point O with a force inversely proportional to, its instantaneous distance from O. If the particle is released from rest, find the time, for it to reach O., At time t = 0 let the particle be located on the x axis at x — a > 0 and let O be the origin., Then by Newton's law, (1), where m is the mass of the particle and k > 0 is a constant of proportionality., Let, , the velocity of the particle., , Then, , and (1) becomes, , (2), upon integrating. Since v = 0 at x — a, we find c = fc In a. Then, (3), , where the negative sign is chosen since x is decreasing as t increases. We thus find that the time, T taken for the particle to go from x = a to x = 0 is given by, , w, Letting In a/x = u or x = ae~u, this becomes, , THE BETA FUNCTION, 9.11. Prove that (a) B(m, n) = E(n, m), (b), (a) Using the transformation x = 1 — y, we have
Page 226 :
216, , GAMMA, BETA AND OTHER SPECIAL FUNCTIONS, (6) Using the transformation, , * = sin2 8, we have, , 9.12. Prove that, Letting, , we have, , S i m i l a r l y , T h e nen, , Transforming to polar coordinates, x = p cos </>, y = p sin <t>,, , using the results of Problem 9.11., , Hence the required result follows., , The above argument can be made rigorous by using a limiting procedure., , 9.13. Evaluate each of the following integrals., (a), (»), , («), , 9.14. Show that, This follows at once from Problems 9.11 and 9.12., , [CHAP. 9
Page 227 :
CHAP. 9], , GAMMA, BETA AND OTHER SPECIAL FUNCTIONS, , 9.15. Evaluate, , (a), , (b), , 217, , (c), , (a) Let 2m -1 = 6, 2w - 1 = 0, i.e. m = 7/2, w = 1/2, in Problem 9.14., Then the required integral has the value, (6) Letting 2m — 1 = 4, Zn — 1 = 5, the required integral has the value, (c) The given integral =, Thus letting 2m — 1 = 0, 2n — 1 = 4 in Problem 9.14, the value is, , 9.16. Prove, , if p is an even posi-, , 2 • 4 • 6 • • • ( » — 1} f, tive integer, (b) —1.3.5...*), * V *s an °<^<J P°sitive integer., From Problem 9.14 with 2m — 1 = p, 2w — 1 = 0, we have, , (a) Up — 2r, the integral equals, , (6) If p = 2r + 1, the integral, , equals, , In both cases, , 9.17. Evaluate, , (a), , as seen by letting e = v/2 — </>., , (b), , (c), , (a) From Problem 9.16 the integral equals * ] 4 ' | = |, | [compare Problem 9.15(«)]., (6) The integral equals, , The method of Problem 9.15(6) can also be used,, (c) The given integral equals, , 9.18. Given, Letting, , and the result follows., , show that, the given integral becomes, , where
Page 228 :
218, , GAMMA, BETA AND OTHER SPECIAL FUNCTIONS, , [CHAP. 9, , 9.19. Evaluate, Let y* = x. Then the integral b e c o m e s b y Problem, 9.18 with P = i-, , by problem, , The result can also be obtained by letting y2 — tan 8., , 9.20. Show that, Letting a3 = 8y or x — 2y1/a, the integral becomes, , 9.21. Prove the duplication formula, Let, Then, Letting 2x = u, we find, , But, , Then since 1 = 3,, and the required result follows., , 9.22. Prove that, We have, , Then, , a), where we have reversed the order of integration and used the integral 16 on page 6., Letting u2 = v in the last integral, we have by Problem 9.18, (2), Substitution of (2) in (1) yields the required result.
Page 229 :
CHAP. 9], , GAMMA, BETA AND OTHER SPECIAL FUNCTIONS, , 219, , STIRLING'S FORMULA, , 9.23. Show that for large n, n\ = \/2irnn"e-n approximately., We have, , W, The function n In x — x has a relative maximum for x = n, as is easily shown by elementary, calculus. This leads us to the substitution x = n + y. Then (1) becomes, (2), , Up to now the analysis is rigorous. The following procedures in which we proceed formally, can be made rigorous by suitable limiting procedures, but the proofs become involved and we shall, omit them., In (2) use the result, , (S), with x — y/n. Then on letting y = \/n v, we find, , (4), When n is large a close approximation is, (*>, , It is of interest that from (4) we can also obtain the miscellaneous result 4 on page 211 l[see, Problem 9.38]., , DIRICHLET INTEGRALS, 9.24. Evaluate, , where V is the region in the first octant, bounded by the sphere x2 + yz + za = 1 and, the coordinate planes., Let *2 = u, y* = v, zz = w. Then, , where ^ is the region in the uvw space bounded, by the plane u + v + w = 1 and the uv, vw and, uw planes as in Fig. 9-2. Thus, , Fig 9.2, , (*)
Page 230 :
220, , GAMMA, BETA AND OTHER SPECIAL FUNCTIONS, , [CHAP. 9, , Letting v = (1 — u)t, we have, , so that (2) becomes, (3), , where we have used (y/2) r(y/2) = r(y/2 + 1)., The integral evaluated here is a special case of the Dirichlet integral (12), page 212. The, general case can be evaluated similarly., 9.25., , Find the mass of the region bounded by a;2 + y2 + z2 = a2 if the density is a — x2y2z2., The required mass = 8, , where V is the region in the first octant bounded, , by the sphere »2 + j/2 + z2 = a2 and the coordinate planes., In the Dirichlet integral (12), page 212, let b = c — a, p = q = r = 2 and a = /? = y = 3., Then the required result, is, , SPECIAL FUNCTIONS AND ASYMPTOTIC EXPANSIONS, , 9.26. (a) Prove that if x > 0, p > 0, then, where, , (b) Prove that, (c) Explain the significance of the result in (b)., (a) Integrating by parts, we have, , Similarly, , so that, , By continuing in this manner the required result follows., (6)
Page 231 :
CHAP. 9], , GAMMA, BETA AND OTHER SPECIAL FUNCTIONS, , 221, , since, , (c) Because of the results in (6), we can say that, , to, i.e. the series on the right is the asymptotic expansion of the function on the left., , 9.27. Show that, , erf (x), , We have, , erf (*), , Now from equation (1) of Problem 9.26 we have on letting p = 1/2 and replacing te by «2,, , which gives the required result., , Supplementary Problems, GAMMA FUNCTION, 9.28., , Evaluate, , (a), , 9.29., , Evaluate, , (a), , 9.30., , Find, , 9.31., , Show that, , 9.32., , Prove that, , 9.33., , Evaluate, , (a), , 9.34., , Evaluate, , (a), , 9.35., , Prove that, , 9.36., , Prove that if m is a positive integer,, , (a), , (6), , (c), (6), , (6), , (c), (c), , (6), , where m = 0,1,2,3, ..., , (c)
Page 232 :
222, 9.37., , GAMMA, BETA AND OTHER SPECIAL FUNCTIONS, Prove that, , [CHAP. 9, , in x dx is a negative number (it is equal to —y, where y = 0.577215..., , is called Euler's constant)., 9.38., , Obtain the miscellaneous result 4 on page 211 from the result (4) of Problem 9.23., [Hint: Expand e"3''3^™' + ••• in a power series and replace the lower limit of the integral by, -».], , BETA FUNCTION, , 9.39., , Evaluate, , (a) B(3,5),, , 9.40., , Find, , 9.41., , Evaluate, , 9.42., , Prove that, , 9.43., , Evaluate, , (a), , 9.44., , Evaluate, , (a), , 9.45., , Prove that, , 9.46., , Prove that, , 9.47., , Prove that, , 9.48., , Prove that, , (a), , (6) B(3/2,2),, , (c) B(l/3,2/3)., , (6), (a), , (c), (6), , «, , (a), , (b), (6), , (b), , where a, b > 0., , [Hint: Differentiate with respect to 6 in Problem 9.47.], DIRICHLET INTEGRALS, , 9.49., , Find the mass of the region in the xy plane bounded by x + y = 1, * = 0, y = 0 if the density is, , 9.50. Find the mass of the region bounded by the e l l i p s o i d i f the density varies as the, square of the distance from its center., , if the density varies as the, , 9.51., , Find the volume of the region bounded by, , 9.52., , Find the centroid of the region in the first octant bounded by, , 9.53., , Show that the volume of the region bounded by xm + ym + zm = am, where m > 0, is given by
Page 233 :
CHAP. 9], 9.54., , GAMMA, BETA AND OTHER SPECIAL FUNCTIONS, , 223, , Show that the centroid of the region in the first octant bounded by xm + y™ + «"> = a™, where, m > 0, is given by, , SPECIAL FUNCTIONS AND ASYMPTOTIC EXPANSIONS, 9.55., , Show that, , 9.56., , Obtain the asymptotic expansion, , 9.57., , Show that, , 9.58., , Obtain the asymptotic expansions, , («), (&), , 9.59. Show that J* ^-dx =, , and derive a similar result for, , 9.60. Show that f sinx*dx = f cos»2<te = ^\l%-., , Answers to Supplementary Problems, 9.28., , (a) 30,, , (6) 16/105,, , (c), , 9.41. (a) 12*-,, , (6) v, , 9.2 (a)24, (6)^-, ( c ) 9 ' 43 - <«>3lr/256- <6)5)r/8, 9.30. (a) £r(£), (6)^2, (c), 9.33., , (a) 24,, , (6) -3/128,, , 9.34., , (a) (16V^)/105,, , 9.39., , (a) 1/105,, , 9.40., , (a) 1/60, (6), , 9.44., , (a) 16/15,, , 9.49., , T/24, , (6) 8/105, , (c), ._ 9-50.fe = constant of proportionality, , (6) -3r(2/3), , (6) 4/15, (c), , 9.51. 4W35, 9.52
Page 234 :
Chapter 10, Bessel Functions, , BESSEL'S DIFFERENTIAL EQUATION, , Bessel functions arise as solutions of the differential equation, (1), which is called Bessel's differential equation. The general solution of (1) is given by, (*), , The solution Jn(x), which has a finite limit as x approaches zero, is called a Bessel function, of the first kind and order n. The solution Yn(x) which has no finite limit [i.e. is unbounded], as x approaches zero, is called a Bessel function of the second kind and order n or Neumann, function., If the independent variable x in (1) is changed to \x where X is a constant, the resulting, equation is, (3), with general solution, , (4), , BESSEL FUNCTIONS OF THE FIRST KIND, , We define the Bessel function of the first kind of order n as, (5), °r, , (6), , where T(n +1) is the gamma function [Chapter 9]. If n is a positive integer, T(n +1) = n\,, r(l) = 1. For n = 0, (6) becomes, (7), , The series (6) converges for all x. Graphs of Jo(x), and Ji(x) are shown in Fig. 10-1., If n is half an odd integer, Jn(x) can be expressed in terms of sines and cosines. See Problems, 10.4 and 10.7., A function J-n(x), n>0, can be defined by replacing n by -n in (5) or (6). If n is an integer, then we can show that [see Problem 10.3], J-n(x) = (-!)«/„ (»), , (8), 224, , Fig, , !„_!
Page 235 :
CHAP. 10], , BESSEL FUNCTIONS, , 225, , If n is not an integer, /»(*) and J-n(x) are linearly independent, and for this case the, general solution of (1) is, (9), , BESSEL FUNCTIONS OF THE SECOND KIND, , We shall define the Bessel function of the second kind of order n as, , (10), For the case where n = 0,1,2,3,... we obtain the following series expansion for Yn(x)., , (11), where y = .5772156... is Euler's constant and, (12), GENERATING FUNCTION FOR Jn(x), , The function, , (18), , is called the generating function for Bessel functions of the first kind of integral order., It is very useful in obtaining properties of these functions for integer values of n which, can then often be proved for all values of n., RECURRENCE FORMULAS, , The following results are valid for all values of n., 1., 2., 3., 4., 5., 6., , If n is an integer these can be proved by using the generating function. Note that results, 3 and 4 are equivalent respectively to 5 and 6., The functions Yn(x) satisfy exactly the same results as those above, where Yn(x) replaces Jn(x).
Page 236 :
BESSEL FUNCTIONS, , 226, , [CHAP. 10, , FUNCTIONS RELATED TO BESSEL FUNCTIONS, 1. Hankel Functions of First and Second Kinds are defined respectively by, 2. Modified Bessel Functions. The modified Bessel function of the first kind of order n, is defined as, (14), If n is an integer,, but if n is not an integer, ln(x) and I-n(x) are linearly independent., , (15), , The modified Bessel function of the second kind of order n is defined as, (16), , These functions satisfy the differential equation, (17), , and the general solution of this equation is, (18), , or if n + 0,1,2,3, ..., , (19), , 3. Ber, Bei, Ker, Kei Functions. The functions Bern(a;) and Bein(a;) are the real and, imaginary parts of Jn(is/2x) where t372 = e3lri/4 = (\/2/2)(l — i), i.e., Jn(i3/2x) = Eern(x)+iEein(x), (20), The functions Kern(a;) and Kein(a;) are the real and imaginary parts of e~nMZKn(iV2x), where i1'2 = e™" = (v^/2)(l + t), i.e., (21), The functions are useful in connection with the equation, (22), , which arises in electrical engineering and other fields. The general solution of this, equation is, (23), EQUATIONS TRANSFORMED INTO BESSEL'S EQUATION, , The equation, (24), , where k, a, r, /? are constants, has the general solution, where, [seepage 76]., , (**), If « = 0 the equation is solvable as an Euler or Cauchy equation
Page 237 :
CHAP. 10], , BESSEL FUNCTIONS, , 227, , ASYMPTOTIC FORMULAS FOR BESSEL FUNCTIONS, For large values of x we have the following asymptotic formulas, , (26), , ZEROS OF BESSEL FUNCTIONS, We can show that if n is any real number, Jn(x) = 0 has an infinite number of roots, which are all real. The difference between successive roots approaches -n- as the roots, increase in value. This can be seen from (26). We can also show that the roots of Jn(x) — 0, lie between those of Jn-i(x) = 0 and Jn+i(x) = 0. Similar remarks can be made for Yn(x)., , ORTHOGONALITY OF BESSEL FUNCTIONS, If A and [i are two different constants, we can show [see Problem 10.21] that, , (27), while [see Problem 10.22], (28), , From (27) we can see that if A and ^ are any two different roots of the equation, , (29), where R and S are constants, then, , (30), which states that the functions, and, are orthogonal in (0,1). Note that, as special cases of (29) we see that A and /n can be any two different roots of Jn(x) = 0 or, Jn(x) — 0. We can also say that the functions Jn(\x), Jn([j.x) are orthogonal with respect to, the density function x., , SERIES OF BESSEL FUNCTIONS, As in the case of Fourier series, we can show that if f(x) satisfies the Dirichlet conditions [page 183] then at every point of continuity of f(x) in the interval 0 < x < 1 there will, exist a Bessel series expansion having the form, , (31), where \\, Aa, As, ... are the positive roots of (29) with R/S § 0, S ¥* 0 and, , (32), At any point of discontinuity the series on the right in (31) converges to £[/(# + 0) + f(x — 0)], which can be used in place of the left side of (31).
Page 238 :
228, , BESSEL FUNCTIONS, , [CHAP. 10, , In case S = 0 so that Ai, A.2,... are the roots of Jn(x) - 0,, (33), , If R = 0 and « = 0, then the series (31) starts out with the constant term, (84), , Solved Problems, BESSEL'S DIFFERENTIAL EQUATION, , 10.1. Use the method of Frobenius to find series solutions of Bessel's differential equation, x*y" + xy' + (x2 - ri*)y = 0., Assuming a solution of the form y — 2 ckxk+® where k goes from — <*> to °° and ck = 0 for, fe < 0, we have, , Then by addition,, and since the coefficients of xk+& must be zero, we find, (1), 2, , 2, , Letting fe = 0 in (1) we obtain, since c_ 2 = 0, the indicial equation (/3 — n )e0 = 0; or assuming, e0 »* 0, ft2 = w2. Then there are two cases, given by /? = — n and /? = n. We shall consider, first the case /? = n and obtain the second case by replacing n by —n., Case 1,, , /} — n., , In this case (1) becomes, , (2), , Putting fe = 1,2, 3,4, ... successively in (2), we have, , Thus the required series is, (*)•, Case 2,, , 0 = — n., , Or replacing n by —n in Case 1, we find, U)
Page 239 :
CHAP. 10], , BESSEL FUNCTIONS, , 229, , Now if n = 0, both of these series are identical. If n = 1,2, ... the second series fails to exist., However, if n¥> 0,1,2, ... the two series can be shown to be linearly independent and so for this, case the general solution is, , (5), , The cases where w = 0,1,2,3, ..., , are treated later [see Problems 10.15 and 10.16]., , BESSEL FUNCTIONS OF THE FIRST KIND, 10.2. Using the definition (5) of Jn(x) given on page 224, show that if n ¥= 0,1,2, ... then, the general solution of Bessel's equation is y = AJn(x) +BJ-n(x)., Note that the definition of Jn(x) on page 224 agrees with the series of Case 1 in Problem 10.1,, apart from a constant factor depending only on n. It follows that (5) can be written y = AJn(x) +, BJ_n(x) for the cases n ¥= 0,1,2,, , 10.3. (a) Prove that /_„(«) = (-!)"/„(x) for « = 1,2,3, ..., (6) Use (a) to explain why AJn(x) + BJ-n(x) is not the general solution of Bessel's, equation for integer values of n., (a) Replacing n by — n in (5) or the equivalent (6) on page 224, we have, , Now since r(—n + r + 1) is infinite for r = 0,1, ..., n — 1, the first sum on the right is zero., Letting r = n + k in the second sum, it becomes, , (6) From (a) it follows that for integer values of n, J-n(x) and Jn(x) are linearly dependent and, so AJn(x) + BJ-n(x) cannot be a general solution of Bessel's equation. If n is not an integer,, then we can show that J-n(x) and Jn(x) are linearly independent so that AJn(x) + BJ^n(x), is a general solution [see Problem 10.10]., , 10.4., , Prove (a), (a), , w
Page 240 :
230, , BESSEL FUNCTIONS, , 10.5. Prove (a), , (&), , [CHAP. 10, , for all n., , (a), , (6), , 10.6. Prove (a), , (6) ,, , for all n., , From Problem 10.5(a),, or, , (1), , From Problem 10.5(6),, or, , (2), , (a) Adding (1) and (2) and dividing by Zx gives, (6) Subtracting (2) from (1) and dividing by « gives, , 10.7. Show that (a), (b), (a) From Problems 10.6(6) and 10.4 we have on letting n — 1/2,, , (6) From Problems 10.6(6) and 10.4 we have on letting n = — -J,, , (b), , 10.8.Evaluatetheintegrals(a)(6), From Problem 10.5,, (a), (6), , Then, Then
Page 241 :
CHAP. 10], , BESSEL FUNCTIONS, , 231, , 10.9. Evaluate (a), (a) Method 1., , Method 2., , Integration by parts gives, , We have, using J^x) = -J'0(x) [Problem 10.27(6)],, , Then, , (&), , Then, , The integral, , cannot be obtained in closed form., , In general, , be obtained in closed form if p + q^ 0 and p + q is odd, where p and q are integers., however, p + q is even the result can be obtained in terms of, , can, If,
Page 242 :
232, , BESSEL FUNCTIONS, , [CHAP. 10, , 10.10. (a) Prove that, (6) Discuss the significance of the result of (a) from the viewpoint of the linear, dependence of Jn(x) and J-n(x),, (a) Since /„(*), and /-„(*), abbreviated /„, J_ n respectively, satisfy Bessel's equation, we have, Multiply the first equation by J_n, the second by /„ and subtract. Then, which can be written, , or, , Integrating, we, , find, , (1), , To determine e use the series expansions for /„ and J-n to obtain, , and then substitute in (1). We find, using the result 1, page 211. This gives the required result., (6) The expression J'nJ-n — J'-nJn in (a) is the Wronskian of Jn and «/_„. If n is an integer, we, see from (a) that the Wronskian is zero so that Jn and «/_„ are linearly independent as is also, clear from Problem 10.3(a). On the other hand if n is not an integer, they are linearly independent since in such case the Wronskian differs from zero., , GENERATING FUNCTION AND MISCELLANEOUS RESULTS, , 10.11. Prove that, Wo Vioiro, , Let r — k = n so that n varies from —» to ». Then the sum becomes, n, , 10.12. Prove, , (a), (&), , Let t = e*» in Problem 10.11. Then, , where we have used Problem 10.3(o). Equating real and imaginary parts gives the required results.
Page 243 :
CHAP. 10], , BESSEL FUNCTIONS, , 233, , 10.13. Prove, Multiply the first and second results of Problem 10.12 by cosw* and sinne respectively and, integrate from 0 to v using, , Then if w is even or zero, we have, , or on adding,, , Similarly if n is odd,, , and by adding,, Thus we have the required result whether n is even or odd, i.e. n = 0,1,2, ...., , 10.14. Prove the result of Problem 10.6(&) for integer values of n by using the generating,, function., Differentiating both sides of the generating function with respect to t, we have, omitting the, limits — oo to °° for n, , or, , i.e., This can be written as, , or, , i.e., Since coefficients of tn must be equal, we have, from which the required result is obtained on replacing n by n — 1., , BESSEL FUNCTIONS OF THE SECOND KIND, 10.15. (a) Show that if n is not an integer, the general solution of Bessel's equation is
Page 244 :
BESSEL FUNCTIONS, , 234, , [CHAP. 10, , (6) Explain how you can use part (a) to obtain the general solution of Bessel's equation in case n is an integer., (a) Since J_ n and «/„ are linearly independent, the general solution of Bessel's equation can be, written, and the required result follows on replacing the arbitrary constants c1( c2 by E, F where, , Note that we define the Bessel function of the second kind if n is not an integer by, , (b) The expression, becomes an "indeterminate" of the form 0/0 for the case when n is an integer. This is because, for an integer n we have cosnir = (—1)" and J-n(x) = (—l)nJn(#) [see Problem 10.3]. This, "indeterminate form" can be evaluated by using L'Hospital's rule, i.e., , This motivates the definition (10) on page 225., , 10.16. Use Problem 10.15 to obtain the general solution of Bessel's equation for n = 0., In this case we must evaluate, (J), , Using L'Hospital's rule [differentiating the numerator and denominator with respect to p] we find, for the limit in (1), , where the notation indicates that we are to take the partial derivatives of Jp(x) and J-p(x) with, respect to p and then put p — 0. Since dJ_p/d(—p) — —dJ_p/dp, the required limit is also equal to, , To obtain dJp/dp we differentiate the series, with respect to p and obtain, (2), , Now if we let, , then, ln<, , so that differentiation with respect to p gives, Then for p — 0, we have, (*)
Page 245 :
CHAP. 10], , BESSEL FUNCTIONS, , 235, , Using (2) and (3), we have, , where the last series is obtained on using the result 6 on page 224. This last series is the series, for Y0(x). We can in a similar manner obtain the series (11), page 225, for Yn(x) where n is an, integer. The general solution if n is an integer is then given by y = CjJre(a;) + c2Yn(x)., , FUNCTIONS RELATED TO BESSEL FUNCTIONS, 10.17. Prove that the recurrence formula for the modified Bessel function of the first kind, In(x) is given by, Prom Problem 10.6(6) we have, , (1), Replace x by ix to obtain, (2), n, , Now by definition In(x) = i~ Jn(ix), , n, , or Jn(ix) — i ln(x), , so that (2) becomes, , Dividing by i™ +1 then gives the required result., , 10.18. If n is not an integer, show that, (a), , (6), , (a) By definition of H™(x) and Yn(x) [see pages 226 and 225 respectively] we have, , (6) Since fl^2)(aj) = JH(x) — iYn(x), we find on replacing i by —i in the result of part (a),, , 10.19. Show that, , (a), (b)
Page 246 :
236, , BESSEL FUNCTIONS, , [CHAP. 10, , We have, , and the required result follows on noting that J0(ia'2x) = Ber0(a;) + i Beic («) and equating real, and imaginary parts. Note that sometimes the subscript zero is omitted in Ber0 (x) and Bei0 (x)., , EQUATIONS TRANSFORMED INTO BESSEL'S EQUATION, 10.20. Find the general solution of the equation xy" + y' + ay = 0., The equation can be written as x2y" + xy' + axy = 0 and is a special case of equation (24),, page 226, where k = 0, a = \fa, r = 1/2, /J = 0. Then the solution as given by (25), page 226, is, , ORTHOGONALITY OF BESSEL FUNCTIONS, 10.21. Prove that, , if A *• /*., , From (S) and (4), page 224, we see that yt — Jn(*.x), equations, , and y^ — Jn(nx), , are solutions of the, , Multiplying the first equation by y2, the second by y1 and subtracting, we find, which on division by x can be written as, , or, , Then by integrating and omitting the constant of integration,, or using yt = Jn(\x), j/2 = Jn(^), , and dividing by /i2 — X2 9* 0,, , Thus, which is equivalent to the required result., , 10.22. Prove that, Let 11 -» \ in the result of Problem 10.21. Then using L'Hospital's rule, we find
Page 247 :
CHAP. 10], , BESSEL FUNCTIONS, , 237, , But since XV^'(X) + \J'n(\) + (X 2 -n 2 ) J n (X) = 0, we find on solving for /£(X) and substituting,, , 10.23. Prove that if A. and ^ are any two different roots of the equation RJn(x) + SxJ'n(x) = 0, where R and S are constants, then, , i.e., , and, , are orthogonal in (0,1)., , Since X and p are roots of RJn(x) + SxJn(x), RJn(\) + S\Jn(\), , = 0,, , = 0, we have, RJM + Spj'M, , = 0, , (1), , Then if R ¥> 0, S ¥> 0 we find from (1),, and so from Problem 10.21 we have the required result, , In case R = 0, S ¥* 0 or R ¥* 0, S = 0, the result is also easily proved., , SERIES OF BESSEL FUNCTIONS, 10.24 If, , where Xp, p = 1,2,3, ..., are the positive roots, , of Jn(x) = 0, show that, , Multiply the series for f(x) by x Jn(\kx) and integrate term by term from 0 to 1. Then, , where we have used Problems 10.22 and 10.23 together with the fact that «/n(^k) = °- ** follows, that, , To obtain the required result from this, we note that from the recurrence formula 3, page 225,, which is equivalent to the formula 6 on that page, we have, or since Jn(\k) = 0,, , 10.25. Expand f(x) = 1 in a series of the form, , i, for 0 < x < 1, if \p, p = 1,2,3, ..., are the positive roots of J0(x) = 0.
Page 248 :
238, , BESSEL FUNCTIONS, , [CHAP. 10, , From Problem 10.24 we have, , where we have made the substitution, 10.8(a) with n — 1., , v = \px in the integral and used the result of Problem, , Thus we have the required series, e, , f, , t, , ,, , -, , », , which can be written, , Supplementary Problems, BESSEL'S DIFFERENTIAL EQUATION, 10.26. Show that if x is replaced by Xa; where X is a constant, then Bessel's equation x2y" + xy' +, (x2 — n2)y — 0 is transformed into x2y" + xy' + (\zx2 — n2)y = 0., BESSEL FUNCTIONS OF THE FIRST KIND, 10.27. (a) Show that, , and verify that the interval of conver-, , gence is — «> < x < °°., (b) Show that, (c) Show that, 10.28. Evaluate (a) Jz/z(x) and (b) J_5/2(x) in terms of sines and cosines., 10.29. Find J3(x) in terms of J9(x) and J^x)., 10.30. Prove that, , (a), (&), , and generalize these results., 10.31. Evaluate, , (a), , (b), , 10.32. Evaluate, , (a), , (b), , 10.33. Evaluate, 10.34. Verify directly the result, , (c)
Page 249 :
CHAP. 10], , BESSEL FUNCTIONS, , 239, , GENERATING FUNCTION AND MISCELLANEOUS RESULTS, 10.35. Use the generating function to prove that J'n(x) = ^[Jn^1(x) + Jn + 1(x)], is an integer., 10.36., , for the case where n, , Use the generating function to work Problem 10.30 for the case where n is an integer., , 10.37. Show that, , 10.38., , Show that, , 10.39., , Show that, , 10.40. Show that, , (a) 1 = J0(x) + 2J2(x) + 2 J 4 ( » ) + • • •, (b) /!<*) - J3(x) + Js(x) - J7(x) + ••• = sinx., , (a), (6), , 10.41., , Show that, , 10.42. Show that, , and thus find, (6) <{/j(6t)},, , (a) 4{J0(bt)},, 10.43., , (c) Jd{Jn(&«)},, , n = 0,1,2, ..., , Show that, , 10.44. Prove that |Jn(&)| = 1 f°r »U integers n. Is the result true if n is not an integer?, BESSEL FUNCTIONS OF THE SECOND KIND, 10.45. Show that (a) IW*) =25.r, n (*)-y»-i<*), (6) F» = ^(Yn^(x) - Yn+l(x)}., tfj, 10.46. Explain why the recurrence formulas for Jn(x) on page 225 hold if Jn(x) is replaced by Yn(x)., 10.47. Prove that Y^x) = -Y^x)., 10.48. Evaluate, 10.49. Prove that, 10.50., , Evaluate, , (a) Y1/2(x),, , (b) r_ 1/2 (w)., , Jn(x) Y'n(x) - J'n(x) Yn(x) (a), , 2/wx., , (b), , (c), , 10.51. Prove the result (11), page 225., , FUNCTIONS RELATED TO BESSEL FUNCTIONS, 10.52. Show that, 10.53., , Show that, , 10.54. Show that, , (a) l'n(x) = ${In_i(x) + /n+J(*)},, e(.x/2->(.t+i/f>, , -, , CO, , ^, In(x)tn, , — 00, , (6) xl'n(x) = xln^(x) - nln(x)., , is the generating function for In(x)., , 10.55. Show that, 10.56. Show that, , (a) sinha; = 2[/!(*) + Is(x) + • • •], (b) cosh* = 70(») + 2[72(*) + 74(«) + • • • ] .
Page 250 :
240, , BESSEL FUNCTIONS, , 10.57. Show that, , (a), , [CHAP. 10, , (6), 2t7, , 10.58. (a) Show that Kn+1(x) = K^^x) + —Kn(x). (b) Explain why the functions Kn(x) satisfy recur00, rence formulas which are the same as those for In(x) with In(x) replaced by Kn(x)., 10.59. Give asymptotic formulas for (a) H™(x), (6) H™(x)., 10.60. Show that, , (a), (b), , 10.61., , Show, , that, , EQUATIONS TRANSFORMED INTO BESSEL'S EQUATION, 10.62. Prove that (25), page 226, is a solution of (24)., , 10.63. Solve, , 4«j/" + 4y' + y = 0., , 10.64. Solve, , (a) xy" + 2y' + xy = 0,, , 10.65., , y" + e2xy = 0., , Solve, , (6) y" + xzy = 0., , [Hint. Let ex = u.}, , 10.66. (a) Show by direct substitution that y — J9 (2^fx), the general solution., , is a solution of xy" + y' + y — 0 and (6) write, , 10.67. (a) Show by direct substitution that y = ^Ji/^x3'2), the general solution., , is a solution of y" + xy = 0 and (6) write, , 10.68. (a) Show that Bessel's equation x2y" + xy' + (x2 — n2)y = 0 can be transformed into, , where y = u/-\fx. (b) Discuss the case where n = ±1/2., (6) Discuss the case where x is large and explain the connection with the asymptotic formulas, on page 227., ORTHOGONAL SERIES OF BESSEL FUNCTIONS, 10.69. Complete Problem 10.23, page 237, for the cases, 10.70. Show that, 10.71., , Prove the results (31) and (32), page 227., , 10.72. Show that, where Xp are the positive roots of «/0(M = 010.73. Show that, where Xp are the positive roots of Ji(X) — 0., 10.74. Show that, where Xp are the positive roots of Ji(\) — 0., , (a) R * 0, S = 0, (6) R = 0, S ¥= 0.
Page 251 :
CHAP. 10], , 10.75. If, , BESSEL FUNCTIONS, , where, , Compare with Parseval's identity for Fourier series., 10.76. Use Problems 10.73 and 10.75 to show that, , where Xp are the positive roots of </ 0 (\) = 0., , Answers to Supplementary Problems, 10.28., , (a), , (6), , 10.29., 10.31. (a), 10.32., , (c), , (a), W, , 10.33., 10.42., , (a), , (c), , 10.48. (a), 10.50., , (a), (6), , 10.59. (a), , (c), , (6), , 10.63., 10.64., 10.65., , (a), , (6), , 241
Page 252 :
Chapter 11, , LEGENDRE'S DIFFERENTIAL EQUATION, , Legendre functions arise as solutions of the differential equation, W, which is called Legendre's differential equation. The general solution of (1) in the case where, w = 0,1,2,3,... is given by, , where Pn(x) are polynomials called Legendre polynomials and Qn(x) are called Legendre, functions of the second kind which are unbounded at x = ±1., LEGENDRE POLYNOMIALS, , The Legendre polynomials are defined by, («), , Note that Pn(x) is a polynomial of degree n. The first few Legendre polynomials are as, follows:, 4, 1.4., 5, 2.5., 3. 6., , 6, , In all cases, '., The Legendre polynomials can also be expressed by Rodrigue's formula given by, (3), GENERATING FUNCTION FOR LEGENDRE POLYNOMIALS, , The function, , u\, , is called the generating function for Legendre polynomials and is useful in obtaining their, properties., RECURRENCE FORMULAS, 1., 2., , 242
Page 253 :
CHAP. 11], , LEGENDRE FUNCTIONS AND OTHER ORTHOGONAL FUNCTIONS, , 243, , LEGENDRE FUNCTIONS OF THE SECOND KIND, If \x\ < 1, the Legendre functions of the second kind are given by the following according as n is even or odd respectively:, , For n > 1, these coefficients are taken so that the recurrence formulas for Pn(x) above apply, also to Qn(x)., , ORTHOGONALITY OF LEGENDRE POLYNOMIALS, The following results are fundamental:, (7), (*), , The first shows that any two different Legendre polynomials are orthogonal in the interval, -1< x < 1., , SERIES OF LEGENDRE POLYNOMIALS, If f(x) satisfies the Dirichlet conditions [Page 183], then at every point of continuity of, f(x) in the interval —Kx<l there will exist a Legendre series expansion having the form, , (9), where, , (10), , At any point of discontinuity the series on the right in (9) converges to %[f(x + 0) + f(x - 0)], which can be used to replace the left side of (9)., ASSOCIATED LEGENDRE FUNCTIONS, The differential equation, , (11), is called Legendre's associated differential equation. If m — 0 this reduces to Legendre's, equation (1). Solutions to (11) are called associated Legendre functions. We consider the, case where m and n are non-negative integers. In this case the general solution of (11) is, given by, (jf), where P™(x) and Q™(x) are called associated Legendre functions of the first and second, kinds respectively. They are given in terms of the ordinary Legendre functions by, , (13), (14), Note that if m > n, P™(x) - 0.
Page 254 :
244, , LEGENDRE FUNCTIONS AND OTHER ORTHOGONAL FUNCTIONS, , [CHAP. 11, , As in the case of Legendre polynomials, the Legendre functions P™ (x) are orthogonal in, -!<«<!, i.e., (15), (16), , We also have, Using these, we can expand a function f(x) in a series of the form, , (17), OTHER SPECIAL FUNCTIONS, The following special functions are some of the important ones arising in science and, engineering., 1. Hermite polynomials., differential equation, , These polynomials, denoted by Hn(x), are solutions of Hermite's, , y" - 2xy' + 2ny = 0, , (18), , The polynomials are given by a corresponding Rodrigue's formula, (19), , Their generating function is given by, , (20), and they satisfy the recursion formulas, (21), , (22), The important results, , (23), (*•*), enable us to expand a function into a Hermite series of the form, (25), , where, , ^6), , 2. Laguerre polynomials. These polynomials, denoted by Ln(x), are solutions of Laguerre's, differential equation, (27), The polynomials are given by the Rodrigue's formula, (28)
Page 255 :
CHAP. 11], , LEGENDRE FUNCTIONS AND OTHER ORTHOGONAL FUNCTIONS, , 245, , Their generating function is given by, (29), , and they satisfy the recursion formulas, (30), (31), , The important results, (32), , (33), enable us to expand a function into a Laguerre series of the form, (34), , where, , (35), , STURM-LIOUVILLE SYSTEMS, , A boundary-value problem having the form, (36), , where ai, a2, 61, 62 are given constants; p(x), q(x), r(x) are given functions which we shall, assume to be diiferentiable and A. is an unspecified parameter independent of x, is called a, Sturm-Liouville boundary-value problem or Sturm-Liouville system., A non-trivial solution of this system, i.e. one which is not identically zero, exists in, general only for a particular set of values of the parameter A. These values are called, the characteristic values, or more often eigenvalues, of the system. The corresponding, solutions are called characteristic functions or eigenfunctions of the system. In general, to each eigenvalue there is one eigenfunction, although exceptions can occur., If p(x), q(x) are real, then the eigenvalues are real. Also the eigenfunctions form an, orthogonal set with respect to the density function r(x) which is generally taken as nonnegative, i.e. r(x) =£ 0. It follows that by suitable normalization the set of functions can be, made an orthonormal set with respect to r(x) in a ^ x ^ b.
Page 256 :
246, , LEGENDRE FUNCTIONS AND OTHER ORTHOGONAL FUNCTIONS, , [CHAP. 11, , Solved Problems, LEGENDRE'S DIFFERENTIAL EQUATION, , 11.1. Use the method of Frobenius to find series solutions of Legendre's differential equation (1 - xz)y" - 2xy' + n(n + l)y = 0., Assuming a solution of the form y = '2,ckxk+& where the summation index fe goes from, —oo to oo and ek = 0 for fc < 0, we have, , Then by addition,, and since the coefficient of xk+& must be zero, we find, (1), , Letting fc = — 2 we obtain, since c_ 2 = 0, the indicial equation 13(13 — I)c0 = 0 or, assuming, c0 ¥= 0, /J = 0 or 1., Case 1, /? = 0., , In this case (1) becomes, (2), , Putting fc = —1,0,1,2,3,... in succession, we find that ct is arbitrary while, , and so we obtain, (3), , Since this leads to a solution with two arbitrary constants, we need not consider Case 2, ft — 1., For an even integer TO S 0, the first of the above series terminates and gives a polynomial, solution. For an odd integer TO > 0, the second series terminates and gives a polynomial solution., Thus for any integer n fe 0 the equation has polynomial solutions. If n = 0,1,2,3, for example,, we obtain from (S) the polynomial, , which are, apart from a multiplicative constant, Legendre polynomials., , LEGENDRE POLYNOMIALS, , 11.2. Derive formula (2), page 242, for the Legendre polynomials., From equation (2) of Problem 11.1 we see that if k = n then c n + 2 = 0 and thus c n+4 = 0,, c n+6 = 0, Then letting k = n — 2, n — 4, ... we find from the equation (2) of Problem 11.1,
Page 257 :
CHAP. 11], , LEGENDRE FUNCTIONS AND OTHER ORTHOGONAL FUNCTIONS, , This leads to the polynomial solutions, , The Legendre polynomials Pn(x) are defined by choosing, , This choice is made in order that Pn(l) = 1., , 11.3. Derive Rodrigue's formula, By Problem 11.2 the Legendre polynomials are given by, , Now integrating this n times from 0 to x, we obtain, , which can be written, , which proves that, , GENERATING FUNCTION, 11.4., , Prove that, Using the binomial theorem, , we have, , and the coefficient of tn in this expansion is, , which can be written as, , i.e. Pn(x)., , The required result thus follows., , 247
Page 258 :
248, , LEGENDRE FUNCTIONS AND OTHER ORTHOGONAL FUNCTIONS, , [CHAP. 11, , RECURRENCE FORMULAS FOR LEGENDRE POLYNOMIALS, 11.5. Prove that, From the generating function of Problem 11.4 we have, (1), Differentiating with respect to t,, Multiplying by 1 - 2xt + t2,, (2), , Now the left side of (2) can be written in terms of (1) and we have, , i.e., Equating the coefficients of t" on each side, we find, which yields the required result., , 11.6., , Given that P0(x) = 1, Pi(x) = x, find (a) P2(x) and (b) P3(x)., Using the recurrence formula of Problem 11.5, we have on letting n = 1,, , Similarly letting n — 2,, , LEGENDRE FUNCTIONS OF THE SECOND KIND, 11.7. Obtain the results (5) and (6), page 243, for the Legendre functions of the second, kind in the case where n is a non-negative integer., The Legendre functions of the second kind are the series solutions of Legendre's equation which, do not terminate. From Problem 11.1, equation (3), we see that if n is even the series which does, not terminate is, , while if n is odd the series which does not terminate is, , These series solutions, apart from multiplicative constants, provide definitions for Legendre, functions of the second kind and are given by (5) and (6) on page 243.
Page 259 :
CHAP. 11], , 11.8., , LEGENDRE FUNCTIONS AND OTHER ORTHOGONAL FUNCTIONS, , 249, , Obtain the Legendre functions of the second kind (a) Qo(x), (b) Qi(x) and (c) Qs(x)., (a) From equation (5), page 243, we have if n — 0,, , where we have used the expansion In (1 + u) = u — u2/2 + w8/3 — tt4/4 + • • • ., (6) From equation (6), page 243, we have if n = 1,, , (c) The recurrence formulas for Qn(x) are identical with those of Pn(x)., , Then from Problem 11.5,, , Putting n = 1, we have on using parts (a) and (6),, , ORTHOGONALITY OF LEGENDRE POLYNOMIALS, , 11.9. Prove that, , 0 if m¥=n., , Since Pm(x),Pn(x) satisfy Legendre's equation,, , Then multiplying the first equation by Pn, the second equation by Pm and subtracting, we find, which can be written, , or, , Thus by integrating we have, , Then since m ¥= n,, , 11.10. Prove that, From the generating function, we have on squaring both sides,
Page 260 :
250, , LEGENDRE FUNCTIONS AND OTHER ORTHOGONAL FUNCTIONS, , [CHAP. 11, , Then by integrating from —1 to 1 we have, , Using the result of Problem 11.9 on the right side and performing the integration on the left side,, , or, , i.e., Equating coefficients of t2™ we have as required, , SERIES OF LEGENDRE POLYNOMIALS, 11.11. If, , fc=0, , show that, , Multiplying the given series by Pm(x) and integrating from —1 to 1, we have on using Problems, 11.9 and 11.10,, , Then as required,, , 11.12. Expand the function, By Problem 11.11,, , Then, , in a series of the forn
Page 261 :
CHAP. 11], , LEGENDRE FUNCTIONS AND OTHER ORTHOGONAL FUNCTIONS, , 251, , etc. Thus, The general term for the coefficients in this series can be obtained by using the recurrence formula, 2 on page 242 and the results of Problem 11.29. We find, , For n even An — 0, while for n odd we can use Problem 11.29(c)., , ASSOCIATED LEGENDRE FUNCTIONS, 11.13. Obtain the associated Legendre functions (a), (a), (6), (c), , 11.14. Verify that Pl(x) is a solution of Legendre's associated equation (11), page 243, for, ra = 2, n = 3., By Problem 11.13,, , o,, , ,, , . Substituting this in the equation, , we find after simplifying,, and so P\(x) is a, , solution., , 11.15. Verify the result (15), page 244, for the functions Pl(x) and Pl(x)., We have from Problem 11.13(a), P\(x) = Sx(l-x*)v*., , Also,, , Then, , 11.16. Verify the result (16), page 244, for the function P\(x)., Since P\(x) = 3*(1 - *2)»/2,, , Now according to (16), page 244, the required result should be, , so that the verification is achieved.
Page 262 :
LEGENDRE FUNCTIONS AND OTHER ORTHOGONAL FUNCTIONS, , 252, , [CHAP. 11, , HERMITE POLYNOMIALS, , 11.17. Use the generating function for the Hermite polynomials to find (a) H0(x), (b) Hi(x),, (c) H2(x), (d) H3(x)., We have, , Now, , Comparing the two series, we have, , 11.18. Prove that H'n(x) = 2nHn-1(x)., Differentiating, , with respect to x,, , or, , Equating coefficients of t" on both sides,, or, , 11.19. Prove that, We have, Then, But, , 11.20. Prove that, We have, Multiplying these,, Multiplying by e~^ and integrating from —« to °°,
Page 263 :
CHAP. 11], , LEGENDRE FUNCTIONS AND OTHER ORTHOGONAL FUNCTIONS, , 253, , Now the left side is equal to, , By equating coefficients the required result follows., The result, can also be proved by using a method similar to that of Problem 11.9., , LAGUERRE POLYNOMIALS, , 11.21. Determine the Laguerre polynomials (a) L0(x), (6) L\(x), (c) Lz(x), (d) L3(x)., We have, , Then, , (a), (6), , (c), (d), , 11.22. Prove that the Laguerre polynomials Ln(x) are orthogonal in (0, «>) with respect to, the weight function e~x., From Laguerre's differential equation we have for any two Laguerre polynomials Lm(x) and, £»(*),, , Multiplying these equations by Ln and Lm respectively and subtracting, we find, , or, Multiplying by the integrating factor, this can be written as, , so that by integrating from 0 to «,, , Thus if m ¥• n,, which proves the required result.
Page 264 :
254, , LEGENDRE FUNCTIONS AND OTHER ORTHOGONAL FUNCTIONS, , [CHAP. 11, , STURM-LIOUVILLE SYSTEMS, 11.23. (a) Verify that the system y" + \y = 0, y(Q) = 0, y(l) = 0 is a Sturm-Liouville system., (6) Find the eigenvalues and eigenfunctions of the system, (c) Prove that the eigenfunctions are orthogonal in (0,1). (d) Find the corresponding set of normalized, eigenfunctions. (e) Expand f(x) = 1 in a series of these orthonormal functions., (a) The system is a special case of (36), page 245, with p(x) = 1, q(x) = 0, r(x) = 1, a = 0, 6 = 1,, <*! = 1, «2 = °> 0i = 1> 02 = ° and tnus is a Sturm-Liouville system., (b) The general solution of y" + \y = 0 is y = A cosVx* + B sin-v/Xw. From the boundary, condition j/(0) = 0 we have A = 0, i.e. y = B sin Vx x. From the boundary condition, j/(l) = 0 we have B sin \/X = 0, so that since B cannot be zero [otherwise the solution will, be identically zero, i.e. trivial] we must have sin Vx = 0. Then \/X = ^TT, X = mV2 where, m = 1,2,3, ... are the required eigenvalues., The eigenfunctions belonging to the eigenvalues X = m2]r2 can be designated by, Bm sin m-n-x, w = 1,2,3, ...., Note that we exclude the value m = 0 or X = 0 as eigenvalue since the corresponding, eigenfunction is zero., (c) The eigenfunctions are orthogonal since, , (d) The eigenfunctions will be orthonormal if, , i.e. if, , taking the, , positive square root. Thus the set V2 sin mirx, m = 1, 2,..., is an orthonormal set., (e) We must find constants c1( c2, ... such that, where f(x) — 1, <f>m(x) = \/2 sin mirx. By the methods of Chapter 8,, , Then the required series [Fourier series] is, , 11.24. Show that the eigenvalues of a Sturm-Liouville system are real., We have, , (1), , (2), Then assuming p(x), q(x), r(x), ttl, «2, plt /J2 are real while X and y may be complex, we have on, taking the complex conjugate,, (S), (4)
Page 265 :
CHAP. 11], , LEGENDRE FUNCTIONS AND OTHER ORTHOGONAL FUNCTIONS, , 255, , Multiplying equation (1) by y, (3) by y and subtracting, we find after simplifying,, , Then integrating from a to 6, we have, (5), , on using the conditions (2) and (4). Since_ r(x) § 0 and is not identically zero in (a, b), the integral, on the left of (5) is positive and so X — X = 0 or X = X so that X is real., , 11.25. Show that the eigenfunctions belonging to two different eigenvalues are orthogonal, with respect to r(x) in (a, 6)., If yl and y2 are eigenfunctions belonging to the eigenvalues X t and X2 respectively,, , (1), (3), (3), <*i2/2(«) + «22/2>) =, , 0., , PiVzQ) + PzVzW, , = 0, , (4), , Then multiplying (1) by yz, (3) by j/t and subtracting, we find as in Problem 11.24,, , Integrating from a to 6, we have on using (2) and (4),, , and since X, , t, , ¥=, , X2, , we, , have, , the, , required, , result, , Supplementary Problems, LEGENDRE POLYNOMIALS, 11.26. Use Rodrigue's formula (3), page 242, to verify the formulas for P0(x), PI(X), ..., Pe(x) on page 242., 11.27., , Obtain the formulas for P±(x) and Ps(x) using the recursion formula., , 11.28., , Evaluate, , 11.29. Show that, , (a), , (6), , (a) P n (l) = 1, , (c), (c) P 2 n-i(0) = 0, , (D), (6) P, for n = 1,2,3, ... using (2), page 242., , 11.30., , p, , Use the generating function to prove that P^+1(cc) — P^-i(x) = (2n +l)P n (c<;)., , 11.31. Prove that, , (a) P»+1(x) - xPfa), , = (n+l)Pn(x),, , (b) xPfa) - P^(x), , = nPn(x)., , 11.32. Show that 2 P, 11.33., , Show that, , (a) P2(cose) = £(1 + 3 cos 20),, , (6) P3(cos«) = 1(3 cos a + 5 cos 30).
Page 266 :
256, , LEGENDRE FUNCTIONS AND OTHER ORTHOGONAL FUNCTIONS, , [CHAP. 11, , LEGENDRE FUNCTIONS OF THE SECOND KIND, 11.34. Prove that the series (5) and (6) on page 243 which are non-terminating are convergent for, -!<«<!., 11.35. Find Q3(«)., 11.36. Write the general solution of (1 - »%" - 2xy' + 2y = 0., , SERIES OF LEGENDRE POLYNOMIALS, 11.37., , Expand or* — Sx2 + x in a series of the form, , 11.38. Expand, nonzero terms., , in a series of the f o r m w r i t i n g the first four, , 11, , 11.39.IfobtainParseval'sidentity, , and illustrate by using the function of Problem 11.37., ASSOCIATED LEGENDRE FUNCTIONS, 11.40. Find, 11.41. Find, 11.42. Verify that the expressions for P\(x) and Q\(x) are solutions of the corresponding differential, equation and thus write the general solution., 11.43. Verify formulas (15) and (16), page 244, for the case where (a) m = 1, n = 1, I = 2, (6) m = 1,, n= 1,1 = 1., 11.44. Obtain a generating function for P™(x)., 11.45. Use the generating function to obtain results (15) and (16) on page 244., 11.46. Show how to expand /(*) in a series of the forn, , and illustrate by using the cases, , 2, , (a) /(«) = x , m — 2 and (6) /(») = x(l — x), m — 1. Verify the corresponding Parseval identity, in each case., HERMITE POLYNOMIALS, 11.47. Use Rodrigue's formula (19), page 244, to obtain the Hermite polynomials H0(x), Hjix), H2(x), H3(x)., 11.48. Use the generating function to obtain the recurrence formula (21) on page 244 and obtain, H2(x), H3(x) given that H0(x) = 1, Ht(x) - 2x., 11.49. Show directly that, , (6), , (a), , 11.50. Evaluate, 11.51. Show that, 11.52., , (a) Show how to expand f(x) in a series of the f o r m a n d, /(*) = «s - So;2 + 2x., , 11.53. If, , obtain Parseval's identity, , and illustrate by using the function of Problem 11.52., , (b) illustrate by using
Page 267 :
CHAP. 11], 11.54., , LEGENDRE FUNCTIONS AND OTHER ORTHOGONAL FUNCTIONS, , Find the general solution of Hermite's differential equation if, , 257, , (a) n — 0, (&) n = 1., , LAGUERRE POLYNOMIALS, 11.55. Find Lt(x) and show that it satisfies Laguerre's equation (27), page 244, for n = 4., 11.56. Use the generating function to obtain the recursion formula (30) on page 245., 11.57., , Use formula (30) to determine L2(x), L3(x) and L4(«) if we define, and 1 when n = 0., , 11.58. Show that, 11.59., , Ln(x) = 0 when, , n = —1, , nLn^(x) = nL'n^(x) - L'n(x)., , Prove the result (33) on page 245., , 11.60. Expand f(x) = xs — Sx2 + 2x in a series of the form, 11.61. Illustrate Parseval's identity for Problem 11.60., 11.62., , Find the general solution of Laguerre's equation if (a) n = 0, (b) n = 1., , STURM-LIOUVILLE SYSTEMS, 11.63. (a) Verify that the system y" + \y - 0, y'(0) = 0, y(l) = 0 is a Sturm-Liouville system., (b) Find the eigenvalues and eigenfunctions of the system., (e) Prove that the eigenfunctions are orthogonal and determine the corresponding orthonormal, functions., 11.64. Show how to write (a) Legendre's equation, (b) Hermite's equation, (e) Bessel's equation and, (d) Laguerre's equation in Sturm-Liouville form and discuss the significance., , Answers to Supplementary Problems, 11.28. (a) 0, , (6) 2/5, , (e) 0, , 11.36., 11.37., 11.38., , 11.40., , (a) 3(1 -x2), , 11.50., , if n — 0,, , 11.52., , (6), , 11.54., , (a), , (c) 105*(1 - *2)3/2, , (b), if n - 2,, , 0 otherwise, , (b), , 11.55. L4(x) = 24 - 96* + 72*2 - 16*3 + ^4, 11.60. 2L0(x) - SL^x) + 6L2(x) 11.62., , (a), , 11.63. ^[2m-l]V2, cos(m-%)*x, , Ls(x), , (b), m = l,2,3, ..., , 1 .64. ( a ) (, , (&), , (d)
Page 268 :
Chapter 12, , SOME DEFINITIONS INVOLVING PARTIAL DIFFERENTIAL EQUATIONS, , A partial differential equation is an equation containing an unknown function of two or, more variables and its partial derivatives with respect to these variables., The order of a partial differential equation is that of the highest ordered derivative, present., Example 1, , ., i, partial, , s, a partial differential equation o f order two, o r a second order, differential equation., , A solution of a partial differential equation is any function which satisfies the equation, identically., The general solution is a solution which contains a number of arbitrary independent, functions equal to the order of the equation., A particular solution is one which can be obtained from the general solution by particular choice of the arbitrary functions., Example 2. As seen by substitution, u = x*y - \xyi + F(x) + G(y) is a solution of the partial, differential equation of Example 1. Because it contains two arbitrary independent, functions F(x) and G(y), it is the general solution. If in particular F(x) = 2 sin x,, G(y) = Sj/4 — 5, we obtain the particular solution u = x*y — ^xy2 + 2 sin « + Sy* — 5., , A singular solution is one which cannot be obtained from the general solution by particular choice of the arbitrary functions., A boundary-value problem involving a partial differential equation seeks all solutions of, a partial differential equation which satisfy conditions called boundary conditions. Theorems relating to the existence and uniqueness of such solutions are called existence and, uniqueness theorems., LINEAR PARTIAL DIFFERENTIAL EQUATIONS, , The general linear partial differential, bles has the form, , equation of order two in two independent varia(*), , where A, B, ..., G may depend on x and y but not on u. A second order equation with, independent variables x and y which does not have the form (1) is called nonlinear., If G = 0 the equation is called homogeneous, while if G ¥• 0 it is called non-homogeneous. Generalizations to higher order equations are easily made., Because of the nature of the solutions of (1) the equation is often classified as elliptic,, hyperbolic or parabolic according as B2 - 4AC is less than, greater than or equal to zero, respectively., 258
Page 269 :
CHAP. 12], , PARTIAL DIFFERENTIAL EQUATIONS, , 259, , SOME IMPORTANT PARTIAL DIFFERENTIAL EQUATIONS, 1. Heat Conduction Equation, Here u(x, y, z, t) is the temperature in a solid at position (x, y, z) at time t. The constant K, called the diffusivity, is equal to K/<TT where the thermal conductivity K, the, specific heat <r and the density (mass per unit volume) r are assumed constant., In case u does not depend on y and z, the equation reduces t o c a l l e d the, one-dimensional heat conduction equation., 2. Vibrating String Equation, This equation is applicable to the small, transverse vibrations of a taut, flexible string,, such as a violin string, initially located on the, x axis and set into motion [see Fig. 12-1]. The, function y(x, t) is the displacement of any point, x of the string at time t. The constant a2 = T/ju,, where T is the (constant) tension in the string, and ti is the (constant) mass per unit length of, the string. It is assumed that no external forces, act on the string but that it vibrates only due, Fig. 12-1, to its elasticity., The equation can easily be generalized to higher dimensions as for example the vibrations of a membrane or drum head in two dimensions. In two dimensions, for example, the equation is, , 3. Laplace's Equation, This equation occurs in many fields. In the theory of heat conduction, for example,, v is the steady-state temperature, i.e. the temperature after a long time has elapsed,, and is equivalent to putting du/dt = 0 in the heat conduction equation above. In the, theory of gravitation or electricity v represents the gravitational or electric potential, respectively. For this reason the equation is often called the potential equation., 4. Longitudinal Vibrations of a Beam, This equation describes the motion of a beam [Fig. 12-2] which can vibrate longitudinally [i.e. in the x direction]. The variable u(x, t) is the longitudinal displacement, from the equilibrium position of the cross section at x. The constant c2 = gE/r where g is, the acceleration due to gravity, E is the modulus of elasticity [stress divided by strain] and, depends on the properties of the beam, r is the, density [mass per unit volume]., Note that this equation is the same as that, for a vibrating string., Fig. 12-2
Page 270 :
260, , PARTIAL DIFFERENTIAL EQUATIONS, , [CHAP. 12, , 5. Transverse Vibrations of a Beam, , This equation describes the motion of a beam [initially located on the x axis, see, Fig. 12-3] which is vibrating transversely (i.e. perpendicular to the x direction). In this, case y(x,t) is the transverse displacement or, deflection at any time t of any point x. The, constant &2 = Elg/p where E is the modulus, of elasticity, / is the moment of inertia of any, cross section about the x axis, g is the acceleration due to gravity and ju is the mass per unit, length. In case an external transverse force, F(x, t) is applied, the right hand side of the, equation is replaced by b2F(x,t)/EI., Fig. 12-3, METHODS OF SOLVING BOUNDARY-VALUE PROBLEMS, , There are many methods by which boundary-value problems involving linear partial differential equations can be solved. The following are among the most important., 1. General Solutions., , In this method we first find the general solution and then that particular solution, which satisfies the boundary conditions. The following theorems are of fundamental, importance., Theorem 12-1 [Superposition principle]. If u\, u2, ...,«„ are solutions of a linear homogeneous partial differential equation, then ciUi + c2«2 + • • • + cnun where, ci, c2, ..., cn are constants is also a solution., Theorem 12-2. The general solution of a linear non-homogeneous partial differential equation is obtained by adding a particular solution of the non-homogeneous, equation to the general solution of the homogeneous equation., We can sometimes find general solutions by using the methods of ordinary differential equations. See Problems 12.8 and 12.9., If A, B, ..., F in (1) are constants, then the general solution of the homogeneous, equation can be found by assuming that u = eax+t>" where a and & are constants to be, determined. See Problems 12.10-12.13., 2. Separation of Variables., , In this method it is assumed that a solution can be expressed as a product of, unknown functions each of which depends on only one of the independent variables., The success of the method hinges on being able to write the resulting equation so that, one side depends only on one variable while the other side depends on the remaining, variables so that each side must be a constant. By repetition of this the unknown, functions can then be determined. Superposition of these solutions can then be used to, find the actual solution. See Problems 12.15-12.17., The method often makes use of Fourier series, Fourier integrals, Bessel series and, Legendre series. For illustrations see Problems 12.18-12.28.
Page 271 :
CHAP. 12], , PARTIAL DIFFERENTIAL EQUATIONS, , 261, , 3. Laplace Transform Methods., In this method the Laplace transform of the partial differential equation and, associated boundary conditions are first obtained with respect to one of the independent, variables. We then solve the resulting equation for the Laplace transform of the, required solution which is then found by taking the inverse Laplace transform. In, cases where Laplace inversion is of some difficulty the complex inversion formula of, Chapter 14 can be used. For illustrations see Problems 12.29 and 14.15-14.19., 4. Complex Variable Methods. [See Chapter 13.], , Solved Problems, CLASSIFICATION OF PARTIAL DIFFERENTIAL EQUATIONS, 12.1. Determine whether each of the following partial differential equations are linear or, nonlinear, state the order of each equation, and name the dependent and independent, variables., (A), , ( a ) l i n e a r , order 2, dep. var. u, ind. var. x,t, , (B), , (b)linear, order 3, dep. var. R, ind. var. x, y, , (C), , ( c ) n o n l i n e a r , order 2, dep. var. W, ind. var. r, s, t, , (D), , ( d ) l i n e a r , order 2, dep. var. <f>, ind. var. x,y,z, , (E), , (e) nonlinear, order 1, dep. var. z, ind. var. u, v, , 12.2. Classify each of the following equations as elliptic, hyperbolic or parabolic., (a), , (b), , (c), , (d), , (e), , Then B2 - 4AC = -4xy., In the region xy > 0 the equation is elliptic., In the region xy < 0 the equation is hyperbolic., If xy = 0, the equation is parabolic.
Page 272 :
262, , PARTIAL DIFFERENTIAL EQUATIONS, , [CHAP. 12, , SOLUTIONS OF PARTIAL DIFFERENTIAL EQUATIONS, , 12.3. Show that u(x,t) = e~Kt sin2x is a solution to the boundary-value problem, u(0,t) - u(v,t) = 0,, , u(x,0) = sin2x, , From u(x, t) = e~8t sin 2x we have, u(Q, t) = e~at sin 0 = 0,, , u(v, t) = e~8« sin 2ir = 0,, , u(x, 0) = «-° sin 2x = sin 2*, , and the boundary conditions are satisfied., Also, Then substituting into the differential equation, we have, which is an identity., , -8e-** sin 2x = 2(-4e-8t sin 2x), , 12.4. (a) Show that v = F(y — Sx), where F is an arbitrary differentiable function, is a, general solution of the equation, (&) Find the particular solution which satisfies the condition v(Q, y) = 4 sin y., (a) Let y — 3x = u. Then v — F(u) and, , Thus, Since the equation is of order one, the solution v = F(u) = F(y — 3#) which involves only, one arbitrary function is a general solution., (6) v(x, y) - F(y - Zx)., , Then v(0, y) = F(y) = 4 sin y., , If F(y) = 4 sin y, then v(x, y) — F(y — 3x) = 4 sin (y — Zx) is the required solution., , 12.5. (a) Show that y(x, t) = F(2x + 5t) + G(2x - 5t) is a general solution of, (6) Find a particular solution satisfying the conditions, y(Q,t) = y(*,t) = 0,, , y(x,Q) = sin2x,, , yt(x,0) = 0, , (a) Let 2x + 5t = u, 2x-5t = v. Then y = F(u) + G(v)., (1), (2), (3), (4)
Page 273 :
CHAP. 12], , PARTIAL DIFFERENTIAL EQUATIONS, , 263, , From (2) and (4),, and the equation is satisfied. Since the equation is of, order 2 and the solution involves two arbitrary functions, it is a general solution., (6) We have from y(x, t) - F(2x + 5t) + G(2x - 5t),, (5), Also, s, , ot, , h, , a, , t, , (, , «, , ), , Differentiating (5),, From (6),, Then, from which, , i.e., Using, , y(0, t) = 0 or Y(n, t) = 0, ct + c2 = 0, y(x> f), , so that, , = % sin (2x + 5t) + $ sin (2x - 5t) = sin 2x cos 5t, , which can be checked as the required solution., , SOME IMPORTANT PARTIAL DIFFERENTIAL EQUATIONS, 12.6. If the temperature at any point (x, y, z) of a solid at time t is u(x, y, z, t) and if K, a and, T are respectively the thermal conductivity, specific heat and density of the solid,, assumed constant, show that, where K =, , Klar, , Let V be an arbitrary volume lying within the solid, and let S denote its surface., flux of heat across S, or the quantity of heat leaving S per unit time, is, , The total, , Thus the quantity of heat entering S per unit time is, (J), by the divergence theorem. The heat contained in a volume V is given by, , Then the time rate of increase of heat is, (2), Equating the right hand sides of (1) and (2),, , and since V is arbitrary, the integrand, assumed continuous, must be identically zero so that, , or if K, <r, T are constants,, The quantity K is called the diffusivity., For steady-state heat flow (i.e. du/dt = 0 or it is, independent of time) the equation reduces to Laplace's equation V2w = 0.
Page 274 :
PARTIAL DIFFERENTIAL EQUATIONS, , 264, , [CHAP. 12, , 12.7. Derive the vibrating string equation on page 259., Referring to Fig. 12-4 assume that As represents, an element of arc of the string. Since the tension is, assumed constant the net upward vertical force acting, on As is given by, W, , Since sin e — tan e approximately for small angles,, this force is, (2), , Fig. 12-4, , using the fact that the slope is tan e = 8y/dx., , By Newton's law this net force is equal to the, 52w, mass of the string G»As) times the acceleration of As which is given by -gp + « where e-»0 as, As -» 0. Thus we have approximately, (3), , If the vibrations are small then As = Aw approximately so that (3) becomes on division by T A*,, , Taking the limit as Ase -» 0 [in which case e -» 0 also], we have, , METHODS OF FINDING SOLUTIONS OF PARTIAL DIFFERENTIAL EQUATIONS, , 12.8. (a) Solve the equation, (6) Find the particular solution for which z(x, 0) = x2, 2(1, y) = cos y., (a) Write the equation as, , Then integrating with respect to «, we find, , where F(y) is arbitrary., Integrating (1) with respect to y,, where G(x) is arbitrary., , W, , (2), , The result (2) can be written, (S), , which has two arbitrary (essential) functions and is therefore a general solution., (6) Since z(x, 0) = a;2, we have from (S), (*), , Thus, , (5), , Since z(L, y) = cos y, we have from (5), or, , Thus using (6) in (5) we find the required solution, , (<?)
Page 275 :
CHAP. 12], 12.9., , PARTIAL DIFFERENTIAL EQUATIONS, , 265, , Solve, Write the equation as, , Integrating with respect to x,, , or, , This is a linear equation having integrating factor, , e^, , (2/od, , * = e 2 1 n t = e tat2 = t2. Then, , Integrating,, the required general solution., , 12.10. Find solutions of, Assume u = ea*+1)». Substituting in the given equation, we find, (a2 + 3a6 + 262)e°*'+t>s = 0 or a2 + Sab + 2&2 = 0, Then (a+ 6)(a + 26) = 0 and a = -6, a = -26. If a = -b, e-bx+by = e«»-*> is a solution, for any value of b. If a — —26, e-2bx+b» — e^u-2*) is a solution for any value of 6., Since the equation is linear and homogeneous, sums of these solutions are solutions. For, example, Ze^v-x) _ z^v-xt + §ev<.y-x) js a solution (among many others) and one is thus led to, F(y — x) where F is arbitrary, which can be verified as a solution. Similarly G(y — 2x) where G, is arbitrary is a solution. The general solution found by addition is then given by, , 12.11. Find a general solution of (a), , (b), , solution., Thus u = exF(2y — 3x) is a general solution., , ig, , a, , (6) Let M = eax+b!/. Then 4a 2 -4a6 + 62 = 0 and 6 = 2a, 2a. From this U = e°<*+2»> and so, F(x + 2y) is a solution., By analogy with repeated roots for ordinary differential equations we might be led to, believe xG(x + 2y) or yG(x + 2y) to be another solution, and that this is in fact true is easy, to verify. Thus the general solution is, M = F(x + 2y) + xG(x + 2y), , or, , u = F(x + 2y) + yG(x + 2y), , 12.12. Solve, The homogeneous e q u a t i o n ) has general solution u = F(x + iy) + G(x — iy) by, Problem 12.42(c)., To find a particular solution of the given equation assume u = aezx+u where a is an unknown, constant. This is the method of undetermined coefficients as in ordinary differential equations., We find a = 2 so that the required general solution is, u = F(x + iy) + G(x-iy) + 2e**+v
Page 276 :
PARTIAL DIFFERENTIAL EQUATIONS, , 266, , [CHAP. 12, , 12.13. Solve, The homogeneous equation has general solution, M = F(2x + y) + G(2x-y), To find a particular solution, we would normally assume M = ae*x+» as in Problem 12.12 but, this assumed solution is already included in F(2x + y). Hence we assume as in ordinary differential, equations that u = axe2x+« (or u = aye2***). Substituting, we find a = J., Then the general solution is, M = F(2x + y) + G(2x -y) + ±xe**+v, , SEPARATION OF VARIABLES, 12.14. Solve the boundary-value problem, by the method of separation of variables., Let u = XY in the given equation. Then, Since X depends only on x and Y depends only on y and since x and y are independent variables,, each side must be a constant, say c., Then X' - 4eX = 0, Y' - cY = 0 whose solutions are X = Ae*<*, Y = Be<v., A solution is thus given by, u(x,y) = XY = ABe««*+»> = Ke^*x+^, From the boundary condition,, , u(0,y) = Ke<» = 8e~3», which is possible if and only if K = 8 and c = —3., Then u(x, y) = 8e~3^4x+V> = %e-i2x-sy is the required solution., , 12.15. Solve Problem 12.14 if u(Q,y) = 8e~3y + 4e~s»., As before a solution is Z6c<te+»>., Then J5T1e<;i<4l+»> and K2ec^4x+^ are solutions and by the principle of superposition so also is, their sum, i.e. a solution is, u(x,y) = JBT1eci<4*+»> + K2eci^x+^, From the boundary condition,, u(0,y), , = K^'in + K2eci", , = 8e~^ + 4e~5», , which is possible if and only if Kl = 8, K2 = 4, cs = —3, c2 = —5., Then u(x,y) = 8e-3<4*+»> + 4e-5<4*+!'> = Se~12x-to + 4e-*<>x-Zy is the required solution., , 12.16. Solve H = 2^, 0 < x < 3, t > 0, given that w(0, t) = u(S, t) = 0,, u(x,Q) = 5sin47ra; - SsinSTra; + 2 sin 1 OTTO;, |«(«,t)| < M, where the last condition states that u is bounded for 0 < x < 3, t > 0., Let u = XT. Then, , XT' = 2X"T, , and, 2, , X"/X = T'/2T, , Each side must be a constant which we call —X . [If we use +X2, the resulting solution obtained, does not satisfy the boundedness condition for real values of X.] Then
Page 277 :
CHAP. 12], , PARTIAL DIFFERENTIAL EQUATIONS, , 267, , with solutions, A solution of the partial differential equation is thus given by, u(x, t) = XT, , = c^e-^* (At cos X» + Bl sin Xz), , =, , e~2>ft (A cos \x + B sin X*), , Since u(0, t) = 0, e~2)A (A) = 0 or A = 0. Then, u(x,t) = Be-2*?* sinXse, Since u(S, t) = 0, Be~2>?t sin 3X = 0. If B = 0, the solution is identically zero, so we must, have sin 3X = 0 or 3X = mv, X = mv/3 where m = 0, ±1, ±2, .... Thus, is a solution., Also, by the principle of superposition,, (1), is a solution., By the last boundary, , condition., , This is possible if and only if Bx = 5, m1 = 12, B2 = -S, m2 = 24, Ba = 2,ms = 30., Substituting these in (1), the required solution is, u(x,t), , = 5e-32^« sin4^o; - 3e-i28^« sinSffa; + 2e-2«o>A sinlOa-a;, , (2), , This boundary-value problem has the following heat flow, interpretation. A bar whose surface, is insulated [Fig. 12-5] has a length of 3 units and a diffusivity of 2 units. If its ends are kept at, temperature zero units and its initial temperature u(x, 0) = 5 sin fax — 3 sin 8vx + 2 sin Wvx,, find the temperature at position x at time t, i.e. find u(x, t). We shall assume that c.g.s. units are, used and that temperature is in degrees centigrade (°C). However, other units could of course be, used., , Fig. 12-5, , SOLUTIONS USING FOURIER SERIES, , 12.17. Find the temperature of the bar in Problem 12.16 if the initial temperature is 25°C., This problem is identical with Problem 12.16 except that to satisfy the initial condition, u(x, 0) = 25 it is necessary to superimpose an infinite number of solutions, i.e. we must replace, equation (1) of the last problem by, , which for t = 0 yields, This amounts to the same thing as expanding 25 in a Fourier sine series. By the methods of, Chapter 7 we then find
Page 278 :
268, , PARTIAL DIFFERENTIAL EQUATIONS, , [CHAP. 12, , The result can be written, , which can be verified as the required solution., This problem illustrates the importance of Fourier series (and orthogonal series in general) in, solving boundary-value problems., , 12.18. Solve the boundary-value problem, This is the same as Problem 12.17 except that the ends of the bar are at temperatures 10°C, and 40°C instead of 0°C. As far as the solution goes this makes quite a difference since we can, no longer conclude that A = 0 and X = mv/3 as in that problem., To solve the present problem assume that u(x, t) — v(x, t) + $(x) where vK#) is to be suitably, determined. In terms of v(x, t) the boundary-value problem becomes, , This can be simplified by choosing, from which we find t(x) — lOa; +10 so that the resulting boundary-value problem is, , As in Problem 12.17 we now find from the first three of these,, , The last condition yields, from which, Since u(x, t) = v(x, t) + ^(x), we have finally, as the required solution., The term lOa; +10 is the steady-state temperature, i.e. the temperature after a long time has, elapsed., , 12.19. A string of length L is stretched between, points (0,0) and (L,0) on the x axis. At, time t = 0 it has a shape given by f(x),, 0 < x < L and it is released from rest., Find the displacement of the string at any, later time., The equation of the vibrating string is, where y(x, t) = displacement from x axis at time t, [Fig. 12-6]., , Fig.12_6
Page 279 :
CHAP. 12], , PARTIAL DIFFERENTIAL EQUATIONS, , 269, , Since the ends of the string are fixed at x = 0 and x = L,, y(0,t) = y(L,t), , = 0, , t >0, , Since the initial shape of the string is given by f(x),, y(x,Q) = f(x), , Q<x<L, , Since the initial velocity of the string is zero,, yt(x, 0) = 0, , 0<x<L, , To solve this boundary-value problem, let y — XT as usual., Then, , XT" = a?X"T, , or, , T"/a?T -, , X"IX, , Calling the separation constant —X2, we have, , T" + X%2r = 0,, and, , T = A! sin Xot + B^ cos \at,, , X" + \2X = 0, , X, , = A2 sin Xa; + Bz cos X*, , A solution is thus given by, y(x, t) = XT — (A2 sin \x + B2 cos \x)(A1 sin Xat + Bl cos Xat), From j/(0, t) = 0, A2 = 0, Then, y(x, t) — B2 sin Xa:(Aj sin Xat + Bl cos Xat) = sin \x(A sin Xat + B cos Xat), From y(L, t) = 0, we have, , sin \L(A sin Xat + B cos Xat) = 0, , so that sin XL = 0, XL = ma- or X = mv/L since the second factor must not be equal to zero. Now, yt(x, t) = sin \x(A\a cos Xat — B\a sin Xat), and yt(x,0) = (sin\x)(A\a) = 0 from which A = 0. Thus, , To satisfy the condition y(x, 0) = f(x), it will be necessary to superimpose solutions., yields, , This, , Then, and from the theory of Fourier series,, , The result is, which can be verified as the solution., The terms in this series represent the natural or normal modes of vibration. The frequency of, the mth normal mode fm is obtained from the term involving cosmiLa and is given by, Li, , Since all the frequencies are integer multiples of the lowest frequency flt the vibrations of the, string will yield a musical tone as in the case of a violin or piano string.
Page 280 :
270, , PARTIAL DIFFERENTIAL EQUATIONS, , [CHAP. 12, , 12.20. A circular plate of unit radius, whose faces are, insulated, has half of its boundary kept at constant, temperature Ui and the other half at constant temperature «2 [see Fig. 12-7]. Find the steady-state, temperature of the plate., In polar coordinates (p, <f>) the partial differential equation for steady-state heat flow is, , W, , Fig. 12-7, , The boundary conditions are, (2), , i.e. M is bounded in the region, Let u(P,<t>) = P# where P is a function of p and * is a function of </>., becomes, , Dividing by P*, multiplying by p2 and rearranging terms,, , Setting each side equal to X2,, which have solutions, [See Problem 3.70(a), page 94.], Since u(p, #) must have period 2ir in <f>, we must have X = m = 0,1,2,3, Also, since « must be bounded at p = 0, we must have B2 = 0. Thus, , By superposition, a solution is, , from which, Then from the theory of Fourier series,, , (3), Then equation (1)
Page 281 :
CHAP. 12], , PARTIAL DIFFERENTIAL EQUATIONS, , 271, , Then, , by Problem 12.52., , SOLUTIONS USING FOURIER INTEGRALS, , 12.21. A semi-infinite thin bar x ^ 0 whose surface is insulated has an initial temperature, equal to f(x). A temperature of zero is suddenly applied to the end x = 0 and, maintained, (a) Set up the boundary-value problem for the temperature u(x,t) at, any point x at time t. (b) Show, that, , (a) The boundary-value problem is, (1), «(*,0) = /(*),, , u(0,t) = 0,, , \u(x,t)\<M, , (2), , where the last condition is used since the temperature must be bounded for physical reasons., (6) A solution of (1) obtained by separation of variables is, u(x, t) =, , e-Ktft, , (A cos \x + B sin X*), , From the second of boundary conditions (2) we find .4 = 0 so that, u(x,t), , = Be-^sinX*, , (3), , Now since there is no restriction on X we can replace B in ($) by a function B(\) and still have, a solution. Furthermore we can integrate over X from 0 to °° and still have a solution. This, is the analog of the superposition theorem for discrete values of X used in connection with, Fourier series. We thus arrive at the possible solution, (4), , From the first of boundary conditions (2) we find, , which is an integral equation for the determination of B(\)., f(x) must be an odd function we have, , From page 201 we see that since, , Using this in (4) we find, , 12.22. Show that the result of Problem 12.21 can be written, , Using, written, , sin \v sin X* = |[cos X(i> - x) — cos X(i> + *)], , the result of Problem 12.21 can be
Page 282 :
272, , PARTIAL DIFFERENTIAL EQUATIONS, , [CHAP. 12, , From the integral, [see Problem 9.9, page 215] we find, , Letting (v — x)/2^ici = w in the first integral and, we find that, , (v + a;)/2\/Kt = w, , in the second integral,, , 12.23. In case the initial temperature f(x) in Problem 12.21 is the constant Uo, show that, , where erf (x/2}/Kt ) is the error function [see page, If f(x) = M0, we obtain from Problem 12.22, , 212]., , SOLUTIONS USING BESSEL FUNCTIONS, 12.24. A circular plate of unit radius [as in Fig. 12-7, page 270] has its plane faces insulated., If the initial temperature is F(p) and if the rim is kept at temperature zero, find the, temperature of the plate at any time., Since the temperature is independent of <f>, the boundary value problem for determining, u(p, t) is, (1), u(l,t) = 0,, , u(P,Q) = F(P),, , \u(P,t)\ < M, , Let u = P(p)T(t) = PT in equation (1). Then, , or dividing by KPT,, from which, These have general solutions [see Chapter 10], T, , =, , Cle-^t>, , P, , =, , A,J0(\P) + B1F0(XP), , Since M = PT is bounded at p = 0, B^ = 0. Then, where A = AjCj.
Page 283 :
CHAP. 12], , PARTIAL DIFFERENTIAL EQUATIONS, , 273, , From the first boundary condition,, u(l,t) = Ae-^t J0(X) = 0, from which J0(X) = 0 and X = X1,X2, ... are the positive roots., Thus a solution is, , m = 1,2,3, ..., , By superposition, a solution is, From the second boundary condition,, Then by the methods of Chapter 10,, , and so, , (2), , which can be established as the required solution., Note that this solution also gives the temperature of an infinitely long solid cylinder whose, convex surface is kept at temperature zero and whose initial temperature is F(p)., , 12.25. A drum consists of a stretched circular membrane of unit radius whose rim represented by the circle of Fig. 12-7, page 270, is fixed. If the membrane is struck so, that its initial displacement is F(p,<t>) and is then released, find the displacement at, any time., The boundary-value problem for the displacement z(p, <f>, t) from the equilibrium or rest position, (the xy plane) is, 2(1, <f,,t) = 0,, , z(p,0,0) = 0,, , zt(p,<t>,0) = 0,, , z(p,>,0) = F(p,<t>), , Let z = P(P) *(0) T(t) = P4>r. Then, , Dividing by a2P*r,, a, , n, , ds, , o, , (, , i, (*), , Multiplying equation (2) by p2, the variables can be separated to yield, , s, , ot, , h, , a, , t, , (, , #, , ), (4), , General solutions of (1), (5) and (4) are, (5), (6), , (7)
Page 284 :
274, , PARTIAL DIFFERENTIAL EQUATIONS, , [CHAP. 12, , A solution z(p, <t>, t) is given by the product of these., Since z must have period 2ir in the variable <j>, we must have /i = m where m = 0,1,2,3, ..., from equation (6)., Also, since z is bounded at p = 0 we must take B3 = 0., Furthermore, to satisfy zt(p, <j>, 0) = 0 we must choose BI — 0., Then a solution is, u(p, <f>, t), , =, , Jm(\p) cos \at (A cos m</> + B sin m<j>), , Since z(l,<j>, t) - 0, J m (X) = 0 so that X = X^, fc = 1,2,3, ..., are the positive roots., By superposition (summing over both m and k),, , (8), , Putting t = 0, we have, (9), , where, , a<?), , But (9) is simply a Fourier series and we can determine Cm and Dm by the usual methods., We find, , From (10), using the results of Bessel series expansions, we have, , Using these values of Amk and Bmk in (5) yields the required solution., Note that the various modes of vibration of the drum are obtained by specifying particular, values of m and k. The frequency of vibration is then given by, , Because these are not integer multiples of the lowest frequency, we would expect noise rather, than a musical tone.
Page 285 :
CHAP. 12], , PARTIAL DIFFERENTIAL EQUATIONS, , 275, , SOLUTIONS USING LEGENDRE FUNCTIONS, 12.26. Find solutions to Laplace's equation in spherical coordinates which are independent, of 0., Laplace's equation V2i> = 0 in spherical coordinates if there is no 0 dependence can be written, (1), Let v = R(r) 6(0) = R® in (1). Then after dividing by RQ, the equation becomes separable, as follows, , find(#), , and we, , W, , Equation (1) can be written, , (4), , a Cauchy equation [see Problem 3.70(e), page 94] having solution, (5), , where we have placed n — —^ + VJ—~X* so that X2 = — n(n + 1)., Equation (S) with X2 = — n(n + 1) can be written, •j-(sme^) + w(n + l)sin«e = 0, ctff \, do j, , (6), , Now if we let cos 9 = x, then, , so that, , Then (6) becomes, or, , (7>), , which is Legendre's differential equation having solution, (*), , [see Chapter 11]. Thus the solution of (6) is, (9), Using (5) and (8) or (9), we then see that a solution of (1) is, where x = cos e., , (10)
Page 286 :
276, , PARTIAL DIFFERENTIAL EQUATIONS, , [CHAP. 12, , 12.27. Find the potential v (a) interior to and (b) exterior to a hollow sphere of unit, radius if half of its surface is charged to potential v0 and the other half to potential, zero., Choose the sphere in the position shown in Fig., 12-8. Then v is independent of <f> and we can use the, results of Problem 12.26. A solution is, , where x = cos 6. Since v must be bounded at e = 0, and IT, i.e. x = ±1, we must choose B2 = 0. Then, , The boundary conditions are, , and v is bounded., , Fig. 12-8, , (a) Interior Potential, 0 S r < 1., Since v is bounded at r = 0, choose B = 0. Then a solution is, , By superposition,, When r = 1,, Then as in Problem 8.16,, , from which, A0 = $v0,, Thus, , V(r,e), , AI = |-w0,, , Az = 0,, , A3 = -T\v0., , 4, , 4 = 0,, , A5 = I|v0, , = Y.[! + frP^cosff) - |r3P3(cos») + flP5(cos«) + • • • ], , (6) Exterior Potential, 1 < r < »., Since v is bounded as r-*<*>, choose A = 0. Then a solution is, , By superposition,, When r = 1,, Then Bn = An of part (a) and so
Page 287 :
CHAP. 12], , PARTIAL DIFFERENTIAL EQUATIONS, , 277, , SOLUTIONS USING LAPLACE TRANSFORMS, 12.28. Solve by Laplace transforms the boundary-value problem, , u(Q, t) = 0,, , u(B, t) = 0,, , u(x, 0) = 10 sin 2irX — 6 sin 4-n-x, , Taking the Laplace transform of the given differential equation with respect to t, we have, , which can be written as, , or, , (1), , where, Using the given condition u(x, 0) = 10 sin 2irx — 6 sin 4vx, (1) becomes, (*), , Taking the Laplace transform of the conditions U(0, t) = 0, 17(3, t) = 0, we have, , or, , (S), , Solving the ordinary differential equation (2) subject to conditions (8) by the usual elementary, methods, we find, Thus taking the inverse Laplace transform we find, , which is the required solution., Note that theoretically we could have taken the Laplace transform of the given differential, equation with respect to * rather than t. However, this would lead to various difficulties as is, evident upon carrying out the procedure. In practice we take Laplace transforms with respect, to the various independent variables and then choose that variable which leads to the greatest, simplification., , Supplementary Problems, CLASSIFICATION OF PARTIAL DIFFERENTIAL EQUATIONS, , 12.29. Determine whether each of the following partial differential equations is linear or nonlinear, state, the order of each equation, and name the dependent and independent variables., (, (6), , a, , ), , (, , c, , ), (d), , (, , e, , )
Page 288 :
278, , PARTIAL DIFFERENTIAL EQUATIONS, , [CHAP. 12, , 12.30. Classify each of the following equations as elliptic, hyperbolic or parabolic., (, , a, , ), , (, , d, , ), , (, , b, , ), , (, , e, , ), , (c), , SOLUTIONS OF PARTIAL DIFFERENTIAL EQUATIONS, 12.31. Show that z(x,y) = 4e~3x cosSy is a solution to the boundary-value problem, , 12.32. (a) Show that v(x, y) — xF(2x + y) is a general solution of, (6) Find a particular solution satisfying v(i, y) = j/2., 12.33. Find a partial differential equation having general solution u = F(x — 3j/) + G(2x + y)., 12.34. Find a partial differential equation having general solution, , (a) z = e*f(2y-3x),, , (b) z = f(2x + y) + g(x-2y), , SOME IMPORTANT PARTIAL DIFFERENTIAL EQUATIONS, 12.35. If a taut, horizontal string with fixed ends vibrates in a vertical plane under the influence of, gravity, show that its equation is, .2, ,2, , a! = "'a**-*, where g is the acceleration due to gravity., 12.36. A thin bar located on the x axis has its ends at x = 0 and x = L. The initial temperature of the, bar is f(x), 0 < x < L, and the ends x = 0, x = L are maintained at constant temperatures TI, Tz, respectively. Assuming the surrounding medium is at temperature U0 and that Newton's law of, cooling applies, show that the partial differential equation for the temperature of the bar at any, point at any time is given by, and write the corresponding boundary conditions., 12.37. Write the boundary conditions in Problem 12.36 if the ends x = 0 and x = L are insulated., 12.38. The gravitational potential v at any point (x, y, z) outside of a mass ra located at the point (X, Y, Z), is defined as the mass m divided by the distance of the point (x, y, z) from (X, Y, Z). Show that, v satisfies Laplace's equation V2v = 0., 12.39. Work Problem 12.38 for a solid., 12.40. Assuming that the tension of a vibrating string is variable, show that its equation is, , METHODS OF FINDING SOLUTIONS OF PARTIAL DIFFERENTIAL EQUATIONS, 12.41. (a) Solve * ~- + ^- = 0. (6) Find the particular solution for which, dx dy, dy, z(x,0) - x* + x,, 2(2, y) = 3y*
Page 289 :
CHAP. 12], , PARTIAL DIFFERENTIAL EQUATIONS, , 279, , 12.42. Find general solutions of each of the following., ( A ), (B), (o), , 12.43. Find general solutions of each of the following., (A), , (E), , (6), , (d), , 12.44. Solve, 12.45. Show that the general solution of, , SEPARATION OF VARIABLES, 12.46. Solve each of the following boundary-value problems by the method of separation of variables., , (, , a, , (, (, (d), , ), , 0, , b, c, , ), ), , z, , u(x, 0 ) = 40-*, , ,, , u(x, 0 ) = Ze~5x + 2e-3*, , ,, t, , (, , 0, , , t) = 0,, , ux(0,t) = 0,, , u(v, t) = 0,, u(2,t) = 0,, , u(x, 0) = 2 sin Zx — 4 sin 5x, w(*,0) = 8cos^-6cos^, , 12.47. Solve the boundary-value problem, y(0,t) = y(5,t) - 0,, if, , (a) f(x) — 5 sin wx,, , y(x,0) = 0,, , yt(x,Q) = /(»), , (b) f(x) = 3 sin Zvx — 2 sin 5irx., , SOLUTIONS USING FOURIER SERIES, 12.48. (a) Solve the boundary-value problem, «(0,t) = M(4,<) = 0,, , M(*,0) - 25«, , where 0 < x < 4, t > 0., (6) Interpret physically the boundary-value problem in (a)., 12.49. (a) Show that the solution of the boundary-value problem, ux(0, t) = ux(ir,t) - 0,, where 0 < x < ir, t>0 is given by, , (b) Interpret physically the boundary-value problem in (a)., , u(x,Q) - f(x)
Page 290 :
280, , PARTIAL DIFFERENTIAL EQUATIONS, , [CHAP. 12, , 12.50. Find the steady-state temperature in a bar whose ends are located at x = 0 and x = 10 if these, ends are kept at 150°C and 100°C respectively., 12.51. A circular plate of unit radius [see Fig. 12-7, page 270] whose faces are insulated has its boundary, kept at temperature 120 + 60 cos 20. Find the steady-state temperature of the plate., 12.52. Show that, , \, , and thus complete Problem 12.20., , 12.53. A string 2 ft long is stretched between two fixed points * = 0 and x = 2. If the displacement, of the string from the x axis at t = 0 is given by f(x) = .03#(2 - *) and if the initial velocity, is zero, find the displacement at any later time., 12.54. A square plate of side a has one side maintained at temperature /(») and the others at zero, as, indicated in Fig. 12-9. Show that the steady-state temperature at any point of the plate is, given by, Obtain the result for f(x) = u0, a constant., 12.55. Work Problem 12.54 if the sides are maintained at temperatures fi(x),gi(y),fz(x),gz(y), [Hint. Use the principle of superposition and the result of Problem 12.54.], , Fig. 12-9, , respectively., , Fig. 12-10, , 12.56. An infinitely long plate of width a indicated by the shaded region of Fig. 12-10 has its two, parallel sides maintained at temperature 0 and its other side at constant temperature u0. (a) Show, that the steady-state temperature is given by, , (6) Use Problem 12.52 to show that, , 12.57. Solve Problem 12.35 if the string has its ends fixed at x — 0 and x = I and if its initial displacement and velocity are given by f(x) and g(x) respectively., 12.58. A square, its edges, vibration., (a) Show, point, , plate [Fig. 12-10] having sides of unit length has, fixed in the xy plane and is set into transverse, that the transverse displacement z(x, y, t) of any, (x, y) at time t is given by, I, 2, , where a is a constant., (6) Show that if the plate is given an initial shape f ( x , y ), arid released with velocity g(x,y), then the displacement, is given by, , Fig. 12-11
Page 291 :
CHAP. 12], , PARTIAL DIFFERENTIAL EQUATIONS, , 281, , where, , and \mn =, 12.59. Show that the natural frequencies for the plate of Problem 12.57 are given by fmn = a^m2 + n2/2ir., Would you expect to get a musical tone from the vibration? Explain., 12.60. A beam has its ends hinged at x = 0 and x ' = I. At time * = 0, a concentrated transverse load, of constant magnitude w is suddenly applied at the midpoint. Show that the resulting transverse, displacement of any point x of the beam at any time t > 0 is, , if 0 < x < 1/2, while the corresponding result for 1/2 < x < I is obtained by symmetry., 12.61. Solve the boundary-value problem, , u(0,t) = ult, , u(l,t) = u2,, , u(x,0) = 0, , where a and I are constants, and interpret physically., 12.62. Show that the solution of the boundary-value problem, , ux(0,t) = hu(0,t),, , ux(l,t) = -hu(l,t),, , u(x,0) = /(*), , where k, h and I are constants is, , where Xn are solutions of the equation, , Give a physical interpretation., , SOLUTIONS USING FOURIER INTEGRALS, 12.63. An infinite thin bar (— °° < x < «>) whose surface is insulated has an initial temperature equal to, f(x). Show that the temperature at any point x at any time t is given by, , 12.64. If f(x)-u0 in Problem 12.63, show that u(x,t) = Uoert&fti/ki), with Problem 12.23., 12.65. Solve the boundary-value problem, , showing that, , u(x, y), , and give a physical interpretation., , and explain the connection
Page 292 :
282, , PARTIAL DIFFERENTIAL EQUATIONS, , 12.66. Show that if, , in Problem 12.65,, , [CHAP. 12, , then, , 12.67. The region bounded by x > 0, y > 0 has one edge * = 0 kept at potential zero and the other, edge y = 0 kept at potential /(*). (a) Show that the potential at any point (*, y) is given by, , (6) If /(«) = !, show that v(x,y) = -tan-i-., H, , Jf, , SOLUTIONS USING BESSEL FUNCTIONS, 12.68. The temperature of a long solid circular cylinder of unit radius is initially zero. At t = 0 the, surface is given a constant temperature u0 which is then maintained. Show that the temperature, of the cylinder is given by, , where \n, n = 1,2,3, ... , are the positive roots of J0(X) = 0 and fe is the diffusivity., 12.69. Show that if F(p) — w0(l — p2), then the temperature of the cylinder of Problem 12.23 is given by, , 12.70. A cylinder 0 < p < a, 0 < z < I has the end z = 0 at temperature /(p) while the other surfaces, are kept at temperature zero. Show that the steady-state temperature at any point is given by, , where 0J(\na) = 0, n = 1,2,3,... ., 12.71. A circular membrane of unit radius lies in the xy plane with its center at the origin. Its edge, p = 1 is fixed in the xy plane and it is set into vibration by displacing it an amount /(p) and then, releasing it. Show that the displacement is given by, , where Xn are the roots of J0(\) = 0., 12.72. (a) Solve the boundary-value problem, , where 0 < p < 1, 0 < <f> < 2v, t > 0 if u is bounded, and, «(1, <&, t) = 0,, u(p, 0,0) = p cos 30,, , ut(p, 0,0) = 0, , (6) Give a physical interpretation to the solution., 12.73. Solve the boundary-value problem, given that y(x, 0) = /(*), yt(x, 0) = 0, y(l, t) = 0 and y(x, t) is bounded for 0 &x SI, t> 0.
Page 293 :
CHAP. 12], , PARTIAL DIFFERENTIAL EQUATIONS, , 283, , 12.74. A chain of constant mass per unit length and length I is suspended, vertically from one end 0 as indicated in Fig. 12-12. If the chain, is displaced slightly at time t = 0 so that its shape is given by, /(*), 0 < x < I, and then released, show that the displacement of, any point x at time t is given by, , where \n are the roots of J0(2\\/l/g), , = 0 and, , Fig. 12-12, 12.75. Determine the frequencies of the normal modes and indicate whether you would expect music or, noise from the vibrations., 12.76. A solid circular cylinder 0 < p < a , 0 < z < I has its bases kept at temperature zero and the, convex surface at constant temperature u0. Show that the steady-state temperature at any point, of the cylinder is, where /0 is the modified Bessel function of order zero., 12.77. A membrane in the form of a circular ring a S p ^ b is set into vibration., frequencies of the various modes of vibration are given by, , Show that the, , where T is the tension per unit length, a is the mass per unit area, and Xmn are roots of the, equation, 12.78. Explain how you would determine the displacement of the membrane in Problem 12.77., 12.79. The surface p = 1 of an infinite cylinder is kept at temperature f(z). Show that the steady-state, temperature everywhere in the cylinder is given by, , SOLUTIONS USING LEGENDRE FUNCTIONS, 12.80. Find the potential i) (a) interior and (6) exterior to a hollow sphere of unit radius with center at, the origin if the surface is charged to potential v0(l + 3 cos e) where v0 is constant., 12.81. Solve Problem 12.80 if the surface potential is 1*0 sin2 e., 12.82. Find the steady-state temperature within the region bounded by two concentric spheres of radius, a and 2a if the temperatures of the outer and inner spheres are u0 and 0 respectively., 12.83. Find the gravitational potential at any point outside a solid uniform sphere of radius a of mass m., 12.84. Is there a solution to Problem 12.82, if the point is inside the sphere? Explain., 12.85. Find the potential at any point due to a thin circular ring of radius a and mass m lying in the, xy plane with center at the origin. [Hint. First find the potential at any point on the axis.], 12.86. Work Problem 12.85 for a solid circular disc., 12.87. Show that a solution of Laplace's equation V2t> = 0 in spherical coordinates is given by, , where P™ and Q™ are the associated Legendre functions.
Page 294 :
284, , PARTIAL DIFFERENTIAL EQUATIONS, , [CHAP. 12, , SOLUTIONS USING LAPLACE TRANSFORMS, 12.88. Solve each of the following boundary-value problems by using Laplace transforms., (a)u(x,0) = 4e~2*, (6), u(x,0) = 10e-* - 6e-4*, ( e ) u ( Q ,t)=0, «(4t)=0, u(x0)=6sin(vxlZ)+3sin, (d) u, 12.89. Work (a) Problem 12.46(d), (b) Problem 12.47, by use of Laplace transforms., 12.90. Show how you would attempt to solve Problem 12.17 by Laplace transforms. Explain what difficulties arise. [Problems of this type can best be solved by using the methods of Chapter 14.], , Answers to Supplementary Problems, 12.29. (a) linear, dep. var. u, ind. var. x, y, order 2, (b) linear, dep. var. T, ind. var. x, y, z, order 4, (c) nonlinear, dep. var. <j>, ind. var. x, y, order 3, , (d) linear, dep. var. y, ind. var. x, t, order 2, (e) nonlinear, dep. var. z. ind. var. r, s, order 1, , 12.30. (a) hyperbolic, (b) hyperbolic (c) elliptic, (d) parabolic, (e) elliptic if x2 + y2 < 1, hyperbolic if a;2 + y2 > 1, parabolic if x2 + y2 = 1, 12.32. (6) x(Zx + y - 2)2, 12.33., 12.34. (a), 12.37. ux(0,t) = 0,, , (b), ux(L,t) = 0, (b) xz = x6 + x2 + Gj/4 - 68, , 12.41. (a) xz = F(x) + G(y), 12.42., , (a) u = F(x + y) + G(x - y), (b) u = eSzj^ _ 2x), (c) u - F(x + iy) + G(x - iy), , 12.43. (a) u = F(y - Zx) + ^, Z, , (6) y = F(x-t) + G(x + t)-t*, , (d) z = F(Bx + y) + G(y - x), (e) z = F(x + y) + xG(x + y), , (c), (d), , 12.44. u = F(x + iy) + G(x - iy) + xH(x + iy) + xJ(x - iy) + 4(x* + y2)*, 12.46. (a), (b), (c), (d)
Page 295 :
CHAP. 12], , PARTIAL DIFFERENTIAL EQUATIONS, , 12.47. (a), , 285, , (b) :, , 12.48. (a), 12.50. 150 - 5x, 12.51., 1253., , 12.72., , where, , are the positive roots of, , Jg(\) = 0, , 12.73., , 12.80. (a), , 12.81., , (a), , 12.82., 12.83. m/r where r > a is the distance from the center of the sphere, if r < a, , 12.85., , if r > a, , i f r > a, , 1 2 . 8 6, , if r > a, , 12.88. (a) u = 4e-2*-»», (b) u - lOe-x-M - 6e-4x-6t, (c) u = 6e~"*v* sin (n-a/2) + 3e~^« sin vx, (d) u = 46-T2' cos w* — 2e-9jr2t/4 cos3ir», , and
Page 296 :
Chapter 13, Complex Variables, and Confer ma I Mapping, FUNCTIONS, If to each of a set of complex numbers which a variable z may assume there corresponds one or more values of a variable w, then w is called a function of the complex, variable z, written w = /(z). The fundamental operations with complex numbers have, already been considered in Chapter 1., A function is single-valued if for each value of z there corresponds only one value of w,, otherwise it is multiple-valued or many-valued. In general we can write w = f(z) =, u(x, y) + iv(x, y), where u and v are real functions of x and y., Example 1. w = z2 = (x + iy)* - x2 - y2 + 2ixy = u + iv so that u(x, y) = x2 — yz, v(x, y) - Zxy., These are called the real and imaginary parts of w = *2 respectively., , Unless otherwise specified we shall assume that f(z) is single-valued. A function, which is multiple-valued can be considered as a collection of single-valued functions., LIMITS AND CONTINUITY, Definitions of limits and continuity for functions of a complex variable are analogous, to those for a real variable. Thus f(z) is said to have the limit I as z approaches za if,, given any « > 0, there exists a 8 > 0 such that \f(z) -l\<c whenever 0 < \z - z»| < 8., Similarly, f(z) is said to be continuous at z0 if, given any e > 0, there exists a 8 > 0, such that, whenever, Alternatively, f(z) is continuous at z0 if, , DERIVATIVES, If f(z) is single-valued in some region of the z plane the derivative of f(z), denoted by, f'(z), is defined as, W, , provided the limit exists independent of the manner in which Az -» 0. If the limit (1), exists for z = z0, then f(z) is called analytic at Zo. If the limit exists for all z in a region <5£,, then /(z) is called analytic in ^ In order to be analytic, /(z) must be single-valued and, continuous. The converse, however, is not necessarily true., We define elementary functions of a complex variable by a natural extension of the, corresponding functions of a real variable. Where series expansions for real functions, f(x) exist, we can use as definition the series with x replaced by z., Example 2. We define, From these we can show that ez = ex+i>> = ex(eos y + i sin y),, as well as numerous other relations., , 286
Page 297 :
CHAP. 13], , COMPLEX VARIABLES AND CONFORMAL MAPPING, , 287, , Example 3. We define ab as e blno even when a and b are complex numbers. Since e2klrf = 1,, it follows that <?** = eK4>+ZM an<j we define In« = In (pe**) = In p + i(<f> + 2kv)., Thus In z is a many-valued function. The various single-valued functions of which, this many-valued function is composed are called its branches., , Rules for differentiating functions of a complex variable are much the same as for, those of real variables., , Thus, , etc., , CAUCHY-RIEMANN EQUATIONS, , A necessary condition that w = f(z) = u(x, y) + iv(x, y) be analytic in a region <K, is that u and v satisfy the Cauchy-Riemann equations, (2), , (see Problem 13.7). If the partial derivatives in (2) are continuous in <R, the equations are, sufficient conditions that f(z) be analytic in <^., If the second derivatives of u and v with respect to x and y exist and are continuous,, we find by differentiating (2) that, (3), , Thus the real and imaginary parts satisfy Laplace's equation in two dimensions. Functions, satisfying Laplace's equation are called harmonic functions., INTEGRALS, , If f(z) is defined, single-valued and continuous in a region ^ we define the integral, of f(z) along some path C in ^ from point Zi to point zz, where Zi — Xi + iyi, z2 = xz + iy%, as, , with this definition the integral of a function of a complex variable can be made to, depend on line integrals for real functions already considered in Chapter 6. An alternative definition based on the limit of a sum, as for functions of a real variable, can also, be formulated and turns out to be equivalent to the one above., The rules for complex integration are similar to those for real integrals. An important, result is, , (-4), where M is an upper bound of |/(z)| on C, i.e. |/(z)| ^ M, and L is the length of the path C., CAUCHY'S THEOREM, , Let C be a simple closed curve. If /(z) is analytic within the region bounded by C, as well as on C, then we have Cauchy's theorem that, (5), where the second integral emphasizes the fact that C is a simple closed curve.
Page 298 :
288, , COMPLEX VARIABLES AND CONFORMAL MAPPING, , [CHAP.13, , Expressed in another way, (5) is equivalent to the statement t h a t h a s a, value independent of the path joining Zi and z2. Such integrals can be evaluated as, F(zz) - F(ZI) where F'(z) = f(z). These results are similar to corresponding results for, line integrals developed in Chapter 6., Example 4. Since f(z) = Zz is analytic everywhere, we have for any simple closed curve C, , Also,, , CAUCHY'S INTEGRAL FORMULAS, If f(z) is analytic within and on a simple closed curve C and a is any point interior, to C, then, («), where C is traversed in the positive (counterclockwise) sense., Also, the %th derivative of f(z) at z = a is given by, (7), , These are called Cauchy's integral formulas. They are quite remarkable because, they show that if the function f(z) is known on the closed curve C then it is also known, within C, and the various derivatives at points within C can be calculated. Thus if a, function of a complex variable has a first derivative, it has all higher derivatives as well., This of course is not necessarily true for functions of real variables., TAYLOR'S SERIES, Let f(z) be analytic inside and on a circle having its center at z = a. Then for all, points z in the circle we have the Taylor series representation of f(z) given by, , (*), See Problem 13.21., SINGULAR POINTS, A singular point of a function f(z) is a value of z at which f(z) fails to be analytic., If f(z) is analytic everywhere in some region except at an interior point z = a, we call, z = a an isolated singularity of f(z)., Example 5. If, , then z = 3 is an isolated singularity of /(«)., , POLES, If, where <f>(z) is analytic everywhere in a region including, z = a, and if n is a positive integer, then f(z) has an isolated singularity at z = a which is, called a pole of order n. If n = 1, the pole is often called a simple pole; if n = 2 it is, called a double pole, etc.
Page 299 :
CHAP. 13], , COMPLEX VARIABLES AND CONFORMAL MAPPING, , Example 6., , has two singularities:, , 289, , a pole of order 2 or double pole at, , 2 = 3, and a pole of order 1 or simple pole at z = —1., Example 7., , has two simple poles at z = ±2i., , A function can have other types of singularities besides poles. For example,, has a branch point at z = 0 (see Problem 13.36). The f u n c t i o n h a s a singularity, Ql fl 4J, , at z = 0. However, due to the fact that lim —^— is finite, we call such a singularity a, z, removable singularity., ~*, LAURENT'S SERIES, If /(z) has a pole of order n at z = a but is analytic at every other point inside and, on a circle C with center at a, then (z -a)nf(z) is analytic at all points inside and on C, and has a Taylor series about z = a so that, (9), , This is called a Laurent series for f(z). The part a0 + ai(z — a) + Oz(z — a)2 + • • • is called, the analytic part, while the remainder consisting of inverse powers of z — a is called the, principal part. More generally, we refer to the series 2 a*(z ~~ a)k as a Laurent series, fc=-oo, where the terms with k < 0 constitute the principal part. A function which is analytic, in a region bounded by two concentric circles having center at z = a can always be, expanded into such a Laurent series (see Problem 13.82)., It is possible to define various types of singularities of a function f(z) from its Laurent, series. For example, when the principal part of a Laurent series has a finite number of, terms and a-n¥= 0 while a-n-i, a-n-z, ... are all zero, then z = a is a pole of order n., If the principal part has infinitely many terms, z = a is called an essential singularity or, sometimes a pole of infinite order., 00, , Example 8. The function, , has an essential singularity at « = 0., , RESIDUES, The coefficients in (9) can be obtained in the customary manner by writing the coefficients for the Taylor series corresponding to (z — a)nf(z). In further developments, the, coefficient a-i, called the residue of /(z) at the pole z = a, is of considerable importance., It can be found from the formula, , (10), where n is the order of the pole. For simple poles the calculation of the residue is of, particular simplicity since it reduces to, , (11), RESIDUE THEOREM, If /(z) is analytic in a region <R except for a pole of order n at z = a and if C is any, simple closed curve in <3{. containing z = a, then /(z) has the form (9). Integrating (9),, using the fact that
Page 300 : 290, , COMPLEX VARIABLES AND CONFORMAL MAPPING, , [CHAP. 13, , (12), , (see Problem 13.13), it follows that, , (IS), i.e. the integral of f(z) around a closed path enclosing a single pole of f(z) is 2?rt times, the residue at the pole., More generally, we have the following important, Theorem 13.1. If f(z) is analytic within and on the boundary C of a region "K except at a, finite number of poles a, b, c, ... within "2^, having residues a-i, 6-1, c-i, ..., respectively, then, (**), , i.e. the integral of f(z) is 2id times the sum of the residues of f(z) at the, poles enclosed by C., Cauchy's theorem and integral formulas are special cases of this result which we call the, residue theorem., EVALUATION OF DEFINITE INTEGRALS, , The evaluation of various definite integrals can often be achieved by using the residue, theorem together with a suitable function f(z) and a suitable path or contour C, the choice, of which may require great ingenuity. The following types are most common in practice., 1., , is an even function., Consider <P F(z)dz along a contour C consisting of the line along the x axis, Jc, from — R to +R and the semi-circle above the x axis having this line as diameter., Then let R^<*>. See Problems 13.29, 13.30., , 2., , G is a rational function of sin 0 and cos 0., Let z = eie Then, , and dz^ieied0 or do =, , dz/iz. The given integral is equivalent to <l> F(z)dz where C is the unit circle, Jc, with center at the origin. See Problems 13.31, 13.32., 3., , is a rational function., Here we c o n s i d e r : where C is the same contour as that in Type 1., See Problem 13.34., , 4. Miscellaneous integrals involving particular contours. See Problems 13.35, 13.37.
Page 301 :
CHAP. 13], , COMPLEX VARIABLES AND CONFORMAL MAPPING, , 291, , CONFORMAL MAPPING, The analytic function w - f(z) = u(x, y) + iv(x, y) defines a transformation u - u(x, y),, v — v(x, y) which establishes a correspondence between points of the uv and xy planes., Suppose that under this transformation point (xo,y<>) of the xy plane is mapped into, point (u0, v0) of the uv plane [see Figs. 13-1 and 13-2] while curves Ci and C2 [intersecting, at (x0,yo)] are mapped respectively into curves C{ and C2' [intersecting at (UO,VQ)]. Then if, the transformation is such that the angle at (xo,y0) between Ci and C2 is equal to the, angle at (uo,v<>) between Ci and C2' both in magnitude and sense, the transformation or, mapping is said to be conformal at (x0, yo). A mapping which preserves the magnitudes, of angles but not necessarily the sense is called isogonal., , Fig. 13-1, , Fig. 13-2, , The following theorem is fundamental., Theorem 13.2. If f(z) is analytic and f'(z) ¥= 0 in a region <K, then the mapping w = f(z), is conformal at all points of "2^., For conformal mappings or transformations, small figures in the neighborhood of a, point zo in the z plane map into similar small figures in the w plane and are magnified, [or reduced] by an amount given approximately by |/'(zo)|2, called the area magnification, factor or simply magnification factor. Short distances in the z plane in the neighborhood, of z0 are magnified [or reduced] in the w plane by an amount given approximately by, |/'(z0)|, called the linear magnification factor. Large figures in the z plane usually map, into figures in the w plane which are far from similar., RIEMANN'S MAPPING THEOREM, Let C [Fig. 13-3] be a simple closed curve in the z plane forming the boundary of a, region •?(. Let C" [Fig. 13-4] be a circle of radius one and center at the origin [the unit circle], forming the boundary of region 9^' in the w plane. The region It? is sometimes called, the unit disk. Then Riemann's mapping theorem states that there exists a function, w = f(z), analytic in ^, which maps each point of "^ into a corresponding point of "5^' and, each point of C into a corresponding point of C', the correspondence being one to one., , Fig. 13-3, , Fig. 13-4
Page 302 :
292, , COMPLEX VARIABLES AND CONFORMAL MAPPING, , [CHAP. 13, , This function f(z) contains three arbitrary real constants which can be determined, by making the center of C' correspond to some given point in <2^ while some point on C', corresponds to a given point on C. It should be noted that while Riemann's mapping, theorem demonstrates the existence of a mapping function, it does not actually produce, this function., It is possible to extend Riemann's mapping theorem to the case where a region bounded, by two simple closed curves, one inside the other, is mapped into a region bounded by, two concentric circles., SOME GENERAL TRANSFORMATIONS, In the following a, ft are given complex constants while a, Go are real constants., 1. Translation., w = z+$, By this transformation, figures in the z plane are displaced or translated in, the direction of vector /?., 2. Rotation., w = ew<>z, By this transformation, figures in the z plane are rotated through an angle 0o., If 0o > 0 the rotation is counterclockwise, while if 0o < 0 the rotation is clockwise., 3. Stretching., w = az, By this transformation, figures in the z plane are stretched (or contracted), in the direction z if a > 1 (or 0 < a < 1). We consider contraction as a special, case of stretching., 4. Inversion., , w = 1/z, , 5. Linear Transformation., w = az + ft, This is a combination of the transformations of translation, rotation and, stretching., 6. Bilinear or Fractional Transformation., This is a combination of the transformations of translation, rotation, stretching, and inversion., MAPPING OF A HALF PLANE ON TO A CIRCLE, Let z0 be any point P in the upper half of the z plane denoted by <•£ in Fig. 13-5 below., Then the transformation, (15), maps this upper half plane in a one to one manner on to the interior <K' of the unit circle, \w\ = 1, Fig. 13-6 below, and conversely. Each point of the x axis is mapped on to the, boundary of the circle. The constant 00 can be determined by making one particular point, of the x axis correspond to a given point on the circle., In the above figures we have used the convention that unprimed points such as, A,B, C, etc., in the z plane correspond to primed points A',B', C', etc., in the w plane., Also, in the case where points are at infinity we indicate this by an arrow such as at, A and F in Fig. 13-5 which correspond respectively to A' and F' (the same point) in, Fig. 13-6. As point z moves on the boundary of "2?. [i.e. the real axis] from —<» (point A), to +00 (point F), w moves counterclockwise along the unit circle from A' back to A'.
Page 303 :
CHAP. 13], , COMPLEX VARIABLES AND CONFORMAL MAPPING, , Fig. 13-5, , 293, , Fig. 13-6, , THE SCHWARZ-CHRISTOFFEL TRANSFORMATION, Consider a polygon [Fig. 13-7] in the w plane having vertices at w\, Wz,..., wn with, corresponding interior angles alt«2, ...« n respectively. Let the points w\, w2, .. ,,wn map, respectively into points Xi,x2, ...,xn on the real axis of the z plane [Fig. 13-8]., , Fig. 13-7, , Fig. 13-8, , A transformation which maps the interior ^ of the polygon of the w plane on to the, upper half <3^' of the z plane and the boundary of the polygon on to the real axis is given by, (16), or, , (17), , where A and B are complex constants., The following facts should be noted:, 1. Any three of the points Xi,x2, .. .,xn can be chosen at will., 2. The constants A and B determine the size, orientation and position of the polygon., 3. It is convenient to choose one point, say xn, at infinity in which case the last factor, of (16) and (17) involving xn is not present., 4. Infinite open polygons can be considered as limiting cases of closed polygons., , SOLUTIONS OF LAPLACE'S EQUATION BY CONFORMAL MAPPING, The problem of determining a function which is harmonic, i.e. satisfies Laplace's, equation, in some region ^ [Fig. 13-7] and which takes prescribed values on the boundary, C is often called a Dirichlet problem. It can often be solved by mapping ^ into the unit, circle or upper half plane. In such case C is mapped into a corresponding boundary C', and boundary conditions on C are transformed into boundary conditions for C'. Since, Laplace's equation in ^ is transformed into Laplace's equation in <2?/ [see Problem 13.49],, the problem is thus reduced to solving Laplace's equation in ^' with boundary conditions, on C", which is generally simple to do. By then transforming back we obtain the required, solution., For illustrations of the procedure see Problems 13.51-13.54.
Page 304 :
294, , COMPLEX VARIABLES AND CONFORMAL MAPPING, , [CHAP. 13, , Solved Problems, FUNCTIONS, LIMITS, CONTINUITY, 13.1. Determine the locus represented by, (o) |«-2| = 8, (6) |«-2| = |* + 4|, (c) |z-8| + |« + 3| = 10., (a) Method 1., circle with center at (2,0) and radius 3., Method 2. \z — 2| is the distance between the complex numbers « = * + iy and 2 + Ot. If this, distance is always 3, the locus is a circle of radius 3 with center at 2 + (K or (2,0)., (b) Method 1., « = —1, a straight line., , sQUARING. WE FING, , Method 2. The locus is such that the distances from any point on it to (2,0) and (-4,0) are, equal. Thus the locus is the perpendicular bisector of the line joining (2,0) and (-4,0), or, * = —1., (e) Method 1. The locus is given by, or, 2, Squarine and simplifying, 25 + 3* = 5V(* + 3) + j/2. Squaring and, simplifying again yields, , an ellipse with semi-major and semi-minor axes of, lengths 5 and 4 respectively., Method 2. The locus is such that the sum of the distances from any point on it to (3,0) and, (-3,0) is 10. Thus the locus is an ellipse whose foci are at (-3,0) and (3,0) and whose major, axis has length 10., , 13.2. Determine the region in the z plane represented by each of the following,, (a) \z\ < 1., Interior of a circle of radius 1. See Fig. 13-9., , (6) K|z + 2i| S2., \z + 2i\ is the distance from z to -2i, so that |« + 2t| = 1 is a circle of radius 1 with center, at -2i, i.e. (0,-2); and \z + 2i\ = 2 is a circle of radius 2 with center at -2i. Then, 1 < \z + 2i\ S 2 represents the region exterior to \z + 2i\ = 1 but interior to or on \z + 2i\ = 2, See Fig. 13-10., ', ', , (c), , 7r/B^argz^Tr/2., Note that arg z = <t>, where z = Pe«*. The required region is the infinite region bounded, by the lines <f> = ?r/3 and </> = v/2, including these lines. See Fig. 13-11., , F, , »S-13-9, , Fig. 13-10, , Fig. 13-11, , 13.3. Express each function in the form u(x, y) + iv(x, y), where u and v are real:, (a) z\ (b) l/(l-z), (c) •&>, (d) In z.
Page 305 :
CHAP. 13], , COMPLEX VARIABLES AND CONFORMAL MAPPING, , 295, , (a) w = z3 = (x + %)3 = Xs + Sx2(iy) + 3a(%)2 + (tj/)3 = a3 + 3&;2j/ - 3wj/2 - tj/3, = »3 - Say2 + i(3a23/ - j/3), Then M(OJ,J/) = x3 - Sxy2, v(x,y) = 3x*y - y3., (b) w, Then, (c) e3z, , =, , eflix+iy), , = e3a:e3i!/, , = e3x (cos gy + i sin Sy), , and, , (d) ln« = ln(pei*) = lap + i</> — In V»2 + J/2 + * tan"1 j//w, 2, , 2, , M = ^ In (w + j/ ),, , M = e3*cos32/, v = e3x sin 3j/, and, , v = tan-ij/Ae, , Note that In z is a multiple-valued function (in this case it is infinitely many-valued) since, <t> can be increased by any multiple of 2jr. The principal value of the logarithm is denned as, that value for which 0 S 0 < 2a- and is called the principal branch of In z., , 13.4. Prove, , (a) sin (a; + iy) = sin a; cosh y + i cos a; sinh y, (b) cos (* + iy) — cos a; cosh y — i sin x sinh y, , We use the relations, , &* = cos z + i sin 2, e~ te = cos z — i sin «, from which, , Then, , Similarly,, , DERIVATIVES. CAUCHY-RIEMANN EQUATIONS, 13.5. Prove that -r-z, where z is the conjugate of z, does not exist anywhere., az, By definition,, , if this limit exists independent of the manner in, , which Az = Ace + tAj/ approaches zero. Then, , If Aj/ = 0, the required limit is, If Aa; — 0, the required limit is, These two possible approaches show that the limit depends on the manner in which Az -» 0, so, that the derivative does not exist; i.e. z is non-analytic anywhere.
Page 306 :
COMPLEX VARIABLES AND CONFORMAL MAPPING, , 296, , 13.6. (v a ) I, ', , f, , f, , i, , n, , [CHAP. 13, , d -^-. x ( b ) Determine where w i s non-analytic., dz, ', , (a) Method 1., provided z ¥= 1, independent of the manner in which Az -* 0., Method 2. The usual rules of differentiation apply provided z¥=l. Thus by the quotient rule, for differentiation,, , (b) The function is analytic everywhere except at z = 1, where the derivative does not exist; i.e., the function is non-analytic at z = 1., , 13.7. Prove that a necessary condition for w = f(z) = u(x, y) + iv(x, y) to be analytic in, a region is that the Cauchy-Riemann e q u a t i o n s b e satisfied in, the region., Since /(z) = f(x + iy) = u(x, y) + iv(x, y), we have, f(z + Az) = /[* + A* + i(y + Aj/)] = u(x + Ax, y + AV) + w(« + Aac, y + Aj/), Then, , If Aj/ = 0, the required limit is, , If A« = 0, the required limit is, , If the derivative is to exist, these two special limits must be equal, i.e.,, , so that we must have, Conversely, we can prove that if the first partial derivatives of u and v with respect to x and y, are continuous in a region, then the Cauchy-Riemann equations provide sufficient conditions for /(z), to be analytic., , 13.8. (a) If f(z) = u(x, y) + iv(x, y) is analytic in a region ^, prove that the one parameter, families of curves2 u(x, y) - Ci and v(x, y) = C2 are orthogonal families, (b) Illustrate, by using f(z) — z ., (a) Consider any two particular members of these families u(x, y) = u0, v(x, y) = v0 which intersect, at the point (x0,y0)., Since du — uxdx + uy dy — 0, we have, Also since
Page 307 :
CHAP. 13], , COMPLEX VARIABLES AND CONPORMAL MAPPING, , 297, , When evaluated at (x0, y0), these, represent respectively the slopes of the, two curves at this point of intersection., By the Cauchy-Riemann equations,, ux = vy, uy = —vx, we have the product, of the slopes at the point (x0, y0) equal to, , so that any two members of the respective families are orthogonal, and thus, the two families are orthogonal., (b) If /(«) = z2, then u - X* - y2, v = 2xy., The graphs of several members of, x2 — y2 = GI, 2xy = C2 are shown in, Fig. 13-12., , Fig. 13-12, , 13.9. In aerodynamics and fluid mechanics, the functions <f> and ^ in f(z) = <j> + i$, where, f(z) is analytic, are called the velocity potential and stream function respectively. If, <j> = x2 + 4x - y2 + 2y, (a) find 4, and (6) find /(«)., (a) By the Cauchy-Riemann equations,, (0, , Then, (2), , Method 1. Integrating (1), $ = 2xy + 4y + F(x)., Integrating (2), $ = 2xy — 2x + G(y)., These are identical if F(x) = —2x + e, G(y) = 4y + c where c is any real constant., ^ = 2xy + 4y — 2x + c., Method 2., , Thus, , Integrating (1), $ = 2xy + 4y + F(x). Then substituting in (2), 2y + F'(x) = 2y — 2 or, F'(x) = -2 and F(x) = -2x + c. Hence ^ = 2xy + ty - 2x + c., (b) From (a),, where ct is a pure imaginary constant., This can also be accomplished by noting that z = x + iy, z = x — iy so that, The result is then obtained by substitution; the terms involving z drop out., , INTEGRALS, CAUCHY'S THEOREM, CAUCHY'S INTEGRAL FORMULAS, 13.10. Evaluate, (a) along the parabola x = t, y = t2 where 1 S t g 2,, (b) along the straight line joining 1 + i and 2 + 4i,, (c) along straight lines from 1 + i to 2 + i and then to 2 + 4i., We have
Page 308 :
298, , COMPLEX VARIABLES AND CONFORMAL MAPPING, , [CHAP. 13, , Method 1., , (a) The points (1,1) and (2,4) correspond to t — 1 and t — 2 respectively. Then the above line, integrals become, , (6) The line joining (1,1) and (2,4) has the equation y — 1 =, we find, , 4, , 1, , (x — 1) or y = 3* — 2. Then, , (c) From 1 + * to 2 +1 [or (1,1) to (2,1)], y = 1, dy = 0 and we have, , From 2 +1 to 2 + 4i [or (2,1) to (2,4)], a; = 2, dx = 0 and we have, , Adding, (f + 3i) + (-30-9t) = -f - 6t., Method 2., , By the methods of Chapter 6 it is seen that the line integrals are independent of the path,, thus accounting for the same values obtained in (a), (b) and (c) above. In such case the integral, can be evaluated directly, as for real variables, as follows:, , 13.11. (a) Prove Cauchy's theorem: If /(z) is analytic inside and on a simple closed curve, C, then, (6) Under these conditions prove t h a t i s independent of the path joining, Pi and P2., («), , By Green's theorem (Chapter 6),, , where ^ is the region (simply-connected) bounded by C., Since /(«) is analytic, ^ = ^-, ^- = -^ (Problem 13.7), and so the above integrals are, -*, ox oy ox, oy, zero. Then Jd> /(z) dz = 0, assuming f'(z) [and thus the partial derivatives] to be continuous., c, , (6) Consider any two paths joining points Pt and P2 (see, Fig. 13-13). By Cauchy's theorem,, , Then, or, Fig. 13-13
Page 309 :
CHAP. 13], , COMPLEX VARIABLES AND CONPORMAL MAPPING, , 299, , i.e. the integral along Pl AP2 (path 1) = integral along Pt BP2 (path 2), and so the integral is, independent of the path joining Pl and P2., This explains the results of Problem 13.10, since f(z) = z2 is analytic., , 13.12. If f(z) is analytic within and on the boundary, of a region bounded by two closed curves Ci, and Cz (see Fig. 13-14), prove that, As in Fig. 13-14, construct line AB (called a crosscut) connecting any point on C2 and a point on Cj., By Cauchy's theorem (Problem 13.11),, Fig. 13-14, , since /(z) is analytic within the region shaded and also on the boundary. Then, <*), But, , i.e., Note that /(«) need not be analytic within curve C2., , 13.13. (a) Prove that, , where C is a simple closed, , curve bounding a region having z — a as interior point., (b) What is the value of the integral if n - 0, -1, -2, -3,... ?, (a) Let Cl be a circle of radius e having center at z = o, (see Fig. 13-15). Since (z — a)~n is analytic within, and on the boundary of the region bounded by C, and Ci, we have by Problem 13.12,, , To evaluate this last integral, note that on, Ci, \z — a\ = e or « — a = eeie and dz = itew de., The integral equals, , Fi, , S-13'15, , If n = 1, the integral equals, (b) For n = 0, —1, —2, ... the integrand is 1, (z — a), (z — a)2, ... and is analytic everywhere inside, Ci, including z = a. Hence by Cauchy's theorem the integral is zero., , 13.14. Evaluate, , where C is (a) the circle \z\ = 1, (b) the circle \z + i\ - 4., , (a) Since z = 3 is not interior to |z| = 1, the integral equals zero (Problem 13.11)., (6) Since z - 3 is interior to \z + i\ = 4, the integral equals 2vi (Problem 13.13).
Page 310 :
300, , COMPLEX VARIABLES AND CONFORMAL MAPPING, , [CHAP. 13, , 13.15. If f(z) is analytic inside and on a simple closed curve C, and a is any point within C,, prove that, Referring to Problem 13.12 and the figure of Problem 13.13, we have, , Letting z — a = eele, the last integral b e c o m e s B u t since /(z) is analytic,, it is continuous. Hence, , and the required result follows., , 13.16. Evaluate (a), , where C is the circle \z —1| = 3., , (a) Since z = ir lies within, , by Problem 13.15 with /(z) = cos z,, , (6), , by Problem 13.15, since z = 0 and z = — 1 are both interior to C., , 13.17. Evaluate, , dz where C is any simple closed curve enclosing 2 = 1., , Method 1. By Cauchy's integral formula,, If n = 2 and f(z) = 5z2 - 3z + 2, then /"(I) = 10. Hence, , Method 2. 5z2 - 3z + 2 = 5(z - 1)2 + 7(z - 1) + 4. Then, , by Problem 13.13.
Page 311 :
CHAP. 13], , COMPLEX VARIABLES AND CONFORMAL MAPPING, , 301, , SERIES AND SINGULARITIES, 13.18. For what values of z does each series converge?, The ttth term, , (a), , Then, , By the ratio test the series converges if |z| < 2 and diverges if |z| > 2. If |z| = 2 the, ratio test fails., However, the series of absolute v a l u e s c o n v e r g e s if |z| = 2, since, » .., i, 2 —5-2 converges., , n=l«, , Thus the series converges (absolutely) for |z| S 2, i.e. at all points inside and on the circle, |z| = 2., (6), , We have, , Then the series, which represents sin z, converges for all values of z., («), , The series converges if \z — i\ < 3, and diverges if |z — i\ > 3., oo, , If \z — i\ = 3, then z — i = 3eie and the series becomes 2 e<"9since the nth term does not approach zero as n -» »., "=1, , This series diverges, , Thus the series converges within the circle |z — i\ = 3 but not on the boundary., , 00, , 13.19. If 2 a"2" ig absolutely convergent for \z\ g R, show that it is uniformly convergent, n=0, , for these values of z., The definitions, theorems and proofs for series of complex numbers are analogous to those for, real series., In this case we have, by the Weierstrass M test that, , Since by hypothesis, , converges, it follows, , ' converges uniformly for |z| § R., , 13.20. Locate in the finite z plane all the singularities, if any, of each function and name, them., (a), , is a pole of order 3., , (6), , is a pole of order 2 (double pole); z = i and z = 1 — 2i are, poles of order 1 (simple poles)., , <c> -2 +"2^1 2 ' ™ < ^ 0 . Since z2 + 2« + 2 = 0 when, can write z» + 2z + 2 = {z - (-1 + i)}{z - (-1 - i)} = (z + 1 - i)(z + 1 + i)., The function has the two simple poles: z = — 1 + i and « — — 1 — i., , we
Page 312 :
302, , COMPLEX VARIABLES AND CONFORMAL MAPPING, appears t o b e a singularity., , (d), , However, s, , i, , n, , [CHAP. 13, c, , e, , i, , t is a, , removable singularity., Another method., Since, , we see that z = 0, , is a removable singularity., («), , This is a Laurent series where the principal part has an infinite number of non-zero terms., Then z = 1 is an essential singularity., , (/) e*., , This function has no finite singularity. However, letting 2 = 1/u, we obtain e1/" which, has an essential singularity at u = 0. We conclude that z = <*> is an essential singularity, of e*., In general, to determine the nature of a possible singularity of f(z) at z = <*>, we let, e = I/M and then examine the behavior of the new function at u — 0., , 13.21. If f(z) is analytic at all points inside and on a circle of radius R with center at a, and, if a + h is any point inside C, prove Taylor's theorem that, , By Cauchy's integral formula (Problem 13.15), we have, , (D, By division,, , (2), Substituting (2) in (1) and using Cauchy's integral formulas, we have, , where, Now when z is on C,, , and \z-a\= R, so that by (4), page 287, we have,, , since 2irR is the length of C,, As n -» », \Rn\ -» o. Then Rn -> 0 and the required result follows., If /(«) is analytic in an annular region rt •& \z - a\ S r2, we can generalize the Taylor series, to a Laurent series (see Problem 13.82). In some cases, as shown in Problem 13.22, the Laurent, series can be obtained by use of known Taylor series.
Page 313 :
CHAP. 13], , COMPLEX VARIABLES AND CONFORMAL MAPPING, , 303, , 13.22. Find Laurent series about the indicated singularity for each of the following, functions. Name the singularity in each case and give the region of convergence of, each series., («), , 2 = 1 is a pole of order 2, or double pole., The series converges for all values of 2 ¥• 1., (6), , 2 = 0 is an essential singularity., The series converges for all values of z ¥* 0., («), , 2 = ir is a removable singularity., The series converges for all values of 2., Then, , (d], , z = —1 is a pole of order 1, or simple pole., The series converges for values of 2 such that 0 < \z +1| < 1., M, , Case 1, 2 = 0., , Using the binomial theorem,, , 2 = 0 is a pole of order 1, or simple pole., The series converges for 0 < |2| < 2.
Page 314 :
304, , COMPLEX VARIABLES AND CONFORMAL MAPPING, Case 2, z = -2., , [CHAP. 13, , Let z + 2 = u. Then, , z = — 2 is a po/e o/ order 5., The series converges for 0 < |z + 2| < 2., , RESIDUES AND THE RESIDUE THEOREM, , 13.23. If /(z) is analytic everywhere inside and on a simple closed curve C except at z = a, which is a pole of order n so that, , where a-n =£ 0, prove that, (a), (6), (a) By integration, we have on using Problem 13.13, , Since only the term involving o_j remains, we call o_! the residue of /(z) at the pole z = a., (b) Multiplication by (z — a)n gives the Taylor series, («-«)»/(«) = a_n + a_ n + 1 (z-o) + • • • + o.^z-a)"-! + •••, Taking the (n — l)st derivative of both sides and letting z -» a, we find, , from which the required result follows., , 13.24. Determine the residues of each function at the indicated poles., («), , These are simple poles., Residue at z — 2 is, Residue at z = i is, Residue at z = —i is, , Then:
Page 315 :
CHAP. 13], , COMPLEX VARIABLES AND CONFORMAL MAPPING, , (&), , 305, , is a simple pole, z — —2 is a pole of order 3. Then:, Residue at 2 = 0 is, Residue at z = —2 is, , Note that these residues can also be obtained from the coefficients of 1/z and l/(z + 2) in, the respective Laurent series [see Problem 13.22(e)]., ze*t, \°> iz _ 3x2 '• * = 3, a pole of order 2 or double pole. Then:, Residue is, , (d) cot,?; z = 5v, a pole of order 1. Then:, Residue is, where we have used L'Hospital's rule, which can be shown applicable for functions of a complex, variable., , 13.25. If f(z) is analytic within and on a simple closed curve C except at a number of poles, a,b,c, ... interior to C, prove that, {sum of residues of /(z) at poles a, b, c, etc.}, Refer to Fig. 13-16., By reasoning similar to that of Problem 13.12, (i.e. by constructing cross cuts from C to C1(C2,C3,, etc.), we have, , For pole a,, , Fig. 13-16, , hence, as in Problem 13.23,, Similarly for pole b,, so that, Continuing in this manner, we see that, (sum of residues)
Page 316 :
306, , COMPLEX VARIABLES AND CONFORMAL MAPPING, , 13.26. Evaluate, , [CHAP. 13, , where C is given by (a) |z| = 3/2, (6) \z\ = 10., , Residue at simple pole, Residue at double pole « = —3 is, , (a) Since \z\ = 3/2 encloses only the pole z = 1,, the required integral, (6) Since |z[ = 10 encloses both poles z = 1 and z — —3,, the required integral, , EVALUATION OF DEFINITE INTEGRALS, M, , 13.27. If \f(z)\ ^^ for z = Re», where k > 1 and, M are constants, prove that lim ( f(z) dz = 0, R-K» JY, , where r is the semi-circular arc of radius R, shown in Fig. 13-17., By the result (4), page 287. we have, , Fig. 13-17, since the length of arc L — vR. Then, , 13.28. Show that for, Ifif R is large enough (say, R > 2, for example) so that M = 2, k = 4., Note that we have made use of the i n e q u a l i t y w i t h Zj = jB^e4** and, *, = !., , 13.29. Evaluate, Consider, , where C is the closed contour of Problem 13.27 consisting of the line from, , —R to R and the semi-circle r, traversed in the positive (counterclockwise) sense., Since «4 + l = 0 when z = e*1/*, eSirf/4, esvi/*t e™/4, these are simple poles of !/(«* +1). Only, the poles e^u and e3lri/4 lie within C. Then using L'Hospital's rule,
Page 317 :
CHAP. 13], , COMPLEX VARIABLES AND CONPORMAL MAPPING, , 307, , Residue at e^1*, , Residue at eSlri/*, , T, , h, , u, , s, , &, , i.e., , (*), , Taking the limit of both sides of (2) as R -» «° and using the results of Problem 13.28, we have, , Since, , the required integral has the value, , 13.30. Show that, The poles of, , enclosed by the contour C of Problem 13.27 are z — i of, , order 2 and z — —1 + i of order 1., Residue at z = i is, Residue at z = — 1 + i is, Then, or, , Taking the limit as R -» « and noting that the second integral approaches zero by Problem, 13.27, we obtain the required result., , 13.31. Evaluate, Let z - e*». Then, , so that, , where C is the circle of unit radius with center at the origin, as shown in Fig. 13-18 below.
Page 318 :
308, , COMPLEX VARIABLES AND CONFORMAL MAPPING, The poles of, , [CHAP. 13, , are the simple poles, , Fig. 13-18, Only —i/S lies inside C., BY l'hOSPITAL'S, , rESIDUE AT, rule., Then, , the required value., , 13.32. Show that, If, Then, , where C is the contour of Problem 13.31., The integrand has a pole of order 3 at z = 0 and a simple pole z = •£ within C., Residue at z — 0 is, Residue at z = \ is, Then, , as required., , M, , 13.33. If \f(z)\ ^ -^ for z- Reie, where k > 0 and M are constants, prove that, where r is the semi-circular arc of the contour in Problem 13.27 and m is a positive, constant., If, , Then
Page 319 :
CHAP. 13], , COMPLEX VARIABLES AND CONFORMAL MAPPING, , 309, , Now sin 9 S 2e/v for 0 S 6 £ v/2 [see Problem 14.3, page 327]. Then the last integral is, less than or equal to, As R -* <*> this approaches zero, since m and k are positive, and the required result is proved., , 13.34. Show that, Consider, , where C is the contour of Problem 13.27., , The integrand has simple poles at « = ±i, but only « = i lies within C., Residue at z = i is, Then, or, , i.e., and so, , Taking the limit as R-* <*> and using Problem 13.33 to show that the integral around T, approaches zero, we obtain the required result., , 13.35. Show that, The method of Problem 13.34 leads us to, consider the integral of eizfz around the contour, of Problem 13.27. However, since z — 0 lies, on this path of integration and since we cannot, integrate through a singularity, we modify that, contour by indenting the path at z = 0, as, shown in Fig. 13-19, which we call contour C', or ABDEFGHJA., Since z = 0 is outside C', we have, Fig. 13-19, or, , Replacing x by —* in the first integral and combining with the third integral, we find,, , or
Page 320 :
310, , COMPLEX VARIABLES AND CONFORMAL MAPPING, , [CHAP. 13, , 13.36. Let w--\fz define a transformation from the z plane to the w plane. A point moves, counterclockwise along the circle \z\ = 1. Show that when it has returned to its, starting position for the first time its image point has not yet returned, but that when, it has returned for the second time its image point returns for the first time., Let z = e™. Then w = Vz = e®'2. Let * = 0 correspond to the starting position. Then, x = 1 and w = 1 [corresponding to A and P in Figs. 13-20 and 13-21]., , Fig. 13-20, , Fig. 13-21, , When one complete revolution in the z plane has been made, 9 = Zir, z = 1 but w = ei9lz =, e = —1 so the image point has not yet returned to its starting position., However, after two complete revolutions in the z plane have been made, e = 4ir, z = 1 and, w = ei6'2 = e2™ = 1 so the image point has returned for the first time., It follows from the above that w is not a single-valued function of « but is a double-valued, function of z; i.e. given z, there are two values of w. If we wish to consider it a single-valued, function, we must restrict e. We can, for example, choose 0 S e < 2v, although other possibilities, exist. This represents one branch of the double-valued function w = Vz. In continuing beyond, this interval we are on the second branch, e.g. 2jr S 8 < 4jr. The point z = 0 about which the, rotation is taking place is called a branch point. Equivalently, we can insure that /(«) = Viz, will be single-valued by agreeing not to cross the line Ox, called a branch line., iv, , 13.37. Show that, Consider, , Since z = 0 is a branch point,, , choose C as the contour of Fig. 13-22 where AB and GH, are actually coincident with the x axis but are shown, separated for visual purposes., The integrand has the pole z = —1 lying within C., Residue at z = — 1 = e17* is, , Then, or, omitting the integrand,, , Fig. 13-22, We thus have, , where we have to use z = xe2*1 for the integral along GH, since the argument of z is increased by, 2v in going around the circle BDEFG.
Page 321 :
CHAP. 13], , COMPLEX VARIABLES AND CONFORMAL MAPPING, , 311, , Taking the limit as r -» 0 and R -» «° and noting that the second and fourth integrals, approach zero, we find, , or, , so that, , CONFORMAL MAPPING, 13.38. Consider the transformation w = f(z) where f(z) is analytic at z0 and /'(«<>) ^ 0., Prove that under this transformation the tangent at z0 to any curve C in the z plane, passing through z0 [Fig. 13-23] is rotated through the angle arg/'(zo)., , Fig. 13-23, , Fig. 13-24, , As a point moves from z0 to z0 + Az along C [Fig. 13-23] the image point moves along C' in the, w plane [Fig. 13-24] from w0 to t00 + Aw. If the parameter used to describe the curve is t, then, corresponding to the path z — z(t) [or x = x(t), y = y(t)] in the z plane, we have the path w = w(t), [or u = u(t), v = v(t)] in the w plane., The derivatives dz/dt and dw/dt represent tangent vectors to corresponding points on C and C'., Now, , and, in particular at Z0 and W0,, , u>, provided /(z) i s analytic a t z = z0. W, have from (1), , r, , i, , t, , i, , n, , g, , w, , e, (2), , so that, as required,, , (3), , Note that if /'(z0) = 0, then a is indeterminate. Points where /'(z) = 0 are called critical points., , 13.39. Prove that the angle between two curves Ci and Cz passing through the point z<> in, the z plane [see Figs. 13-1 and 13-2, page 291] is preserved [in magnitude and sense], under the transformation w = f(z), i.e. the mapping is conformal, if f(z) is analytic, at 20 and f'(z0) *£ 0., By Problem 13.38 each curve is rotated through the angle arg /'(z0). Hence the angle between, the curves must be preserved, both in magnitude and sense, in the mapping.
Page 322 :
312, , COMPLEX VARIABLES AND CONFORMAL MAPPING, , [CHAP. 13, , 13.40. If w = f(z) = u + iv is analytic in a region <!R, prove that, If /(«) is analytic in <R, then the Cauchy-Riemann equations, are satisfied in % Hence, , 13.41. If z<> is in the upper half of the z plane,, show that the bilinear transformation, maps the upper half, of the z plane into the interior of the, unit circle in the w plane, i.e. \w\ = 1., We have, , From Fig. 13-25 if z is in the upper half plane,, \z — z0\ S \z — z0\, the equality holding if and, only if z is on the x axis. Hence \w\ § 1, as, required., , Fig. 13-25, , THE SCHWARZ-CHRISTOFFEL TRANSFORMATION, , 13.42. Establish the validity of the Schwarz-Christoffel transformation., We must show that the mapping function obtained from, , ^ = A(z-x1)«i'"-i(z-x2)<*i'«-i---(z-xn)<'n'«-i, , W, , maps a given polygon of the w plane [Fig. 13-26] into the real axis of the z plane [Fig. 13-27]., To show this observe that from (1) we have, W, , As z moves along the real axis from the left toward xlt let us assume that w moves along a side, of the polygon toward wt. When z crosses from the left of x^ to the right of xv el — arg (z — xt), changes from JT to 0 while all other terms in (2) stay constant. Hence axgdw decreases by, (ajAr-1) arg (z —*!> = (<*!/"•-IV = «i — r or, what is the same thing, increases by ir-a-^ [an, increase being in the counterclockwise direction]., , Fig. 13-26, , Fig. 13-27
Page 323 :
CHAP. 13], , COMPLEX VARIABLES AND CONFORMAL MAPPING, , 313, , It follows from this that the direction through wt turns through the angle v — alt and thus w, now moves along the side w^w^ of the polygon., When z moves through xz> ei —ar& (z ~ *i) an^ *2 —arS (g ~ #2) change from v to 0 while, all other terms stay constant. Hence another turn through angle TT — <*2 in *he w plane is made., By continuing the process we see that as z traverses the x axis, w traverses the polygon, and, conversely., We can prove that the interior of the polygon (if it is closed) is mapped on to the upper, half plane by (1)., , 13.43. Prove that for closed polygons the sum of the exponents, in the Schwarz-Christoffel transformation (16) or (17), page 293, is equal to —2., The sum of the exterior angles of any closed polygon is 2jr. Then, (IT —at]) + (ir — a2) + ••• + (v — <*„), and dividing by —v, we obtain as required,, , =, , 2ir, , 13.44. If in the Schwarz-Christoffel transformation (16) or (17), page 293, one point, say x»,, is chosen at infinity, show that the last factor is not present., In (16), page 293, let A = Jf/(—*„)««/»-! where K is a constant. Then the right side of (16), can be written, fx — z\"»/'r~ *, K(z - «!)«!/"-1 (z - xz)aiiv-1 • • • (z - *n_t) «„-1/*-1 ( —x, ), \, , n /, , As *„-»», this last factor approaches 1; this is equivalent to removal of the factor., , 13.45. Determine a function which maps the region of Fig. 13-28 in the w plane on to the, upper half of the z plane of Fig. 13-29., , Fig. 13-28, , Fig. 13-29, , Let points P,Q,S and T [Fig. 13-28] map respectively into P',Q',S' and 2" [Fig. 13-29]., We can consider PQST as a limiting case of a polygon (a triangle) with two vertices at Q and, S and the third vertex P or T at infinity., By the Schwarz-Christoffel transformation, since the angles at Q and S are equal to w/2,, we have, , Integrating,, When g = l, w = b., , Hence, , When z = -l,w = -b., , Hence,, , (1) 6 = K sin-i (1) + B = Kjr/2 + B., (2) -b = K sin-i (-1) + B = -Kir/2 + B., , Solving (1) and (2) simultaneously, we find B = 0, K = 2b/v. Then
Page 324 :
314, , COMPLEX VARIABLES AND CONFORMAL MAPPING, , [CHAP. 13, , SOLUTIONS OF LAPLACE'S EQUATION BY CONFORMAL MAPPING, , 13.46. Show that the functions (a) x2-y2 + 2y and (6) sin x cosh y are harmonic in any, finite region "5?. of the z plane., ( a ) I f * = a2 - y2 + 2 y , w e h, monic in 9{., ( 6 ) I f * = s i n x cosh y , w e h, and * is harmonic in %, , a, a, , v, , e, v, , T, e, , h, , e, , T, , n, h, , a, e, , n, , d * i s harn, , 13.47. Show that the functions of Problem 13.46 are harmonic in the w plane under the, transformation z = w3., If, , z = w3, then x + iy = (u + iv)3 = u3 — 3«i>2 + i(Su2v — vs) and x = u3 — Buv2, y = 3w2v — v3., , (a) * = x2 - y2 + 2y - (u3- 3w-y2)2 - (Bu2v - v3)2 + 2(3w2v - v3), = u6 - 15w*v2 + 15MV - v6 + 6w2v - 2v3, Then, and, , as required., , (6) We must show that * = sin (u3 - 3wt>2) cosh (3w2i> - v3) satisfies |^ + |-f = 0. This can, readily be established by straightforward but tedious differentiation., This problem illustrates a general result proved in Problem 13.49., , 13.48. Prove that, , where w = f(z) is analytic and f(z) ¥* 0., , The function *(«, y) is transformed into a function &[x(u,v), y(u,v)] by the transformation., By differentiation we have, , Similarly,, , Adding,, , W
Page 325 :
CHAP. 13], , COMPLEX VARIABLES AND CONFORMAL MAPPING, Since u and v are harmonic,, , equations,, , 315, , Also, by the Cauchy-Riemann, Then, , Hence (1) becomes, , 13.49. Prove that a harmonic function §(x, y) remains harmonic under the transformation, w = f(z) where f(z) is analytic and f'(z) ¥= 0., This follows at once from Problem 13.48, since if, , and /'(z) ¥• 0, then, , 13.50. If a is real, show that the real and imaginary parts of w = In(2-a) are harmonic, functions in any region ^ not containing z = a., Method 1. If 'R, does not contain a, then w = In (z — a) is analytic in <^. Hence the real and, imaginary parts are harmonic in «£., Method 2. Let z — a — reie. Then if principal values are used for 9, w — u + iv = ln(z — a) =, In r + ie so that u = In r, v = 6., In the polar coordinates (r, e), Laplace's equation is, and by direct, substitution we find that u = In r and v = e are solutions if ^. does not contain r = 0, i.e., z — a., Method 3. If z — a = rete, then x — a = r cos 6, y = r sin e and, , r - V(* - «)2 + V2,, e = tan-1 {y/(x — a)}, Then w — u + iv = £ In {(x — a)2 + y2} + i tan"1 {y/(x — a)} and u = £ In {(x — a)2 + y2},, , v—, , 1, , tan* {y/(x — a)}. Substituting these into Laplace's equation —r + 7-5-z = 0, we find after, ox'' oy, straightforward differentiation that u and v are solutions if z ¥° a., , 13.51. Find a function harmonic in the upper half of the z plane, Im {z} > 0, which takes, the prescribed values on the x axis given by G(x) = J.f l x > 0, We must solve for 4>(a;, y) the boundary-value problem, , [0, , x <0, , V > 0;, This is a Dirichlet problem for the upper half plane [see Fig. 13-30],, The function A9 + B, where A and B are real, constants, is harmonic since it is the imaginary, part of A In z + Bi., To determine A and B note that the boundary, conditions are * = 1 for x > 0, i.e. 9 = 0 and, * = 0 for x < 0, i.e. e = ir. Thus, (1) 1 = A(0) + B,, (2) 0 = A(v) + B, from which A = —I/B-, B = 1., Then the required solution is, Pig. 13-30
Page 326 :
316, , COMPLEX VARIABLES AND CONFORMAL MAPPING, , [CHAP. 13, , 13.52. Solve the boundary-value problem, , where T0, Ti, T2 are constants., This is a Dirichlet problem for the upper half, plane [see Pig. 13-31]., The function A0t + Be2 + C where A, B and C, are real constants, is harmonic since it is the imaginary part of A In (* + 1) + B In (z - 1) + Ci., , Fig. 13-31, , To determine A, B, C note that the boundary conditions are: * = T2 for * > 1, i.e. ev = ez — 0;, 4> = Tl for —1 < * < 1, i.e. e^ = 0, 62 = ir; * = T0 for * < —1, i.e. ffj = v, e2 = v. Thus, (1), , T2 = A(0) + B(0) + C, , (2) Tt = A(0) + B(v) + C, , (3) T0 = A(w) + B(w) + C, , from which C - T2, B = (Tl - T2)/v, A = (T0- Tj/v., Then the required solution is, , 13.53. Find a function harmonic inside the unit circle \z\ = 1 and taking the prescribed, values given by, , on its circumference., , This is a Dirichlet problem for the unit circle [Fig. 13-32] in which we seek a function satisfying, Laplace's equation inside |z| = 1 and taking the values 0 on arc ABC and 1 on arc CDE., , Fig. 13-32, , Fig. 13-33, , We map the interior of the circle |«| = 1 on to the upper half of the w plane [Fig. 13-33] by, using the mapping f u n c t i o n o r o b t a i n e d by using (15), page 292, with, w and z interchanged., Under this transformation, arcs ABC and CDE are mapped on to the negative and positive real, axis A'B'C' and C'D'E' respectively of the w plane. Then the boundary conditions * = 0 on, arc ABC and * = 1 on arc CDE become respectively # = 0 on A'B'C' and <J> = 1 on C'D'E'., Thus we have reduced the problem to finding a function * harmonic in the upper half w plane, and taking the values 0 for u < 0 and 1 for u > 0. But this problem has already been solved in, Problem 13.51 and the solution (replacing x by u and y by v) is given by, (1), Now from w, , we findu, , these in (1), we find the required solution, , Then substituting
Page 327 :
CHAP. 13], , COMPLEX VARIABLES AND CONFORMAL MAPPING, , 317, , (2), or in polar coordinates (r, 6), where x = r cos e, y — r sin e,, , (3), , 13.54. A semi-infinite slab (shaded in Fig. 13-34) has its boundaries maintained at the, indicated temperatures where T is constant. Find the steady-state temperature., , Fig. 13-34, , Fig. 13-35, , We can solve this problem by methods of conformal mapping since the steady-state temperature, satisfies Laplace's equation., The shaded region of the z plane is mapped into the upper half of the w plane [Fig. 13-35], by the mapping function w = sin (a-z/a) which is equivalent to u = sin (irx/a) cosh (iryla), v —, cos (vx/a) sinh (try/a)., We must now solve the equivalent problem in the w plane. We use the method of Problem 13.52, to find that the solution in the w plane is, , and the required solution to the problem in the z plane is therefore, , Supplementary Problems, FUNCTIONS, LIMITS, CONTINUITY, 13.55. Describe the locus represented by (a) |« + 2 — 3t| = 5, (6) \z + 2| = 2|z -1|, (c) \z + 5\ - \z - 5| = 6., Construct a figure in each case., 13.56. Determine the region in the z plane represented by each of the following:, (a) |«-2 + i|i=4, (6) |z| S3, OSarg«s|,, Construct a figure in each case., , (c) |z-3| + |z + 3| < 10., , 13.57. Express each function in the form u(x, y) + iv(x, y), where u and v are real,, (a) «« + 2iz, (6) z/(3 + z), (c) e*1, (d) ln(l + «)., 13.58. Prove that (a) lim «« = z*, Z-+ZQ, , (6) /(«) = «2 is continuous at z = «0 directly from the definition., , 13.59. (a) If z = a, is any root of z5 - 1 different from 1, prove that all the roots are 1, w, w2, u3, w4., (6) Show that 1 + a + «2 + os + a* = 0., (c) Generalize the results in (a) and (6) to the equation z» = 1.
Page 328 :
318, , COMPLEX VARIABLES AND CONFORMAL MAPPING, , [CHAP. 13, , DERIVATIVES, CAUCHY-RIEMANN EQUATIONS, 13.60. (a) If w = f(z) = z + -, find ^ directly from the definition., z, az, (6) For what finite values of z is /(z) non-analytic?, 13.61. Given the function w = z*. (a) Find real functions u and v such that w = u + iv. (b) Show that, the Cauchy-Riemann equations hold at all points in the finite z plane, (c) Prove that u and v are, harmonic functions, (d) Determine dw/dz., 13.62. Prove that /(z) = z\z\ is not analytic anywhere., 13.63. Prove that /(z) =, , z, , &, , is analytic in any region not including z = 2., , 13.64. If the imaginary part of an analytic function is 2#(1 — y), determine (a) the real part, (6) the, function., 13.65. Construct an analytic function /(z) whose real part is e~x(x cos y + y sin y) and for which /(O) = 1., 13.66. Prove that there is no analytic function whose imaginary part is x2 — 2y., 13.67. Find /(z) such that /'(z) = 4z - 3 and /(I +i) = -3i., 13.68. If z = pe'* and /(z) = u(p, <f>) + iv(p, <t>), where p and 0 are polar coordinates, show that the, Cauchy-Riemann equations are, , INTEGRALS, CAUCHY'S THEOREM, CAUCHY'S INTEGRAL FORMULAS, 13.69. Evaluate, (a) along the path x = 2t + 1, y = 4t2 - t - 2 0 S t § 1,, (b) along the straight line joining 1 — 2i and 3 + i,, (c) along straight lines from 1 — 2i to 1 + i and then to 3 + i., 13.70. E v a l u a t e w h e r e C is the upper half of the circle \z\ = 1 traversed in the positive, direction., 13.71. Evaluate, , , where C is the circle (a) |z| = 2, (6) \z — 3| = 2., , 13.72. E v a l u a t e d z ,, , where C is: (a.) a square with vertices at —1 — i, — 1 + i, —3 +1,, , —B-i; (b) the circle |z + i| = 3; (c) the circle |z| = \/2., 13.73. Evaluate, , (a), , (6), , where C is any simple closed curve enclosing z = 1., , 13.74. Prove Cauchy's integral formulas., [Hint. Use the definition of derivative and then apply mathematical induction.], 13.75. If f(z) is analytic inside and on the circle |z — a| = R, prove Cauchy's inequality, namely,, , where |/(z) = M\ on the circle., , [Hint: Use Cauchy's integral formulas.], , SERIES AND SINGULARITIES, 13.76. For what values of z does each series converge?, («), , (6), , (c)
Page 329 :
CHAP. 13], , COMPLEX VARIABLES AND CONFORMAL MAPPING, , 13.77. Prove that the s e r i e s i s (a) absolutely convergent,, |*| S 1., , 319, , (6) uniformly convergent for, , 13.78. Prove that the s e r i e s c o n v e r g e s uniformly within any circle of radius R such that, |z + i\ < R < 2., 13.79. Locate in the finite z plane all the singularities, if any, of each function and name them:, , (a), , (6), , (c), , (d), , (/), , (e), , 13.80. Find Laurent series about the indicated singularity for each of the following functions, naming the, singularity in each case. Indicate the region of convergence of each series., («), , (6), , (c), , 13.81. Find a Laurent series expansion for the function, 1< |z| < 2 and diverges elsewhere., Hint:, , /(«) = . , •.u, , (z + )(z, , • »-,, ', , which converges, , for, , Write, , 13.82. If o + h is any point in the annular region bounded by Ct and C2, and /(z) is analytic in this region,, prove Laurent's theorem that, , where, C being any closed curve in the angular region surrounding Cj., , Hing wrint, , and expand, , in two, , different ways., RESIDUES AND THE RESIDUE THEOREM, 13.83. Determine the residues of each function at its poles:, , (a), , (b), , (c), , (d), , 13.84. Find the residue of ez* tan z at the simple pole z — 3n-/2., 13.85. Evaluate, , where C is a simple closed curve enclosing all the poles., , 13.86. If C is a simple closed curve enclosing z — ±i, show that, , 13.87. If f(z) = P(z)/Q(z),, , where P(z) and Q(z) are polynomials such that the degree of P(z) is at least, , two less than the degree of Q(z), prove that J<j> /(«) dz = 0, where C encloses all the poles of /(«)., , c, , EVALUATION OF DEFINITE INTEGRALS, Use contour integration to verify each of the following, 1 3 . 8 8, , 13.90, , 13.89., , 13.91.
Page 330 :
320, , COMPLEX VARIABLES AND CONFORMAL MAPPING, , 13.92., , 13-95., , 13.93., , 13.96., , 13.94., , 13.97., , [CHAP. 13, , 13.98., 13.99., 13.100., , 13.103., , 13.101., , 13.104., , 13.102., , 13.105., , 13.106., , [Hint. C o n s i d e r w h e r e C is a rectangle with vertices, at (-#,0), (R,0), (R,v), (-R,*-). Then let R -» <*>.], , CONFORMAL MAPPING, 13.107. Prove that (a) w = z + 0 represents a translation,, az + b represents a stretching [or contraction]., , (6) w — eie«z represents a rotation, (c) w =, , 13.108. Prove that (a) w — az + ft represents a combination of translation, rotation and stretching., 13.109. Prove that, , represents a combination of translation, rotation, stretching and inversion, , if aS — /3y ¥> 0. Discuss the case a8 — /3y = 0., 13.110. (a) Prove that under the transformation w = (z — i)/(iz — 1) the region Im {z} S 0 is mapped into, the region \w\ S 1. (6) Into what region is Im {z} £ 0 mapped under the transformation?, 13.111. Determine the equation of the curve in the w plane into which the straight line x + y = 1 is, mapped under the transformations (a) w = z2, (6) w — 1/z., 13.112. Show t h, graphically., , a, , t, , m, , a, , p, , s the unit circle on to a wedge-shaped region and illustrate, , THE SCHWARZ-CHRISTOFFEL TRANSFORMATION, 13.113. Use the Sehwarz-Christoffel transformation to determine a function which maps each of the, indicated regions in the w plane on to the upper half of the z plane., (a), , Fig. 13-36, , Fig. 13-37
Page 331 :
CHAP. 13], , COMPLEX VARIABLES AND CONFORMAL MAPPING, , Fig. 13-38, , Fig. 13-39, , Fig. 13-40, , Fig. 13-41, , Fig. 13-42, , Fig. 13-43, , 321, , 13.114. Find a function which maps the upper half plane on to the interior of a triangle with vertices at, w = 0,1, i corresponding to z = 0,1, » respectively., 13.115. Show that the functions (a) 2xy + y3 — Sx2y, (b) e~x sin y are harmonic., 13.116. Show that the functions of Problem 13.115 remain harmonic under the transformations, (b) z = sin w., , (a) z — w2,, , 13.117. Find a function harmonic in the upper half z plane Im {z} > 0 which takes the prescribed values, on the x axis given by G(x) =, , 13.118. Work Problem 13.117 if G(x) -, , 13.119. Find a function harmonic inside the circle |z| = 1 and taking the values F(e) =, on its circumference., 13.120. Work Problem 13.119 if F(e) =
Page 332 :
322, , COMPLEX VARIABLES AND CONFORMAL MAPPING, , [CHAP. 13, , 13.121. Find the steady-state temperature at the point (5,2) in the shaded region of Fig. 13-44 below if the, temperatures are maintained as shown., , Fig. 13-44, , Fig. 13-45, , 13.122. An infinite conducting plate has in it a circular hole ABCD of unit radius [Fig. 13-45 above]., Temperatures of 20°C and 80°C are applied to arcs ABC and ADC and maintained indefinitely., Find the steady-state temperature at any point of the plate., , Answers to Supplementary Problems, 13.55. (a) Circle (x + 2)2 + (y - 3)2 = 25, center (-2,3), radius 5., (6) Circle (x-2)* + y* = 4, center (2,0), radius 2., (c) Branch of hyperbola »2/9 — j/2/16 = 1, where * B 3., 13.56. (a) Boundary and exterior of circle (x - 2)2 + (y + I)2 = 16., (6) Region in the first quadrant bounded by w2 + y2 = 9 the x axis and the line y = x., (c) Interior of ellipse «2/25 + j/2/16 = 1., 13.57. (a) M = x3 - Sxj/2 - Zy, v = Sx*y -ys + 2x, , W, (e) u = ex ~v cos 2xy, v = ex*~y* sin 2xy, (d) u = £ln{(l + a02+j,2}, „ = tan-»j-2~£+2fcr, fe = 0,±1,±2,..., 13.60. (a) 1-1/z2, (6) z = 0, 13.61. (a) u = x* - 6»2j/2 + y\ v = 4x3y-4xy3, , (d) 4z», , 13.64. (a) y*-x*-2y + c, (b) Ziz -z* + c, where c is real, 13.65. ze~» + 1, 13.67. /(«) = 2z2 - 3« + 3 - 4t, 13.69. 17 + 19i in all cases, 13.71. (a) 0,, , (6) 5«/2, , 13.72. (a) -8«/3, , (6) -2vi, , (c) 2*i/3~, , 13.73. (a) -2vi, , (b) vie/3, , 13.76. (a) all?, , (6) |z-i| < 1, , (c) z = -l±i
Page 333 :
CHAP. 13], , COMPLEX VARIABLES AND CONPORMAL MAPPING, , 13.79. (a) z = — £, pole of order 4, (6) « = 1, simple pole; z = —2, double pole, (c) Simple poles z — —1 ± i, 13.80.(a),simplepole,al z¥•ir, , (d) z = 0, essential singularity, (e) z — ir/3, removable singularity, (/) z = —2t, double poles, simple pole, , (b), , (b)essential singularity, all z¥= 0, , (c), , double pole, 0 < \z - 1| < 4, , 13.81., 13.83. (a) z = 2; 7/4, « = -2; 1/4, (6) « = 0; 8/25, « =-5; -8/25, , (c) 2 = 2; £<2e2«, (d) z = i; 0, z = -t; 0, , 13.84. -«8Tt/2, 13.85., , -8«, , 13.111. (a) w2 + 2v = 1,, 13.113. (a) w = z3,, , (6) w2 + 2wv + 2v2 = u + v, , (b) w = cosh Grz/2),, , 13.114., 13.117. 1 - (2M tan-i (y/x), 13.118., , 13.119., 13.121. 45.9° C, , (c) w = e",, , (d) w = z*™, , 323
Page 334 :
Chapter 14, , THE COMPLEX INVERSION FORMULA, If F(s) = £ {/(*)}, then £-* (F(s)} is given by, (*), , and f(t) = 0 for t < 0. This result is called the complex inversion integral or formula. It, is also known as Bromwich's integral formula. The result provides a direct means for, obtaining the inverse Laplace transform of a given function F(s)., The integration in (1) is to be performed along a line s = y + iy in the complex plane, where s = x + iy. The real number y is chosen so that s = y lies to the right of all the, singularities (poles, branch points or essential singularities) but is otherwise arbitrary., THE BROMWICH CONTOUR, , In practice, the integral in (1) is evaluated, by considering the contour integral, (*), where C is the contour of Fig. 14-1. This contour, sometimes called the Bromwich contour,, is composed of line AB and the arc BJKLA of, a circle of radius R with center at the origin O., If we represent arc BJKLA by r, it follows, from (1) that since T = -\/R" — y2,, , Fig. 14-1, , (3), , USE OF RESIDUE THEOREM IN FINDING INVERSE LAPLACE TRANSFORMS, , Suppose that the only singularities of F(s) are poles all of which lie to the left of the, line s = y for some real constant y. Suppose further that the integral around r in (5), approaches zero as /?-*<». Then by the residue theorem we can write (3) as, /(*) =, =, , sum of residues of est F(s) at poles of F(s), 2 residues of estF(s) at poles of F(s), 324, , ^
Page 335 :
CHAP. 14], , COMPLEX INVERSION FORMULA FOR LAPLACE TRANSFORMS, , 325, , A SUFFICIENT CONDITION FOR THE INTEGRAL AROUND r, TO APPROACH ZERO, , The validity of the result (4) hinges on the assumption that the integral around r in (3), approaches zero as R-* °°. A sufficient condition under which this assumption is correct, is supplied in the following, Theorem 14-1. If we can find constants M > 0, k > 0 such that on r (where s = Re19),, (5), then the integral around r of estF(s) approaches zero as R-*•<*>, i.e.,, (*), , The condition (5) always holds if F(s) = P(s)/Q(s) where P(s) and Q(s) are polynomials, and the degree of P(s) is less than the degree of Q(s)., The result is valid even if F(s) has other singularities besides poles., MODIFICATION OF BROMWICH CONTOUR, IN CASE OF BRANCH POINTS, If F(s) has branch points, extensions of the, above results can be made provided that the, Bromwich contour is suitably modified. For, example, if F(s) has only one branch point at, s = 0, then we can use the contour of Fig. 14-2., In this figure, BDE and LNA represent arcs of, a circle of radius R with center at origin O,, while HJK is the arc of a circle of radius e with, center at 0. For details of evaluating inverse, Laplace transforms in such cases see Prob. 14.9., , Fig. 14-2, , CASE OF INFINITELY MANY SINGULARITIES, , If we wish to find the inverse Laplace transform of functions which have infinitely, many isolated singularities, the above methods can be applied. In such case the curved, portion of the Bromwich contour is chosen to be of such radius Rm so as to enclose only a, finite number of the singularities and so as not to pass through any singularity. The, required inverse Laplace transform is then found by taking an appropriate limit as m-*•<*>., See Problems 14.13 and 14.14., APPLICATIONS TO BOUNDARY-VALUE PROBLEMS, , The method of Laplace transforms combined with the complex inversion formula provide powerful tools in the solution of various boundary-value problems arising in science, and engineering. See Problems 14.15-14.19.
Page 336 :
326, , COMPLEX INVERSION FORMULA FOR LAPLACE TRANSFORMS, , [CHAP. 14, , Solved Problems, THE COMPLEX INVERSION FORMULA, 14.1. Establish the validity of the complex inversion formula., We have, by definition,, , Then, , Letting s = y + iy, ds = i dy, this becomes, , by Fourier's integral theorem [see Chapter 8]. Thus we find, as required., In the above proof, we assume that e~yuf(u), , is absolutely integrable in (0,«°), i.e., , converges, so that Fourier's integral theorem can be applied., , To insure this, , condition it is sufficient that f(t) be of exponential order y where the real number y is chosen so that, the line # = y in the complex plane lies to the right of all the singularities of F(s). Except for, this condition, y is otherwise arbitrary., , 14.2. Let r denote the curved portion BJPKQLA, of the Bromwich contour [Fig. 14-3] with, equation s = Rew, 00 S= 6 S= 2ir — 00, i.e. r is, the arc of a circle of radius R with center, at O. Suppose that on r we have, , where k > 0 and M are constants. Show, that, Fig. 14-3, , If r1( r2, T3 and T4 represent arcs BJ, JPK,, KQL and LA respectively, we have, , Then if we can show that each of the integrals on the right approach zero as R -» «, proved the required result. To do this we consider these four integrals., Case 1., , Integral over T1 or BJ., , Along Tj we have, since s = Reie, e0 g 8 ^ ir/2,, , we, , will have
Page 337 :
CHAP. 14], , COMPLEX INVERSION FORMULA FOR LAPLACE TRANSFORMS, , 327, , Then, , where we have used the given condition \F(s)\ § M/Rk, where 00 = ir/2 — 00 = sin"1 (y/R)., , on r t and the transformation, , e = ir/2 — $, , Since sin <f> S sin <f>0 S cos «0 = y/R, this last integral is less than or equal to, , But as R-*<*>, this last quantity approaches zero [as can be seen by noting, for example, that, sin-1 (y/R) ~ y/R for large R], Thus lim 7X = 0., R-+ oo, , Case 2., , Integral over T2 or JPK., , Along T2 we have, since s = Rew, ir/2 £ e S v,, , Then, as in Case 1, we have, , upon letting 0 = v/2 + <f>., Now sin 0 S 2<f>/ir for 0 S <f> S ir/2 [see Problem 14.3], so that the last integral is less than, or equal to, which approaches zero as, Case 3., , Thus, , Integral over Ts or KQL., , This case can be treated in a manner similar to Case 2 [see Problem 14.28(a)]., Case 4., , Integral over l\ or LA., , This case can be treated in a manner similar to Case 1 [see Problem 14.28(6)]., , 14.3. Show that sin 0 § 2<£/ir for 0 ^ <£ g ir/2., Method 1., , Geometrical proof., , From Fig. 14-4, in which curve OPQ represents, an arc of the sine curve y = sin 0 and y = 2<j>/ir, represents line OP, it is geometrically evident that, sin 0 g 2(6/jr for 0 S 4, S jr/2., Method 2., , Analytical proof., , Fig. 14-4, , Consider G(<&) = &B£. We have, , <t>, , |?, , =, , G'(0), , =, , *CO»^-riH», , (i)
Page 338 :
328, , COMPLEX INVERSION FORMULA FOR LAPLACE TRANSFORMS, If ., , [CHAP. 14, , then, (2), , Thus for 0 g -f < T/2, H'(<j>) S 0 and #(0) is a decreasing function. Since H(0) = 0, it follows, that H(<t>) g 0. Then from (1) we see that G'(<f>) £ 0, or <?(#) is a decreasing function. Denning, G(0) — lim G(0) = 1, we see that G(0) decreases from 1 to 2Ar as <f> goes from 0 to ?r/2. Thus, cf>-»0, , from which the required result follows., , USE OF RESIDUE THEOREM IN FINDING INVERSE LAPLACE, , (, TRANSFORMS, , 14.4. Suppose that the only singularities of F(s) are poles which all lie to the left of the line, x = y for some real constant y. Suppose further that F(s) satisfies the condition, given in Problem 14.2. Prove that the inverse Laplace transform of F(s) is given by, f(t), , =, , sum of residues of estF(s) at all the poles of F(s), , We have, where C is the Bromwich contour of Problem 14.2 and T is the circular arc BJPKQLA of Fig. 14-3., By the residue theorem,, —: <fc est F(s) ds =, 2rtJ c, =, , sum of residues of est F(s) at all poles of F(s) inside C, 2 residues inside C, , Thus, Taking the limit as R -» °°, we find by Problem 14.2,, f(t), , = sum of residues of est F(s) at all the poles of F(s), , 14.5. (a) Show that, (6) Find the residue of, (c) Evaluate, (a) For s = Re^, we have, , satisfies the condition in Problem 14.2., at the pole s — 2., by using the complex inversion formula.
Page 339 :
CHAP. 14], , COMPLEX INVERSION FORMULA FOR LAPLACE TRANSFORMS, , 329, , for large enough R (e.g. R > 4). Thus the condition in Problem 14.2 is satisfied when fc = 1,, M = 2. Note that in establishing the above we have used the result |«t — z2| ^ |zi| — |«2| which, follows from the result [a + b\ S \a\ + |6| if a = «i — «2, & = «2(6) The residue at the simple pole s — 2 is, , (c) By Problem 14.4 and the results of parts (a) and (6), we see that, sum of residues of est f(s), , =, , e2t, , Note that the Bromwich contour in this case is chosen so that y is any real number greater, than 2 and the contour encloses the pole s — 2., , 14.6. Evaluate, , by using the method of residues., , Since the function whose Laplace inverse is sought satisfies condition (5) of the theorem on, page 325 [this can be established as in Problem 14.5], we have, , =, , 2 residues of, , at poles s — —1 and s = 2, , Now the residue at simple pole s = —1 is, , and the residue at double pole s = 2 is, , Then, , 14.7. Evaluate, As in Problem 14.6, the required inverse is the sum of the residues of, , at the poles « = —1 and s = 1 which are of orders three and two respectively.
Page 340 :
330, , COMPLEX INVERSION FORMULA FOR LAPLACE TRANSFORMS, Now the residue at s — —1 is, , and the residue at 8 = 1 is, , Then, , 14.8. Evaluate, We have, The required inverse is the sum of the residues of, , at the poles s = i and s = —i which are of order two each., Now the residue at s = i is, , and the residue at s = —i is, , which can also be obtained from the residue at s = i by replacing i by —i. Then, , Compare with Problem 4.37, page 113., , INVERSE LAPLACE TRANSFORMS OF, FUNCTIONS WITH BRANCH POINTS, , 14.9. Find, , by use of the complex, , inversion formula., By the complex inversion formula, the required inverse Laplace transform is given by, , w, Since 8 = 0 is a branch point of the integrand,, we consider, , [CHAP. 14
Page 341 :
CHAP. 14], , COMPLEX INVERSION FORMULA FOR LAPLACE TRANSFORMS, , 331, , where C is the contour of Fig. 14-5 consisting of the line AB (s = y + iy), the arcs BDE and LNA, of a circle of radius R and center at origin O, and the arc HJK of a circle of radius e with center, at 0., Since the only singularity s = 0 of the integrand is not inside C, the integral on the left is zero, by Cauchy's theorem. Also, the integrand satisfies the condition of Problem 14.2 so that on taking, the limit as R -* <*> the integrals along BDE and LNA approach zero. It follows that, , (2), , Along J3H, s = xe™, V« — V* «rt/2 = *V* and as s goes from —R to —«, x goes from R to e., Hence we have, , Similarly, along KL, s = xe-™, \fs = -\fx e~'!ri/2 = —i\fx, from « to R. Then, , and as s goes from —e to —R, x goes, , Along HJK, s = eeli and we have, , Thus (2) becomes, , Since the limit can be taken underneath the integral sign, we have, , and so we find ($), , (3), , This can be written (see Problem 14.10) as, , U)
Page 342 :
332, , COMPLEX INVERSION FORMULA FOR LAPLACE TRANSFORMS, , 14.10. Prove that, , an, , d, , tnus, , [CHAP. 14, , establish the final result, , (4) of Problem 14.9., Letting x = t&, the required integral becomes, , Then differentiating with respect to a and using Problem 9, we have, , Hence, using the fact that / = 0 when a, = 0,, , and the required result is established., , 14.11. Find XT1 {<roVr}., H -C {/(*)} = F(s), then we have £{f'(t)} = «F(s)-/(0) = »/(«) if F(0) = 0. Thus if, •C-itfX*)} = /(«) and /(O) = 0, then ^-i{«^(«)} = /'(*)., By Problems 14.9 and 14.10, we have, , so that /(O) = 0 and, Then it follows that, , INVERSE LAPLACE TRANSFORMS OF FUNCTIONS WITH INFINITELY, MANY SINGULARITIES, 14.12. Find all the singularities of, Because of the presence of V«, it would appear that * = 0 is a branch point. That this is, not so, however, can be seen by noting that, , from which it is evident that there is no branch point at 8 = 0. However, there is a simple pole, at 8 = 0., The function F(s) also has infinitely many poles given by the roots of the equation
Page 343 :
CHAP. 14], , COMPLEX INVERSION FORMULA FOR LAPLACE TRANSFORMS, , These occur where, from which, , e 2 ^ = -1 = 6'rt+2k1ri, Vs = (k + $)*i, , or, , k = 0, ±1, ±2, ..., s = -(fc + £)V2, , These are simple poles., Thus F(s) has simple poles at, s = 0 and s = sn, , where, , sn = — (n — £)2ir2, w = 1,2,3,..., , 14.13. Find, The required inverse can be found by using, the Bromwich contour of Fig. 14-6. The line AB, is chosen so as to lie to the right of all the poles, which, as seen in Problem 14.12, are given by, We choose the Bromwich contour so that the, curved portion BDEFGHA is an arc of a circle, rm with center at the origin and radius, where m is a positive integer. This choice insures, that the contour does not pass through any of, the poles., , Fig. 14-6, , We now find the residues of, at the poles. We have:, Residue at 8 = 0 is, Residue at s = -(n-$)W, n = 1,2,3,... is, , If Cm is the contour of Fig. 14-6, then, , Taking the limit as m -> » and noting that the integral around rm approaches zero, we find, , 333
Page 344 :
334, , COMPLEX INVERSION FORMULA FOR LAPLACE TRANSFORMS, , 14.14. Find, i.e.,, , [CHAP. 14, , where 0 < x < a., , The f u n c t i o n h a s poles at s = 0 and at values of s for which cosh sa = 0,, , Because of the presence of sz, it would appear that 8 = 0 is a pole of order two. However, by, observing that near 8 = 0,, , we see that s = 0 is a pole of order one, i.e. a simple pole. The poles sk are also simple poles., Proceeding as in Problem 14.13, we obtain the residues of es*/(s) at these poles., Residue at s — 0 is, , using L'Hospital's rule., Residue at s = sk is, , By an appropriate limiting procedure similar to that used in Problem 14.13, we find on taking, the sum of the residues the required result,
Page 345 :
CHAP. 14], , COMPLEX INVERSION FORMULA FOR LAPLACE TRANSFORMS, , 335, , APPLICATIONS TO BOUNDARY-VALUE PROBLEMS, 14.15. A semi-infinite solid x > 0 [see Fig. 14-7], is initially at temperature zero. At time, t = 0, a constant temperature w» > 0 is, applied and maintained at the face x = 0., Find the temperature at any point of the, solid at any later time t > 0., The boundary-value problem for the determination of the temperature u(x, t) at any point x, and any time t is, , u(x, 0) = 0,, , u(0, t) = MO,, , Fig. 14-7, , \u(x, t)\ < M, , where the last condition expresses the requirement that the temperature is bounded for all x and t., Taking Laplace transforms, we find, (1), where, , (2), , and U = U(x, s) is required to be bounded., Solving (1), we find, Then we choose cl = 0 so that u is bounded as « -» «, and we have, (*), , From (2) we have c2 = tt</s, so that, Hence by Problems 14.9 and 14.10 we find, , 14.16. Work Problem 15 if at t = 0 the temperature applied is given by g(t), t>0., The boundary-value problem in this case is the same as in the preceding problem except that, the boundary condition u(Q, t) - MO is replaced by w(0, t) = g(t). Then if the Laplace transform, of ff(t) is G(s), we find from (3) of Problem 14.15 that c2 = G(s) and so, , Now by Problem 14.11,, Hence by the convolution theorem,, , on letting v — x2/iicu.
Page 346 :
336, , COMPLEX INVERSION FORMULA FOR LAPLACE TRANSFORMS, , [CHAP. 14, , 14.17. A tightly stretched flexible string has its ends fixed at x = 0 and x = I. At time, t = 0 the string is given a shape defined by f(x) = nx(l — x), where ju. is a constant,, and then released. Find the displacement of any point x of the string at any time, t>0., The boundary-value problem is, , Taking Laplace transforms, we find, if Y(x, s) = „£ {y(x, t)},, , o r, , (1), , where, , (2), , The general solution of (1) is, (S), Then from conditions (2) we find, (4), so that (S) becomes, By using residues we find, , or, , 14.18. A semi-infinite beam which is initially at rest on the x axis is at time t — 0 given a, transverse displacement h at its end x = 0. Determine the transverse displacement, y(x, t) at any position x > 0 and at any time t > 0., The boundary-value problem is, , W, (*), , Taking Laplace transforms, we find, Y(0,s) - his,, , Yxx(0,s) = 0,, , Y(x, s) is bounded, , The general solution of the differential equation is, From the boundedness condition we require ct = c2 = 0 so that, , (3)
Page 347 :
CHAP. 14], , COMPLEX INVERSION FORMULA FOR LAPLACE TRANSFORMS, , From the first and second boundary conditions in (3), we find c4 = 0 and c3 = h/s so that, , The inverse Laplace transform is, by the complex, inversion formula,, , To evaluate this we use the contour of Fig. 14-8 since, s = 0 is a branch point. Proceeding as in Problem 14.9,, we find, omitting the integrand for the sake of brevity,, that, , Along EH,, , and we, , Along KL,, , find, , Fi, , S-14'8, , and we find, , Along HJK, s = eeU and we find, , Then (4) becomes, , Letting u/2b = v2, this can be written, , The result can also be written in terms of Fresnel integrals as, , 14.19. An infinitely long circular cylinder of unit, radius has a constant initial temperature Uo., At t — 0 a temperature of 0°C is applied to, the surface and is maintained. Find the temperature at any point of the cylinder at any, later time t., If (r, <f>, z) are cylindrical coordinates of any point, of the cylinder and the cylinder has its axis coincident with the z axis [see Fig. 14-9], it is clear that, the temperature is independent of <f> and z and can, thus be denoted by u(r, t). The boundary-value, problem is, (1), , u(l, t) = 0, u(r, 0) = u0, \u(r, t)\ < M, , (2), , Fig. 14-9, , 337
Page 348 :
338, , COMPLEX INVERSION FORMULA FOR LAPLACE TRANSFORMS, , [CHAP. 14, , It is convenient to consider instead of (1) the equation, , and then to replace t by xt., Taking Laplace transforms, we find, , £7(1, s) = 0,, , U(r, s) is bounded, , The general solution of this equation is given in terms of Bessel functions as, , Since Y0(i\fsr) is unbounded as r-» 0, we must choose c2 = 0. Then, , From 17(1, s) = 0, we find, , Thus, By the inversion formula,, , Now J0(i\fs) has simple zeros where i^/s = X1( X2, ... \n, .... Thus the integrand has simple poles, at s = —x|, re = 1,2,3, ... and also at s = 0. Furthermore it can be shown that the integrand, satisfies the conditions of Problem 14.2 so that the method of residues can be used., We have:, Residue of integrand at s = 0 is, , Residue of integrand at s = —X^ is, , where we have used L'Hospital's rule in evaluating the limit and also the fact that J'0(u) = —J^u)., Then, , Replacing t by «*, we obtain the required solution
Page 349 :
CHAP. 14], , COMPLEX INVERSION FORMULA FOR LAPLACE TRANSFORMS, , 339, , Supplementary Problems, THE COMPLEX INVERSION FORMULA AND USE OF RESIDUE THEOREM, 14.20. Use the complex inversion formula to evaluate, , (a), 14.21., , Find the inverse Laplace transform of each of the following using the complex inversion formula:, , 14.22. (a) Show that F(s) = -=, JL~i{F(s)}., *, 14.23. Evaluate, , p , 0, 3s + 2, , satisfies the conditions of the inversion formula., , (b) check your answer., 14.25. (a) Evaluate, , V, ^<-__ ,_^^, , justifying all steps., , 14.24. (a) E v a l u a t e j u s t i f y i n g all steps and, , ds around the contour C, , (6) Find, , 9, , /, /, T, , N*, , D, , \v, , _\^, , . y ., , x_, , \^, , shown in Fig. 14-10 where R is 3 and y > 1., (6) Give an interpretation of your answer as far as, Laplace transform theory is concerned., , ^""""~ ~"^, Fig. 14-10, , 14.26. Use the inversion formula to e v a l u a t e w h e r e a and b are any positive, constants., 14.27. Use the inversion formula to work: (a) Problem 4.39, page 113,, (c) Problem 4.41, page 114, (d) Problem 4.78(c), page 118., , (6) Problem 4.40, page 113,, , 14.28. Complete the proofs of (a) Case 3 and (6) Case 4 of Problem 14.2., INVERSE LAPLACE TRANSFORMS OF FUNCTIONS WITH BRANCH POINTS, 14.29. Find, 14.30. Find ., 14.31.Showthatbyusingtheinversionformula., 14.32. Find, 14.33., , using the complex inversion formula., by the inversion formula., by using the inversion formula, by using the complex inversion formula., , (a) Use the complex inversion formula to evaluate J^*1 {s~1/3} and (6) check your result by another, method., , 14.34. E v a l u a t e b y using t h e inversion formula., 14.35. Evaluate, , by the inversion formula., , INVERSE LAPLACE TRANSFORMS OF FUNCTIONS WITH INFINITELY MANY SINGULARITIES, 14.36. Find, 14.37. Prove that, 14.38. Find, , asing the complex inversion formula.
Page 350 :
340, , COMPLEX INVERSION FORMULA FOR LAPLACE TRANSFORMS, , [CHAP. 14, , 14.39. By using the complex inversion formula, prove that, , 14.40. Show that, 14.41. Show that for, , APPLICATIONS TO BOUNDARY-VALUE PROBLEMS, , 14.42. A semi-infinite solid « S 0 is initially at temperature zero. At t = 0 the face x = 0 is suddenly, raised to a constant temperature u0 and kept at this temperature for a time t0, after which the, temperature is immediately reduced to zero. Show that after an additional time t0 has elapsed,, the temperature is a maximum at a distance given by * = 2V«*o m 2 where K is the diffusivity,, assumed constant., 14.43. A semi-infinite solid x > 0 has its initial temperature equal to zero. A constant heat flux A is, applied at the face * = 0 so that —Kux(Q,t) = A. Show that the temperature at the face after, time t is, 14.44. Find the temperature at any point x > 0 of the solid in Problem 14.43., 14.45. A solid 0 S x S I is insulated at both ends x = 0 and x = I. If the initial temperature is equal, to ax(l — x) where a is a constant, find the temperature at any point x and at any time t., 14.46. A tightly stretched flexible string has its ends fixed at * = 0 and x = I. At t = 0 its midpoint, is displaced a distance h and released. Find the resulting displacement at any time t > 0., 14.47. Solve the boundary-value problem, , and interpret physically., 14.48. At t = 0, a semi-infinite solid x > 0 which is at temperature zero has a sinusoidal heat flux, applied to the face x = 0 so that —Kux(Q, t) = A + B sin«t, t > 0. Show that the temperature, of the face at any later time is given by, , 14.49. A solid, 0 < x < I, is initially at constant temperature u0 while the ends * = 0 and x = I are, maintained at temperature zero. Show that the temperature at any position * at any time t is, given by, , 14.50. A beam has its ends hinged at x = 0 and x = I. At time t = 0, a concentrated transverse load, of magnitude w is suddenly applied at the midpoint. Show that the resulting transverse displacement of any point x of the beam at any time t > 0 is, , if 0 < x < 1/2, while the corresponding result for 1/2 < x < I is obtained by symmetry.
Page 351 :
CHAP. 14], , COMPLEX INVERSION FORMULA FOR LAPLACE TRANSFORMS, , 341, , 14.51. An infinite circular cylinder of unit radius has its initial temperature zero. A constant flux A is, applied to the convex surface. Show that the temperature at points distant r from the axis at any, time t is given by, , where .Xn are the positive roots of «70 (X) = 0., 14.52. A cylinder of unit radius and height has its circular ends maintained at temperature zero while its, convex surface is maintained at constant temperature u0. Assuming that the cylinder has its axis, coincident with the z axis, show that the steady-state temperature at any distance r from the axis, and z from one end is, , Answers to Supplementary Problems, 14.20., 14.21., 14.22., 14.23., 14.34., 14.35., , 14.38., , 14.44., , 14.45., , 14.46., , 14.47.
Page 352 :
Chapter 15, , <M*ti|*i;, DEFINITION OF A MATRIX, , A matrix of order m x n, or m by n matrix, is a rectangular array of numbers having, m rows and n columns. It can be written in the form, (1), , Each number ajk in this matrix is called an element. The subscripts j and k indicate, respectively the row and column of the matrix in which the element appears., We shall often denote a matrix by a letter, such as A in (1), or by the symbol (a^) which, shows a representative element., A matrix having only one row is called a row matrix [or row vector] while a matrix, having only one column is called a column matrix [or column vector]. If the number of, rows m and columns n are equal the matrix is called a square matrix of order n x n or, briefly n. A matrix is said to be a real matrix or complex matrix according as its elements, are real or complex numbers., SOME SPECIAL DEFINITIONS AND OPERATIONS INVOLVING MATRICES, , 1. Equality of Matrices. Two matrices A = (ajk) and B = (bik) of the same order [i.e., equal numbers of rows and columns] are equal if and only if aik = bik., 2. Addition of Matrices. If A = (aik) and B = (bjk) have the same order we define the, sum of A and B as A + B = (aik + bjk)., Example 1. If, , then, , Note that the commutative and associative laws for addition are satisfied by, matrices, i.e. for any matrices A, B, C of the same order, A + B = B + A,, A + (B + C) = (A +B) + C, (2), 3. Subtraction of Matrices. If A = (a}k), B = (bik) have the same order, we define the, difference of A and B as A - B = (ajk - bik)., Example 2. If A and B are the matrices of Example 1, then, , 342
Page 353 :
CHAP. 15], , MATRICES, , 343, , 4. Multiplication of a Matrix by a Number. If A - (aik) and X is any number [or scalar],, we define the product of A by A. as XA = A\ = (Attjic)., Example 3. If A is the matrix of Example 1 and X = 4, then, , 5. Multiplication of Matrices. If A = (ajfc) is an mxn matrix while B = (&#) is an, nxp matrix, then we define the product A-B or AB of A and 5 as the matrix, C = (Cjk) where, (3), , and where C is of order mxp., Note that matrix multiplication is defined if and only if the number of columns of, A is the same as the number of rows of B. Such matrices are sometimes called, conformable., Example 4. Let, , Then, , Note that in general AB ¥* BA, i.e. the commutative law for multiplication of, matrices is not satisfied in general. However, the associative and distributive laws, are satisfied, i.e., A(BC) = (AB)C,, , A(B + C) = AB + AC,, , (B + C)A = BA + CA, , (4), , A matrix A can be multiplied by itself if and only if it is a square matrix. The, product A - A can in such case be written A2. Similarly we define powers of a square, matrix, i.e. A3 = A'A2, A4 = A«A 3 , etc., 6. Transpose of a Matrix. If we interchange rows and columns of a matrix A, the, resulting matrix is called the transpose of A and is denoted by AT. In symbols, if, A = (ajfc) then AT = (akj)., Example 5. The transpose of, , is, , We can prove that, (5), , 1. Symmetric and Skew-Symmetric Matrices. A square matrix A is called symmetric if, AT — A and skew-symmetric if AT = —A., Example 6. The matrix, , is symmetric while, , is skew-symmetric., , Any real square matrix [i.e. one having only real elements] can always be expressed, as the sum of a real symmetric matrix and a real skew-symmetric matrix.
Page 354 :
344, , MATRICES, , [CHAP. 15, , 8. Complex Conjugate of a Matrix. If all elements ajk of a matrix A are replaced by, their complex conjugates aik, the matrix obtained is called the complex conjugate of A, and is denoted by A., ., 9. Hermitian and Skew-Hermitian Matrices. A square matrix A which is the same as, the complex conjugate of its transpose, i.e. if A = AT, is called Hermitian. If, A = —AT, then A is called skew-Hermitian. If A is real these reduce to symmetric, and skew-symmetric matrices respectively., 10. Principal Diagonal and Trace of a Matrix. If A = (aik) is a square matrix, then the, diagonal which contains all elements ajk for which j = k is called the principal or, main diagonal and the sum of all such elements is called the trace of A,, Example 7. The principal or main diagonal of the matrix, , is indicated by the shading, and the trace of the matrix is 5 + 1 + 2 = 8., A matrix for which ajk = 0 when j ¥= k is called a diagonal matrix., , 11. Unit Matrix. A square matrix in which all elements of the principal diagonal are, equal to 1 while all other elements are zero is called the unit matrix and is denoted by, /. An important property of 7 is that, AI •= IA = A,, , I* = I,, , % = 1,2,3, ..., , (6), , The unit matrix plays a role in matrix algebra similar to that played by the, number one in ordinary algebra., 12. Zero or Null Matrix. A matrix whose elements are all equal to zero is called the, null or zero matrix and is often denoted by O or simply 0. For any matrix A having, the same order as 0 we have, A + 0 = 0 + A=A, (7), Also if A and 0 are square matrices, then, , AO = OA = 0, , (8), , The zero matrix plays a role in matrix algebra similar to that played by the number, zero of ordinary algebra., DETERMINANTS, If the matrix A in (1) is a square matrix, then we associate with A a number denoted by, , W, , called the determinant of A of order n, written det (A). In order to define the value of a, determinant, we introduce the following concepts., 1. Minor. Given any element ajk of A we associate a new determinant of order (n — 1), obtained by removing all elements of the jth row and Mh column called the minor of ajk.
Page 355 :
CHAP. 15], , MATRICES, , 345, , Example 8. The minor corresponding to the element 5 in the 2nd row and 3rd column of the, fourth order determinant, , which is obtained by removing the elements shown shaded., , 2. Cofactor. If we multiply the minor of ajfc by (-l)'+k, the result is called the coj'actor, of cijk and is denoted by A}k., Example 9. The cofactor corresponding to the element 5 in the determinant of Example 8 is, (—l) 2+s times its minor, or, , The value of a determinant is then defined as the sum of the products of the elements, in any row [or column] by their corresponding cofactors and is called the Laplace expansion., In symbols,, (10), We can show that this value is independent of the row [or column] used [see Problem 15.7],, THEOREMS ON DETERMINANTS, , Theorem 15-L The value of a determinant remains the same if rows and columns are, interchanged. In symbols, det (A) = det (AT)., Theorem 15-2. If all elements of any row [or column] are zero except for one element, then, the value of the determinant is equal to the product of that element by its, cofactor. In particular, if all elements of a row [or column] are zero the, determinant is zero., Theorem 15-3. An interchange of any two rows [or columns] changes the sign of the, determinant., Theorem 15-4. If all elements in any row [or column] are multiplied by a number, the, determinant is also multiplied by this number., Theorem 15-5. If any two rows [or columns] are the same or proportional, the determinant, is zero., Theorem 15-6. If we express the elements of each row [or column] as the sum of two terms,, then the determinant can be expressed as the sum of two determinants, having the same order., Theorem 15-7. If we multiply the elements of any row [or column] by a given number and, add to corresponding elements of any other row [or column], then the value, of the determinant remains the same., Theorem 15-8. If A and B are square matrices of the same order, then, det (AS) = det (A) det (B), , (11)
Page 356 :
346, , MATRICES, , [CHAP. 15, , Theorem 15-9. The sum of the products of the elements of any row [or column] by the, cofactors of another row [or column] is zero. In symbols,, , (12), If p = q, the sum is det (A) by (10)., Theorem 15-10. Let v\, v2,..., vn represent row vectors [or column vectors] of a square, matrix A of order n. Then det (A) - 0 if and only if there exist constants, [scalars] Ai, A2, ..., \n not all zero such that, , (13), where O is the null or zero row matrix. If condition (13) is satisfied we, say that the vectors v\, Vz,..., vn are linearly dependent. Otherwise they, are linearly independent. A matrix A such that det (A) = 0 is called a, singular matrix. If det (A) ¥° 0, then A is a non-singular matrix., In practice we evaluate a determinant of order n by using Theorem 15-7 successively, to replace all but one of the elements in a row or column by zeros and then using Theorem, 15-2 to obtain a new determinant of order » —1. We continue in this manner, arriving, ultimately at determinants of orders 2 or 3 which are easily evaluated., INVERSE OF A MATRIX, If for a given square matrix A there exists a matrix B such that AB — I, then B is, called an inverse of A and is denoted by A"1. The following theorem is fundamental., Theorem 15-11. If A is a non-singular square matrix of order n [i.e. det(A)^0], then, there exists a unique inverse A"1 such that AA"1 = A"1 A = 7 and we, can express A"1 in the following form, , (U), where (Ajk) is the matrix of cofactors ASk and, transpose., , (Ajfc)T = (Aki), , is its, , The following express some properties of the inverse:, (15), , ORTHOGONAL AND UNITARY MATRICES, A real matrix A is called an orthogonal matrix if its transpose is the same as its, inverse, i.e. if AT = A"1 or ATA = 7., A complex matrix A is called a unitary matrix if its complex conjugate transpose is, the same as its inverse, i.e. if A T = A"1 or ATA = 7. It should be noted that a real, unitary matrix is an orthogonal matrix., ORTHOGONAL VECTORS, In Chapter 5 we found that the scalar or dot product of two vectors ad + a2j + ask and, &ii + &2J + &sk is ai&i +tt2&2+ as&s and that the vectors are perpendicular or orthogonal if, «i&i + oz&2 + a3&s = 0. From the point of view of matrices we can consider these vectors, as column vectors
Page 357 :
CHAP. 15], , MATRICES, , 347, , from which it follows that, This leads us to define the scalar product of real column vectors A and B as ATB and to, define A and B to be orthogonal if ATB = 0., It is convenient to generalize this to cases where the vectors can have complex components and we adopt the following definition:, Definition 1. Two column vectors A and B are called orthogonal if ATB = 0, and ATB, is called the scalar product of A and B., It should be noted also that if A is a unitary matrix then ATA = 1, which means that the, scalar product of A with itself is 1 or equivalently A is a unit vector, i.e. having length 1., Thus a unitary column vector is a unit vector. Because of these remarks we have the, following, Definition 2. A set of vectors Xi, X2, ... for which, , is called a unitary set or system of vectors or, in the case where the vectors, are real, an orthonormal set or an orthogonal set of unit vectors., SYSTEMS OF LINEAR EQUATIONS, A set of equations having the form, (16), , is called a system of m linear equations in the n unknowns x\,xz,...,xn. If r\,r2,...,rn, are all zero the system is called homogeneous. If they are not all zero it is called nonhomogeneous. Any set of numbers x\,xz, ...,%«. which satisfies (16) is called a solution of, the system., In matrix form (16) can be written, (17), , or more briefly, , (18), , where A,X,R represent the corresponding matrices in (17)., SYSTEMS OF n EQUATIONS IN n UNKNOWNS. CRAMER'S RULE, If m = n and if A is a non-singular matrix so that A"1 exists, we can solve (17) or, (18) by writing, and the system has a unique solution., , (iy)
Page 358 :
348, , MATRICES, , [CHAP. 15, , Alternatively we can express the unknowns x\,xz,.. .,xn as, (20), , where, called the determinant of the system, is given by (9) and Afc,, k = 1,2, ..., n is the determinant obtained from A by removing the fcth column and, replacing it by the column vector R. The rule expressed in (20) is called Cramer's rule., The following four cases can arise., Case 1,, In this case there will be a unique solution where not all x* will, be zero., Case 2,, In this case the only solution will be x\ = 0, x% = 0, ..., xn = 0,, i.e. X = 0. This is often called the trivial solution., Case 3, A = 0, R = 0. In this case there will be infinitely many solutions other than, the trivial solution. This means that at least one of the equations can be obtained, from the others, i.e. the equations are linearly dependent., Case 4, A = 0, R ¥* 0. In this case infinitely many solutions will exist if and only if all, of the determinants Afe in (20) are zero. Otherwise there will be no solution., The cases where m ¥* n are considered in Problems 15.93-15.96., EIGENVALUES AND EIGENVECTORS, Let A = (Ojk) be an n x n matrix and X a column vector. The equation, AX = AX, , (21), , where A is a number can be written as, (22), , or, (23), , The equation (23) will have non-trivial solutions if and only if, (24), , which is a polynomial equation of degree n in A. The roots of this polynomial equation are, called eigenvalues or characteristic values of the matrix A. Corresponding to each eigenvalue there will be a solution X ¥* 0, i.e. a non-trivial solution, which is called an eigenvector or characteristic vector belonging to the eigenvalue. The equation (24) can also, be written, det(A-A7) = 0, , and the equation in A is often called the characteristic equation., , (25)
Page 359 :
CHAP. 15], , MATRICES, , 349, , THEOREMS ON EIGENVALUES AND EIGENVECTORS, Theorem 15-12. The eigenvalues of a Hermitian matrix [or symmetric real matrix] are, real. The eigenvalues of a skew-Hermitian matrix [or skew-symmetric, real matrix] are zero or pure imaginary. The eigenvalues of a unitary, [or real orthogonal matrix] all have absolute value equal to 1., Theorem 15-13. The eigenvectors belonging to different eigenvalues of a Hermitian matrix, [or symmetric real matrix] are orthogonal., Theorem 15-14 [Cayley-Hamilton]. A matrix satisfies its own characteristic equation, [see Problem 15.40]., Theorem 15-15 [Reduction of matrix to diagonal form]. If a non-singular matrix A has, distinct eigenvalues Xi, A2, A.S, ... with corresponding eigenvectors written, as columns in the matrix, , then, i.e. B~1AB, called the transform of A by B, is a diagonal matrix containing, the eigenvalues of A in the main diagonal and zeros elsewhere. We say, that A has been transformed or reduced to diagonal form. See Problem, 15.41., Theorem 15-16 [Reduction of quadratic form to canonical form]., Let A be a symmetric real matrix, for example,, , Then if x, , Then if , we obtain the quadratic form, , The cross product terms of this quadratic form can be removed by letting X = BU, where U is the column vector with elements u\, u2, u3 and B is an orthogonal matrix which, diagonalizes A. The new quadratic form in u\, Uz, u3 with no cross product terms is called, the canonical form. See Problem 15.43. A generalization can be made to Hermitian quadratic forms [see Problem 15.114]., OPERATOR INTERPRETATION OF MATRICES, If A is an n x n matrix, we can think of it as an operator or transformation acting on a, column vector X to produce AX which is another column vector. With this interpretation, equation (21) asks for those vectors X which are transformed by A into constant multiples, of themselves [or equivalently into vectors which have the same direction but possibly, different magnitude].
Page 360 :
MATRICES, , 350, , [CHAP. 15, , If case A is an orthogonal matrix, the transformation is a rotation and explains why, the absolute value of all the eigenvalues in such case are equal to one [Theorem 15-12], since, an ordinary rotation of a vector would not change its magnitude., The ideas of transformation are very convenient in giving interpretations to many, properties of matrices., , Solved Problems, OPERATIONS WITH MATRICES, , 15.1. If, , find, , (a) A + B, (b) A-B,, , (c) 2A - 3C, (d) 3A + 2B- 4(7, (e) AB, (/) BA, (g) (AB)C, (h) A(BC), (i) AT + BT,, (j) BTAT., , (a), (6), <«), , (d), , ('), (/), Note that AB ¥= BA using (e), illustrating the fact that the commutative law for products, does not hold in general., (9), (K), , Note that (AB)C = A(BC), products holds., , using (g), illustrating the fact that the associative law for, , (i), , Note that AT + B* = (A + B)? using (a)., 0), , Note that BTAf = (AB)? using (e).
Page 361 :
CHAP. 15], , MATRICES, , 15.2. If A =, , show that, , B =, (A+B)2, , 351, , = A2 + AB + BA + B2, , We have, , T h e n ( A + B ), , 2, , = (A+B)(A+B), , =, , Now, , Thus, , A* + AB + BA + B2 =, , \ =, , (A + B)2, , 153. Prove that any real square matrix can always be expressed as the sum of a real, symmetric matrix and a real skew-symmetric matrix., If A is any real square matrix, then, , A = $(A+AT) + %(A-AT), But since (A+A?)? - AT + A = A + AT, it follows that £(A+AT) is symmetric. Also, since, (A-AT)i = AT -A = -(A-AT), it follows that %(A-AT) is skew-symmetric. The required, result is thus proved., , 15.4. Show that the matrix, We have, , is Hermitian., Thus A is Hermitian.
Page 362 :
MATRICES, , 352, , [CHAP. 15, , 15.5. Prove that a unit matrix 7 of order n commutes with any square matrix A of order, n and the resulting product is A., We illustrate the proof for n = 3. In such case, , Then, , i.e. IA = AI = A., Extensions are easily made for n > 3., , DETERMINANTS, , 15.6. Use the definition of a determinant [Laplace expansion] as given on page 345 to, evaluate a determinant of (a) order 2, (&) order 3., (a) L e t t h e determinant b, cofactors are, , «, , U, , s, , e t h e elements o f t h e f i r s t row., , T h e corresponding, , Then by the Laplace expansion the determinant has the value, The same value is obtained by using the elements of the second row [or first and second, columns]., (6) Let the determinant be, , The cofactors of the elements in the first row are, , Then the value of the determinant is, , The same value is obtained by using elements of the second or third rows [or first, second and, third columns].
Page 363 :
CHAP. 15], , MATRICES, , 353, , 15.7. Prove that the value of a determinant remains the same regardless of which row [or, column] is taken for the Laplace expansion., Consider the determinant A = (ajk) of order n. The result is true for n = 2 by Problem 15.6., We use proof by induction, i.e. assuming it to be true for order n — 1 we shall prove it true for, order n. The plan will be to expand A using two different rows p and q and show that the expansions are the same., Let us first expand A by elements in the pth row. Then a typical term in the expansion is, (1), , where Mpk is the minor corresponding to the cofactor Apk of Oj,k., n — 1, any row can be used in its expansion., , Since this minor is of order, , We shall use the qth row where we assume that q > p since a similar argument holds if, q < p. This row consists of elements aqr where r ¥= k and corresponds to the (q — l)st row of Mpk., Now if r < k, aqr is located in the rth column of Mpk so that in the expansion the term corresponding to aqr is, (2), where Mpkqr is the minor corresponding to the element aqr in Mpk. From (1) and (2) it follows that, a typical term in the expansion of A is, (»), , If r > k then aqr is located in the (r — l)st column and so there is an additional minus sign in (3)., If we now expand A by elements in the qih row, a typical term is, (4), , We can expand Mqr by elements in the pth row where p > q. As before if k> r, a typical term, in the expansion of Mqr is, (5), From (4) and (5) we see that a typical term in the expansion of A is, («), , which is the same as (3). If k < r an additional minus sign appears in (6), agreeing with the, case corresponding to r > k using the first expansion. Thus the required result is proved., In a similar manner we can prove that expansion by columns is the same and gives the same, result as the expansion by rows [Theorem 15-1, page 345]., , 15.8. Evaluate by the Laplace expansion the determinant, in the first row and (&) using elements in the second row., (a) Using elements in the first row, the expansion is, , (6) Using elements in the second row, the expansion is, , (a) using elements
Page 364 :
354, , MATRICES, , [CHAP. 15, , 15.9. Prove Theorem 15-4, page 345., Let the determinant be, , a), and suppose that the elements in the fcth row are multiplied by X to give the determinant, , (2), , Expanding (1) and (2) according to elements in the fcth row, we find respectively, (S), (4), , from which ., , as required., , 15.10. Prove Theorem 15-5, page 345., (a) If two rows have the same elements, then the value of the determinant will not change if the, rows are interchanged. However, according to Theorem 15-3, page 345, the sign must change., Thus we have A = —A or A = 0., (6) If the two rows have proportional elements, then they can be made the same by factoring out, the proportionality constants and thus the determinant must be zero by (a)., , 15.11. Prove Theorem 15-6, page 345., Write the determinant as, , in which the first row has each element expressed as the sum of two terms. Then by the Laplace, expansion we have, W, , where An, A12,.. .,Aln are the cofactors of the corresponding elements in the first row. But (1), can be written as, , as required. A similar procedure proves the result if any other row [or column] is chosen.
Page 365 :
CHAP. 15], , MATRICES, , 355, , 15.12. Prove Theorem 15-7, page 345., Suppose we multiply the elements of the second row of A = (ajk) by X and add to the elements, of the first row [a similar proof can be used for any other rows or columns]. Then the determinant, can be written as, , But by Problem 15.11 this can be written as, , Then the required result follows since the second determinant is zero because the elements of its, first and second rows are proportional [Theorem 15-5]., , 15.13. Evaluate, , Multiplying the elements of the first row by —3,2,3 and adding to the elements of the second,, third and fourth rows respectively, we find, , which by Theorem 15-7 has a value equal to that of the given determinant. Note that this new, determinant has three zeros in the 2nd column, which was precisely our intention in choosing the, numbers —3,2,3 in the first place., Multiplying each element in the second column by its cofactor, we see that the value of the, determinant is, , on removing the factor 5 from the second row, using Theorem 15-4., Now multiplying the elements in the second row by 5 and —1 and adding to the elements of, the first and third rows respectively, we find, , which on expanding by the elements in the second column gives
Page 366 :
MATRICES, , 356, , [CHAP. 16, , 15.14. Verify Theorem 15-8 if, The theorem states that det (AB) = det (A) det (B)., , Then since, , it states that, or, , (7)(84) =, , (17)(14), , But since this is correct, the theorem is verified for this case., , 15.15. Let vi - (2-1 3), vz = (1 2-1), va = (-3 4-7). (a) Show that Vi,v2fva are, linearly dependent. (6) Illustrate Theorem 15-10, page 346, by showing that, , (a) We must show that there exist constants XL X2, X3 not all zero such that \lvl + X2v2 + X3v8 =, 0 = (0 0 0). Now, when, , Assuming that X3 = 1, for example, the equations become 2Xj + X2 = 3, Xj — 2X2 = 4,, 3Xj — X2 = 7. Solving any two of these simultaneously, we find Xt = 2, X2 = —1. Thus, Xt = 2, X2 = —1, X3 = 1 provide the required constants., (6) Multiplying the elements of the second row by —2,3 and adding to the first and third rows, respectively, the given determinant equals, , 15.16. Prove Theorem 15-9, page 346., By definition the determinant, , when expanded according to the elements of the pth row has the value, (1), , Let us now replace the elements opfc in the pth row of A by corresponding elements aqk of the gth, row where p ¥° q. Then two rows will be identical and the new determinant thus obtained will be, zero by Theorem 15-5. Since apk — a,qk, (l) is replaced by, i.e., , (2)
Page 367 :
CHAP. 15], , MATRICES, , 357, , Similarly by using columns rather than rows we can show that, (»), , If p = q, then (2) and (3) become respectively, (4), (5), , INVERSE OF A MATRIX, 15.17. Prove that, We must show that AA~l — I, the unit matrix., , To do this consider the product, , Now by the rule for multiplying determinants [which is the same as that for multiplying matrices],, the element cpq in the resulting determinant is found by taking the sum of the products of elements, in the gth row of the first determinant and the pth column of the second determinant. We thus, have, But by the results of Problem 15.16,, , It follows that, , Then if det (A) ¥> 0, this can be written, , and it thus follows that AB = I where, , 15.18. (a) Find the inverse of the m, direct multiplication., , a, , t, , (a) The matrix of cofactors of A is given by, , r, , i, , x, , a, , n, , d (6) check the answer by
Page 368 :
358, , MATRICES, , [CHAP. 15, , The transpose of this matrix, , is, , Since det (A) = 35 [see Problem 15.8], we have, , (6), , We can also show that A~1A = /. This supplies the required check., , 15.19. Prove that, , (AB)~l = B^A'1., , Let X = (AB)-1. Then (AB)X = I where / is the unit matrix. By the associative law this, becomes A(BX) = I. Multiplying by A"1, we have A~1[A(BX)] = A~ll — A~l which again using, the associative law becomes (A~1A)(BX) = A"1 or I(BX) = A"1, i.e. BX - A'1. Multiplying by, B~l and using the associative law once more, we have B~1(BX) — B~1A~1> (B~1B)X = B~1A~1,, IX = B~iA-i, i.e. X = B-1A~1, as required., , 15.20. Prove that if A is a non-singular matrix, then det, Since A A ~ 1 = I, det (A A -') = det (/) = !. But by Theorem 15-8, det (A A ~») = det (A) det (A -')., Thus det (A-1) det (A) = 1 and the required result follows., , ORTHOGONAL AND UNITARY MATRICES. ORTHOGONAL VECTORS, 15.21. Show that, , is an orthogonal matrix., , We have, using the fact that A is real,, , since cos2 6 + sin2 e = 1., , Thus A is an orthogonal matrix., , 15.22. Show that, , is a unitary matrix., , Since A is complex, we must show that ATA = /., , so that A is a unitary matrix., , We have
Page 369 :
CHAP. 15], , MATRICES, , 359, , 15.23. If A is an orthogonal matrix, prove that det (A) = ±1., If A is orthogonal, then ATA = I so that by Theorem 15-8, page 345,, det (A^A) = det (AT) det (A) = det/ = 1, , (1), , T, , But det(A ) = det (A) so that (1) becomes, , [det (A)]* = 1, , or, , det (A) = ±1, , 15.24. Show that the vectors, , form an orthonormal set or system of vectors., Since the vectors are real, we must show that, , If j = k = 1, we, , have, , T, , T, , Similarly we find if j — k = 2 and j = k — 3, A 2 A 2 = 1, A 3 A 3 = 1. Thus A 1( A 2 ,A s are unit, vectors., To show the orthogonality of any two of the vectors consider, for example, j = 1, k = 2., Then we have, , T, , T, , Similarly A 1 A 3 = 0, A 2 A 3 = 0 and so the vectors are mutually orthogonal., form an orthonormal system., , Thus the vectors, , SYSTEMS OF LINEAR EQUATIONS, 15.25. Prove Cramer's rule (20), page 348, for solving the system of equations (16), page, 347, in the case where m = n., The system of equations can be written, , Multiplying by A kp and adding from k = 1 to «, we have, , and, , This can be written, Now by equations (3) and (5) of Problem 15.16, we have, , (i)
Page 370 :
360, , MATRICES, , Thus (1) becomes, , [CHAP. 15, , det, , so that if A = det (A),, (2), Now the numerator of (2) is a determinant in which the pth column is replaced by the column, vector (TI r2 ... rn)T, and so Cramer's rule follows., , 15.26. Work Problem 15.25 by using the inverse matrix., As on page 347, we solve the system (17) or (18) in the form (19), i.e., , X = A~1R, Now, , Thus we have, , from which it follows that, , agreeing with (2) of Problem 15.25., , 15.27. Solve the system of equations, , (a) by Cramer's rule and (6) by using inverse matrices., (a) By Cramer's rule,, , where the determinant of the coefficients is, , See Problem 15.8., xs = —1., , Evaluation of the other determinants yields the solution xl = 2, x2 = —3,
Page 371 :
CHAP. 15], , MATRICES, , 361, , (6) The system can be written in matrix form as, CO, , Now the inverse of the first matrix A in (1) was found in Problem 15.18, so that multiplying, both sides of (1) by this matrix we have on using the fact that A~1A = I,, , Thus »i - 2, x2 = —3, xs — —1., Geometrically the equations with xt = x, xz = y, xs = « represent three planes intersecting in the point (2, —3, —1)., , 15.28. Solve, Cramer's rule gives, , where, Evaluation of the determinants gives formally, , U), illustrating the fact that the system has no solution., On multiplying the first of the given equations by 2, the second by 3, and adding, we obtain, 7*1 + 4a;2 — Sxa = 12 which is not consistent with the third equation given, i.e. 7xl + 4#2 — Sx3 — —4., Thus the system of equations is inconsistent., Geometrically the first two equations represent two planes which intersect in a line. The, third equation represents a plane which is parallel to this line. Theoretically the planes meet at, a point at infinity, which is a possible interpretation of (1)., , 15.29. Solve, In this case a formal application of Cramer's rule gives, , since theoretically 0/0 can represent any number, our result illustrates the fact that the system, has infinitely many solutions.
Page 372 :
362, , MATRICES, , [CHAP. 15, , On multiplying the first equation by 2, the second by 3 and adding, we obtain the third, equation. Thus the third equation can be obtained from the first two and so is not needed. We, call the system of equations dependent or more precisely linearly dependent., Geometrically the planes represented by the first two equations intersect in a line. The plane, represented by the third equation passes through the line., To obtain possible solutions, assign different values to x3 for example. Thus if xs = 1, then, and we have a point on the line whose coordinates are, Other solutions, can be obtained similarly., , 15.30. Solve, Cramer's rule gives the solution [see Problem 15.8], , so that the only solution is the trivial solution., Geometrically the equations represent three planes which intersect in the point (0,0,0)., , 15.31. Solve, Formal application of Cramer's rule gives, , illustrating the fact that there are infinitely many solutions besides the trivial and obvious one, xl = 0, x2 — 0, #3 = 0. Such solutions can be found by assigning different values to x3 as in, Problem 15.29. Note that the third equation is obtained by adding twice the first equation to, three times the second equation, so the equations are dependent., , 15.32. For what values of k will the system, , have non-trivial solutions?, Solving formally by Cramer's rule, we would have, , Then if A ¥= 0, the system would have the trivial solution x — 0, y = 0, z = 0. In order that, the system have non-trivial solutions, we must then have A = 0, i.e., or, , Solving, we find that k - 1, 9/4.
Page 373 :
CHAP. 15], , MATRICES, , 363, , EIGENVALUES AND EIGENVECTORS, 15.33. Find the eigenvalues of the matrix, Method 1., If, , we must consider the equation AX = \X, i.e., , or, , Equating corresponding elements of these matrices, we find, , w, This system will have non-trivial solutions if, , (2), Expansion of this determinant yields, , X« - 6X2 + 11X - 6 = o, , or, , (X - 1)(\ - 2)(X - 3) = 0, , Then the eigenvalues are X = 1,2,3., Method 2., We can write AX — \X, zero matrix and, , as AX = \IX, , or (A — \I)X = 0 where / and 0 are the unit and, , Then non-trivial solutions will exist if det (A — X/) = 0 and we can then proceed as in method 1., Note that equation (2) can be written at once by subtracting X from each of the diagonal elements, of A., , 15.34. (a) Find eigenvectors corresponding to the eigenvalues of the matrix A in Problem, 15.33 and (6) determine a set of unit eigenvectors., (a) Corresponding to X = 1, equations (1) of Problem 15.33 become, , Solving for x1 and xs in terms of x2 we find xs — 3x2, xl = 2x2. Then an eigenvector is, , since any eigenvector is a scalar (constant) multiple of this.
Page 374 :
364, , MATRICES, , [CHAP. 15, , Similarly, corresponding to X = 2 equations (1) of Problem 15.33 lead to, Xi — sc2 which in turn leads to the eigenvector, , x3 = 2x2,, , Finally if X = 3 we obtain x3 = x2, xl = —xz, giving the eigenvector, , (b) The unit eigenvectors have the property that they have length 1, i.e. the sum of the squares, of their components = 1. To obtain such eigenvectors we divide each vector by the square, root of the sum of the squares of the components. Thus the above become respectively, , 15.35. Find the (a) eigenvalues and (&) eigenvectors of, , (a) The eigenvalues are solutions of, , which gives X = 1,4,6., , (6) From the equations (A — \I)X = 0 we obtain, , Then corresponding to X = 1 we find the eigenvector, , Corresponding to X = 4 we find the eigenvector, , Corresponding to X = 6 we find the eigenvector, , 15.36. Find the (a) eigenvalues and (6) eigenvectors of A =, (a) By the usual procedure the eigenvalues are solutions of, or, , Then
Page 375 :
CHAP. 15], , MATRICES, , 365, , (6) The equations for determining the eigenvectors are found from, , i.e., , (-0, , Using X = e'e = cos e + i sin e we find from (1), x2 = —ite, so that a corresponding eigenvector is, or simply, Using X = e~w = cose — i sin $ we find xz = ixt so that a corresponding eigenvector is, or simply, , THEOREMS ON EIGENVALUES AND EIGENVECTORS, 15.37. Prove that the eigenvalues of a Hermitian matrix [or symmetric real matrix] are real., Let A be a Hermitian matrix and X an eigenvalue., eigenvector X such that, , Then by definition there is a non-trivial, , Multiplying by XT,(1), , (1), , Taking the conjugate,, , (#), , Taking the transpose, using the second and third equations in (5), page 343, we find, (3), , Now since A is Hermitian, AT = A so that (3) becomes, , (•*), Subtracting (4) from (1) we thus obtain, Then since XTX cannot be zero, it follows that X = X or that X must be real., , 15.38. Prove that the eigenvectors of a Hermitian matrix [or symmetric real matrix], belonging to different eigenvalues are orthogonal., Let Xj and X2 be eigenvectors belonging to eigenvalues X1( X2. Then denoting the matrix by, A, we have, (1), , Multiplying these by X* and X* respectively, we find, (2), Taking the conjugate of the first equation in (2), we find since X t is real,, (*), , Now taking the transpose of (3),, W
Page 376 :
MATRICES, , 366, , [CHAP. 15, , Since A is Hermitian, i.e. AT = A, (4) becomes, , Subtracting this from the second of equations (2),, , Then since Xj n4 \2, we have, , or Xl and X2 are orthogonal., , 15.39. (a) Illustrate by an example the results of Problems 15.37 and 15.38., (b) If a matrix has real eigenvalues, must it be Hermitian? Explain., (a) The matrix A of Problem 16.35 is real and symmetric and thus Hermitian., problem, the eigenvalues are all real. Also the eigenvectors, , As shown in that, , are mutually orthogonal as is easily verified., (6) A matrix can have real eigenvalues without being Hermitian., of Problem 15.33., , See, for example, the matrix, , 15.40. Verify the Cayley-Hamilton Theorem 15-14, page 349, for the matrix, The characteristic equation is, or, , X2 - 7X + 14 = 0, , To verify the theorem we must show that the matrix A satisfies, A2 - 7A + 147 = 0, where X in the characteristic equation is replaced by A, the constant term [in this case 14] is, replaced by 147 and 0 is replaced by 0., We have, , as required., , 15.41. Verify Theorem 15-15, page 349, by transforming the matrix of Problem 15.33 into, diagonal form., The eigenvectors for the matrix of Problem 15.33 are the columns of, , as shown in Problem 15.34. The inverse of B is then given by
Page 377 :
CHAP. 15], , MATRICES, , 367, , Thus, , 15.42. Prove Theorem 15-15., We prove the theorem for the case of a third order matrix, since the proof for any square, matrix is exactly analogous. Denote the eigenvectors of A by the columns in, , and the corresponding distinct eigenvalues by X lf X2, X$. Then by definition,, , from which, , Thus multiplying by B~l we have, as required,, , 15.43. (a) Show that the quadratic form 2x\ + 4x22 + 5x1 ~ 4%iXs = XTAX, , where, , (6) Find a linear transformation from x\, Xz, Xz to Ui,U2,us which will remove the, cross product term in the quadratic form of (a) and thus write the resulting, quadratic form in u\, Uz, Ua., (a) We have
Page 378 :
368, , MATRICES, , ICHAP. 15, , Note that the coefficients of x\,x\, *| namely 2, 4, 5 appear in the main diagonal while half, the coefficients of XjXk, j ¥* k appear as elements in the /th row and feth column., (6) A linear transformation from xlt X2, xa to ult U2,u3 can be written as, , X = BU, , where, , and B is a 3 X 3 matrix. Then we have, XTAX - (BU)TA(BU), , = UT(BTAB)U, , (1), , T, , Now the right side of (1) will not have cross product terms if B AB is a diagonal matrix. Thus, we see that if BT = B~l [i.e. if B is an orthogonal matrix] the problem becomes one of finding, the eigenvalues and eigenvectors of A. This has already been done in Problem 15.35. We choose, B as the matrix of unit eigenvectors, i.e., , from which we easily find that BT = B~l so that B is orthogonal and we have, , as required. Then (1) becomes, , which is the required quadratic form, called the canonical form., U is X = BU, from which we find, , The transformation from X to, , Supplementary Problems, OPERATIONS WITH MATRICES, 15.44. (a) If, , verify that A(B + C) = AB + AC,, T, , (A + B)(A - B) = A* - £2 + BA - AB, and (ABC), , = CTBTAT., , (b) Find 2A - SB - C and (A - 2B)(C + SB) where A,B, C are the matrices in (a)., 15 4, , - 5. If, (e)A(Bi+C),, , 15, , -46., , 15 47, , - -, , B = (4 -2),, , (d) ATB + B*A., , find, , (a) AC,, , (b) CA, (c) BC, (d) CTBT,, , (f)BBT., , If, , If, , C = (~1}, , (a) CTA, (b) A^C, (c) AA^BC., , find, , (a)(A-B)(A+B),, , (b)A*-B*, (c) AB - BA,
Page 379 :
CHAP. 15], , MATRICES, , 369, , 15.48. Prove that for any m X n matrices (a) A + B = B + A,, (c) X(A + B) = \A + \B where X is any scalar., , (b) A + (B+ C) = (A + B) + C,, , 15.49. Find x and y such that, 15.50. If A and B are square matrices such that AB = 0, prove that we can have A ¥* 0, B ¥= 0., the result true for non-square matrices?, , Is, , 15.51. If AB = AC, is it true that B = C? Explain., 15.52. If A,B and C are any square matrices of the same order, prove that (a) A(BC) = (AB)C,, (6) A(B + C)=AB + AC, (c) (ABC)T - CTBfAf, and generalize these results., 15.53. A linear transformation from an (xlt xz) to a (ylt j/2) coordinate system is denned as j/j = a-uXi +, a-12%2, 2/2 — &2ix\ + «22*2', , (a) ^, , transformation can be written, , Y = AX., , show that the, (b) If, , X = BU, , where, , show that xl = bnUi + bl2u2, x2 = &2iwi + &22M2of, , Thus obtain ylt j/2 in terms, , M I( «2 and explain how you can use the approach to motivate a definition of AB., , 15.54. Generalize the ideas of Problem 15.53 to 3 or more dimensions., 15.55. Let P [Fig. 15-1] have coordinates (xtf/) relative to an xy coordinate system and (x1, y') relative to, an x'y' coordinate system which is rotated through angle 8 relative to the xy coordinate system., (a) Prove that the relationship between the coordinates or transformation from (x,y) to (x',y'), is given by, , (6) Show that the square matrix in (a) is skew-symmetric., , Fig. 15-1, , Fig. 15-2, , 15.56. A vector with components (xt, yt) in an xy coordinate system [Fig. 15-2] is rotated through angle 6, so that its new components are (x2,y2). Show that, , and explain the relationship with Problem 15.55., 15.57. Let A(e) denote the square matrix in Problems 15.55 or 15.56. Show that (a) A(el + $2) =, A(e1)A(e2), (b) A(»l +9Z+••• + en) = A(e1)A(e2)-•-A(en), (c) [A(«)]» = A(ne) and discuss the, significance of these results in terms of the transformations in Problems 15.55 and 15.56.
Page 380 :
370, , MATRICES, , 15.58. Let, , [CHAP. 15, , (a) Show that, , which, , is called a quadratic form in xlt ...,xn. (b) Show that if A is a real symmetric matrix, i.e., °jfc = akj> then XTAX = an»\ + o^l + • • • + «„„«£ + 2ai2*i»2 + 2a13a;1*s + • • •, which is called, a symmetric quadratic form, (c) What does the quadratic form become if A is skew-symmetric?, 15.59. Write the quadratic forms, in terms of matrices., 15.60. If A is Hermitian or skew-Hermitian, the quadratic form XTAX is called a Hermitian or skewHermitian form respectively. Prove that for every choice of X (a) the value of a Hermitian form, is always real, (b) the value of a skew-Hermitian form is zero or pure imaginary., 15.61., , Prove that every square matrix C can be written as A + B where A is Hermitian and B is, skew-Hermitian., , 15.62. The concept of a matrix whose elements are real or complex numbers can be extended to one whose, elements themselves are matrices. In such case the elements are called submatrices. Rules for, addition, multiplication, etc., analogous to those on pages 312-313 can be made. Prove that if, for, example,, , where the elements are matrices, then, , assuming the submatrices are conformable., DETERMINANTS, 15.63. Evaluate the determinants, , 15.64. Evaluate, , 15.65. Express each of the following as a single determinant and verify the result:, , 15.66. Illustrate each step of the proof in Problem 15.7 by referring to the determinant of Problem, 15.64(a)., 15.67. Prove Theorem 15-1 and illustrate by an example., 15.68. A triangular matrix A has all elements above [or below] the main diagonal equal to zero., that det (A) = 0., 15.69. If A and B are the matrices of Problem 15.50, prove that at least one of them is singular., 15.70. Prove Theorem 15-2, page 345, and illustrate by an example., 15.71. Prove Theorem 15-3, page 345, and illustrate by an example., , 15.72. (a) Show that, , [Hint: Use Problem 15.7.], , Prove
Page 381 :
CHAP. 15], , MATRICES, , 371, , (6) By transforming the fourth order determinant in (a) using Theorem 15-7, page 345, show that, it is equal to, , Thus prove Theorem 15-8, page 345, for 2 X 2 matrices., 15.73. Generalize Problem 15.72 and thus prove Theorem 15-8 for square matrices of any order., 15.74. Illustrate Theorem 15-9, page 346, by an example., 15.75. Prove Theorem 15-10, page 346., INVERSE OF A MATRIX, 15.76. Find the inverse of each of the following matrices and check results., , 15.77. (a) Prove that if AB — I, then B — A~l., a n d, , Use this result to find directly the inverses of, , H i n t :, , i n, , ( a ), , a s s u m e, , and find, , B, , 6, , 11> 612> &21> 622-], , 15.78. Prove that (.A-1)-1 — A where A is a non-singular matrix and illustrate by an example., 15.79. Is it true that if d e t ( A ) ^ 0, statements., , (a) (A"1)2 = (A2)-i,, , (&) (Am)n - (An)m = Amnl, , Justify your, , 15.80. Prove that (ABC)-1 = C~1B-^A-1 and generalize., 15.81. Discuss the significance of the inverse of a matrix with special reference to (a) Problem 15.53,, (6) Problems 15.55 and 15.56., ORTHOGONAL AND UNITARY MATRICES. ORTHOGONAL VECTORS, 15.82. Show that, , are orthogonal matrices., , 15.83. Show that, , are unitary matrices., , 15.84. Determine the form of the most general second order unitary square matrix of (a) order 2,, (6) order 3., 15.85. If A is a unitary matrix, prove that det (A) = eia for some constant a. Illustrate by an example., 15.86. (a) Show that the vectors, , are mutually orthogonal. (6) From the vectors, , in (a) determine a set of mutually orthonormal vectors.
Page 382 :
372, , MATRICES, , [CHAP. 15, , 15.87. Find a unit vector which is orthogonal to each of the vectors, , 15.88. If, matrix., , are mutually orthogonal, prove that, , is an orthogonal, , Can you determine an analogous result for a unitary matrix?, , SYSTEMS OF LINEAR EQUATIONS, 15.89. Solve the systems of equations, , 15.90. The currents /i,/2.^s, , an, , d I* in an electric network satisfy the system of equations, , Find Is., 15.91. Classify each of the following systems of equations according as they (i) have a unique solution,, (ii) are inconsistent (iii) are dependent. Determine solutions where they exist., , 15.92. For what value or values of k will the system of equation, , have solutions other than the trivial one? Determine some of these solutions., , 15.93. Given the system of equations, , show that any two of xlt x&x3 can be, , solved in terms of the remaining one and thus that there are infinitely many solutions., , 15.94. Given the system of equations, can be solved in terms of the remaining one., , determine whether any two of xltx2,xs
Page 383 :
CHAP. 15], , MATRICES, , 373, , 15.95. Investigate each of the following systems for possible solutions., , EIGENVALUES AND EIGENVECTORS, 15.96. Find eigenvalues and corresponding eigenvectors for each of the following matrices., , 15.97. Determine sets of unit eigenvectors corresponding to the matrices of Problem 15.96., 15.98. (a) Prove that if the eigenvalues of a matrix A are X1( X2, Xj, x|,... . (6) Generalize the result in (a)., , then the eigenvalues of A* are, , 15.99. Prove that the eigenvalues of a skew-Hermitian matrix [or skew-symmetric real matrix] are, either zero or pure imaginary., 15.100. Illustrate the result of Problem 15.99 by means of an example., 15.101. Prove that the eigenvalues of a unitary [or real orthogonal matrix] all have absolute value equal, to one., 15.102. Illustrate the result of Problem 15.101 by means of an example., 15.103. Find matrices which transform those of Problem 15.96 to diagonal form., 15.104. Prove that the eigenvalues of A and B~1AB are the same and illustrate by an example., 15.105. If the eigenvector corresponding to a given eigenvalue of a matrix A is X, prove that the eigenvector corresponding to the same eigenvalue of B~1AB [see Problem 15.104] is B~1X. Illustrate, by an example., 15.106. (a) Write the quadratic form 5x* — 2x\ — Sx* + IZx^Xz — 8xtx3 + Z0xzxs m *ne matrix form, XTAX. (b) Find the eigenvalues and eigenvectors of A. (c) Find the matrix B and the transformation equations X = BU when U = (% u.2 us)T so that the quadratic form in (a) is reduced, to canonical form, (d) Write the new quadratic form., 15.107. (a) Find a transformation which removes the xy term in x2 + xy + y2 — 16 and (b) give a geometric, interpretation to the result., 15.108. Discuss the relationship between Problem 15.107 and the problem of finding the maximum or, minimum of x2 + y2 subject to the condition x2 + xy + y2 = 16. [Hint. Use the method of Lagrange, multipliers.], 15.109. (a) Discuss the relationship between the problem of removing the cross product terms in, 5x1 ~ 2a!2 ~ 3a;3 + 12^1*2 — SajXg + 20^2*3 = 100 [Problem 15.106] and finding the maximum or, minimum of x\ + x\ + x\ subject to this constraint. Give geometric interpretations. (6) What, is a corresponding problem for reduction of any quadratic form to canonical form?
Page 384 :
374, , MATRICES, , [CHAP. 15, , 15.110. (a) Verify that the eigenvalues of, , are X = 1,1,2., , (6) Show that an eigenvector corresponding to X = 2 is, independent eigenvectors, namely, , and that there are two linearly, , , corresponding to the single eigenvalue X = 1., , (o) Can Theorem 15-15 be used to transform the matrix in (a) to diagonal form? Justify your, answer., 15.111. Reduce, , to canonical form., , 15.112. Explain how to reduce a Hermitian matrix to diagonal form and illustrate by an example., 15.113. Explain how to reduce a Hermitian quadratic form [see Problem 15.60] to canonical form and, illustrate by an example., 15.114. (a) Prove that the sum of the diagonal elements of any square matrix, i.e. the trace, is equal to, the sum of the eigenvalues of the matrix, (b) Illustrate by using the matrices of Problems, 15.33, 15.35, 15.96. (o) Is there a relationship between the trace of a matrix A and the matrix, B~1AB1 Explain your answer., 15.115. (a) Verify the Cayley-Hamilton theorem for a third order matrix., theorem for any nth order matrix., , 15.49. * = -33/5,, 15.76. (a), , (6) Prove the Cayley-Hamilton, , Answers to Supplementary Problems, y = -26/5, (b), , (c), , (d), , (e), , 15.77. (6), 15.89. (a) *! = 3, *2 = -4, , (b) x^ = 4, x2 - —1, xa = 2, , (c) art = -2, *2 = 4, *3 = -3, , 15.90. 40, 15.91. (a) inconsistent. (6) has unique solution x1 = —1, »2 = 2, x3 = —4. (c) has only trivial solution, xl = 0, *2 = 0, *3 = 0. (d) dependent, (e) dependent and has solutions other than trivial one., , 15.96. (a) 3,4; (i),(j), , (6> °' 5 = (i). (4), , (c) 5> 3± 4i, , ', , (d) 1,-1,2;|, , 15.106. (o), , (c), , / 5 6 -A AA, (»! xz xa) I 6 -2 10 ) x2 ), \-4 10 -8/\*,/, B, , ~, , (6), , 5,9, -15, , , x\ = i(-2ttx + 2tt2 + M3), *2 = £(MI + 2w2 - 2u3),, xa = £(2Ml + «2 + 2u3), , (d) Gul + 9«1 - I5ul
Page 385 :
Chapter 16, Calculus of Variations, , MAXIMUM OR MINIMUM OF AN INTEGRAL, One of the main problems of the calculus of variations is to determine that curve connecting two given points which either minimizes or maximizes some given integral. For, example, the problem of determining that curve connecting two points (#1,2/1) and (#2,2/2), whose length is a minimum is the same as that of finding the curve Y = y(x) where, 2/(#i) = 2/1, 2/(#2) = 2/2 such that, , W, is a minimum., In the general case we want to find the curve Y = y(x) where y(xi) = y\, y(x£) = 2/2, such that for some given function F(x, y, y'),, (2), , is either a maximum or minimum, also called an extremum or stationary value. A curve, which satisfies this property is called an extremal. An integral such as (2) which assumes, a numerical value for some class of functions y(x) is often called a functional., EULER'S EQUATION, In order to find the required curve Y — y(x), we, consider the effect on the integral (2) of neighboring, curves [see Fig. 16-1], (3), , where i)(x) is an arbitrary function and e is an arbitrary parameter. In order for the curve (3) to pass, through (xi, 2/1) and (xz, 1/2), we require that, W, , We can show [see Problems 16.1 and 16.2] that the, required curve Y - y(x) satisfies the equation, , Fig.l6-l, (5), , which is called Euler's equation. The condition (5) is a necessary condition for y = y(x), to be an extremal but it is not a sufficient condition., In case F(x,y,y') does not contain x explicitly, a first integral of (5) is found to be, [see Problem 16.4], W, 375
Page 386 :
CALCULUS OF VARIATIONS, , 376, , [CHAP. 16, , CONSTRAINTS, , In certain problems we want to find that curve which makes a given integral, (7), , a maximum or minimum but at the same time keeps the integral, (*), , equal to some constant. This type of problem in the calculus of variations is one which, involves a constraint condition, namely that the integral (8) is a constant. A special case, of this is the problem of determining that curve having a given perimeter which encloses, the largest area [Problem 16.8]. Because of this special case, we often refer to this class, of problems as isoperimetric problems., Such problems can generally be solved by using the method of Lagrange multipliers., To do this we consider the integral formed by adding (7) to A. multiplied by (8), whose A. is, the Lagrange multiplier. The resulting integral given by, (9), must be an extremum and leads to the Euler equation, where, , (10), , Using this we can find the required extremal. Extensions can be made to cases where there, are more constraint conditions., THE VARIATIONAL NOTATION, , It is often convenient to use a variational symbol 8 having properties which are in many, ways analogous to the differential d of the calculus., Given a function F(x, y(x), y'(x)), or briefly F(x,y,y') where we consider x as fixed,, we define, (11), where « and 17 = rj(x) have the same meaning given on page 375. Using the Taylor, expansion, (12), (11) can be written, , (IS), The sum of the first two terms on the right of (13) is denoted by 8F called the variation, of F, i.e., (U), If in particular F = y or F = y' in (14), we have, (15), , so that (14) can be written, (16)
Page 387 :
CHAP. 16], , CALCULUS OF VARIATIONS, , 377, , From (15) we see that, (17), , i.e., , (18), , showing that the operators 8 and d/dx are commutative., The variational symbol and its properties provide approaches alternative to those, involving t and r)(x) for dealing with problems of finding extrema of integrals. Thus we, can show for example that a necessary condition for the integral (2) to be an extremum is, , (19), which in turn leads to the Euler equation. See Problem 16.11., GENERALIZATIONS, , The ideas above can be extended. An example is the problem of finding curves, Xi = Xi(t), Xz = xz(t), ..., Xn = xn(t) such that using Xi = dxi/dt, ..., xn = dxjdt,, (20), , is a maximum or minimum. A necessary condition for this is that the Euler equations, (21), , are satisfied. A solution of these equations leads to the required curves. See Problem 16.12., Generalizations to cases where there are constraints are also possible. See Problem 16.13., It is also possible to generalize to cases where multiple integrals are used rather than, single integrals, and also where endpoints may not be fixed. See Problems 16.14-16.16., HAMILTON'S PRINCIPLE, , According to Newton's laws, a particle of mass m moves in a path according to the, equation, (22), , where F is the external force acting on the particle, r is the position vector with respect to, the origin of some fixed coordinate system, and * is the time., Now if the force field is conservative, then there exists a potential function V such that, , (23), The kinetic energy of the particle is defined as, (24), if r = xi + yj + zk.
Page 388 :
[CHAP. 16, , CALCULUS OF VARIATIONS, , 378, , We can then show that equation (22) follows as a consequence of the problem of finding, the path of the particle such that the integral, (25), , is an extremum, actually a minimum. In this integral ti and fe are two specified times and, the path is required to join the positions of the particles at these times., The principle that a particle moves in such a way that (25) is a minimum is often called, Hamilton's principle. It can be generalized to systems of two or more particles., LAGRANGE'S EQUATIONS, , Very often in practice the position of a particle at any time can be described by a certain, minimum number of variables called generalized coordinates. For example, if we consider, the pendulum bob of a pendulum to be a point mass [Fig. 16-2] then its position is defined, by using the generalized coordinate 0 given by the angle made between the pendulum rod, and the vertical., , Fig. 16-2, , Fig. 16-3, , Similarly in the case of a double pendulum with two masses mi, mz, as in Fig. 16-3, the, positions are specified by using two angles 9i, 02 which are the generalized coordinates., The potential and kinetic energy can be expressed in terms of these generalized, coordinates which are often denoted by q\, QZ, ...,QN. The number N of required coordinates, is often called the number of degrees of freedom of the system., According to Hamilton's principle the system moves so that, (26), , is an extremum where, , (27), , is called the Lagrangian of the system. Euler's equations then become, (28), , which are then called Lagrange's equations. From these equations the motion of the, system can be obtained., STURM-LIOUVILLE SYSTEMS AND RAYLEIGH-RITZ METHODS, , The calculus of variations often provides important methods for solving boundaryvalue problems. For example, the eigenvalues and eigenfunctions of the Sturm-Liouville, system
Page 389 :
CHAP. 16], , CALCULUS OF VARIATIONS, , 379, , (29), , <hy(a) + a2y'(a), , =, , 0,, , biy(b) + b2y'(b), , =, , 0, , (80), , considered in Chapter 11 can be formulated as a problem in finding extrema of suitable, functionals., Methods for finding approximate solutions of boundary-value problems by use of, variational principles are called Rayleigh-Ritz methods. See Problems 16.22-16.25., , Solved Problems, EULER'S EQUATION AND APPLICATIONS, 16.1. Let Y = y(x) be the curve joining points (xi, y\), (x2,3/2) which m a k e s d x, an extremum and Y = y(x) + ^(x), 77(0:1) = 0, r)(x?) = 0 be a neighboring curve joining these points. Prove that a necessary condition for this extremum is, , The value of the integral along the neighboring curve is, , a), Now this function of e is a maximum or minimum for the curve Y = y(») when, at, , e = 0, , (2), , If we denote the integrand in (1) by Fc, we have by differentiating under the integral sign, [Leibnitz' rule], , At « = 0 we thus have, , (S), , using integration by parts and the fact that u(*i) = 0, ij(«2) = 0. Since this is equal to zero from, (2), the required result follows.
Page 390 :
380, , CALCULUS OF VARIATIONS, , [CHAP. 16, , 16.2. Prove that a necessary condition for the extremum in Problem 16.1 is, , From Problem 16.1 we must show that the condition, , W, where TI(X) is arbitrary implies that the integrand is also zero. i.e., (2), , To show this we suppose the contrary, i.e. the coefficient of ii(x) is not identically zero. Then, since ii(x) is arbitrary, we can always choose it to be positive where, negative where, , and, , In such case the left side of (1) will be positive and so will, , give us a contradiction. Thus (2) follows., , 16.3. Show that Euler's equation can be written in the, , form, , We have, Also, Then by subtraction, and using Euler's equation gives the required result., , 16.4. If F does not involve x explicitly, show that the Euler equation can be integrated to, yield, If F does not depend explicitly on x, then dF/dx = 0 so that from Problem 16.3 we have, , 16.5. A curve C joining points (x\, yi) and (xa, y2), [see Fig. 16-4] is revolved about the x axis., Find the shape of the curve so that the surface thus generated is a minimum., The surface area is given by, , Since the integrand is independent of x, we can, use Problem 16.4 with F — y\fl + y'% to show, that the required curve is a solution of, Fig. 16-4
Page 391 :
CHAP. 16], , CALCULUS OF VARIATIONS, , 381, , which becomes on simplification, U), , Separating the variables and integrating,, , i.e., , (2), , where the constants e and fc are determined from points (*j, j/i), («%, 1/2). The curve (2) is often, called a catenary from the Latin meaning chain, since this is the shape in which a chain would, hang if suspended from the points (xlt j/x) (x2, yz). The problem is also of importance in connection, with soap films which are known to take shapes having minimum surfaces., It should be mentioned that we have not proved that the surface is actually a minimum, which, requires further analysis., , 16.6. Work Problem 16.5 without using the results of Problems 16.3 and 16.4., In this case we use, , in Euler's equation to obtain, , which simplifies to, Letting, integrating,, , (1), Letting y' = p so that(1) becomes on separating variables and, , Solving for p,, and the result can be found as in Problem 16.5., , 16.7. A wire in a vertical plane connects origin O and point, Pz(xz,y2) as indicated in Fig. 16-5. A bead of mass, m placed at O slides without friction down the wire, to P2 under the influence of gravity. Find the shape, of the wire so that the bead goes from O to P2 in, the least time., Assume that at time t the bead is at P(x, y) and that arc, OP = s. Then from mechanics we have, Kinetic energy at 0 + Potential energy at O = Kinetic energy at P + Potential energy at P, or, , i.e., Thus the time for the bead to go from O to Pl is
Page 392 :
382, , CALCULUS OF VARIATIONS, , [CHAP. 16, , The time will be a minimum when the integral is a minimum. Letting, using Problem 16.4,, , and, , which simplifies to, Letting, , and solving for y',, , Separating variables and integrating,, (1), L, , e, , t, , t, , i, , n, , g, , (, , 2, , ), , in (1), we have, , Thus the parametric equations of the curve are given by, , x = 6(0 — sin 0) + k,, , y = 6(1 — cos 0), , (3), , where 6 = a/2, 0 = 2s. Since the curve passes through the origin we must have fc = 0 so that, the required equations are, x = 6(0 — sin 0),, y = 6(1 — cos 0), (4), The constant 6 is determined from the fact that the curve must pass through (xz, y2)., The curve represented by (4) is a cycloid and is the path of a fixed point A on a circle of radius, 6 as it rolls along the x axis [see Fig. 16-6]., The problem is often called the brachistochrone problem from the Greek words brachistos, meaning shortest and ehronos meaning time., , Fig. 16-6, , CONSTRAINTS, 16.8. Find that curve C having given length I which encloses a maximum area., By Problem 6.17, page 164, the area bounded by C is, U), while the arc length is, , (2), , Using the method of Lagrange multipliers, we consider, (3)
Page 393 :
CHAP. 16], F r o m, , CALCULUS OF VARIATIONS, t h e, , E u l e r, , 383, ( 4 ), , e q u a t i o n, , (5), , w h r e, , we find, (6), , o r, , i.e., Solving for y',, Then on integrating,, , i.e., , (7), , which is a circle. A problem of this type is often referred to as an isoperimetric problem., Another method., , By carrying out the integration in (6), we have, , which shows that the curvature of C must be constant, i.e. C is a circle., , THE VARIATIONAL NOTATION, 16.9. Let Fi and F2 be functions of x,y,y'., (b) 8(FiF2) = FiSFi + FzSFi., (a) By definition,, , 8(Fj + F2) =, , (6) By definition,, S(F1Fli), , 16.10. Prove that, Method 1., , =, , Prove that (a) 8(Fi+F2) = 8Fi + $F2,
Page 394 :
384, , CALCULUS OF VARIATIONS, , [CHAP. 16, , Then by definition we have as required,, , Method 2., , where we have used Leibnitz's rule for differentiating under the integral sign., , 16.11. Show that a necessary condition for, , to be an extremum is, , By Problem 16.1, equation ($), we see on multiplying by e that a necessary condition for an, extremum is, But this can be written, as required., , *2, , / *i, , auier equation as in rrooiem 10.2., , F dx = 0, we can arrive at the, , GENERALIZATIONS, 16.12. Show that a necessary condition for, to be an extremum [maximum or minimum] is that, , As in the one dimensional case, a necessary condition for an extremum is, , i.e., Using the fact that S^ = d(Sxl)/dt, ..., Sxn = d(Sxn)/dt, , and integrating by parts, we have, , and since Sxlt .. ,,8xn are arbitrary we, , have, , k - 1,2, ...,n, The result can also be found without the use of 8.
Page 395 :
385, , CALCULUS OF VARIATIONS, , CHAP. 16], , 16.13. Show that a necessary condition for, to be an extremum subject to the constraint condition G(XI, #2) = 0 is, , where A is a Lagrange multiplier which may be a function of t., We must have as in Problem 16.12,, , or, , (1), , Also from G(x\, x%) = 0 we have, (*), , Multiplying (2) by X, which may be a function of t, and integrating we have, (3), Adding (1) and (3) gives, , Then since Sa^ and Sx2 are arbitrary the required equations follow. Generalizations to any number, of variables xlt .. ,,xn are immediate., , 16.14. If G — G(x, y) and ^ is a region bounded by a simple closed curve C in the xy plane,, show that a necessary condition, for, to be an extremum is that G satisfy Laplace's equation, i.e. V2G = 0, in ^., A necessary condition is, W, Now ", , ^, , Let <£ be a region such as indicated in Fig. 16-7, where lines parallel to the x and y axes meet C in, no more than two points., We have, , Fig. 16-7
Page 396 :
[CHAP. 16, , CALCULUS OP VARIATIONS, , 386, , using the fact that &G = 0 on C., , Similarly we find, , It follows from (1) that, , so that since 8G is arbitrary,, , 16.15. If y(x) makes, , dx an extremum where y(xi) is fixed but y(x2) may vary,, , prove that we must have, , As in Problem 16.1 we find the necessary condition for an extremum to be, , w, Then since *! is fixed but x2 is not, we have TI(XJ) = 0 while q(x2) is not necessarily zero. Thus, (1) becomes, (2), , Now since i» is arbitrary subject only to i)(*i) = 0, it must in particular hold for i/(*2) = 0, i.e., (3), , from which, , (4), , Using (4) in (£), we must then have, (S), , Thus if we are to satisfy (5) for y(x2) not necessarily zero, we must have, , 16.16. A frictionless wire in a vertical plane connects the origin 0 of an xy coordinate, system to a point P2 located somewhere on a given vertical line x = #2 [see Fig. 16-8]., Find the shape of the wire so that a bead of mass m placed on it at O will slide down, under the influence of gravity to Pz in the least time., The problem is identical with that of Problem, 16.7 except that instead of being fixed at (xz, y2) the, endpoint P2 can vary along the line x = x2 where x2, is prescribed., As in Problem 16.7 the total time taken to go, from 0 to Po is, <*), , Then b y Problem 16.15, t, have, , a, , k, , i, , n, , g, , , we, , Fig. 16-8
Page 397 :
CALCULUS OF VARIATIONS, , CHAP. 16], , 387, , (*>, , The first condition in (2) leads as in Problem 16.7 to the fact that the shape of the wire must be a, cycloid. The second condition in (2), i.e., , shows that y' = 0 at x — x2 or that the tangent to the cycloid at P2 must be parallel to the, x axis or, in other words, that the cycloid must be perpendicular to the line « — «2 at P2. These, conditions are enough to enable us to obtain the equations of the cycloid., , HAMILTON'S PRINCIPLE AND LAGRANGE'S EQUATIONS, 16.17. Derive Hamilton's principle for a system of n particles from Newton's laws., Assume the n particles have masses mk, k = 1,2, ..., n, that they have position vectors rk,, k = 1,2, ..., n, relative to an xyz coordinate system and that the forces acting on them are, Fk, k = 1,2, ..., n. Then the path Ck of the fcth particle is determined from the equation, (1), , Assume that we vary the path of the feth particle without changing the endpoints and let this, variation, often called virtual displacement, be 3rfc; then from (1),, (2), Summing over all particles, we obtain, (3), where the right hand side of (8) is the total work SW done under the displacement of the path, i.e., on using (1)., , (4), , Now the total kinetic energy of the system is, (5), , from which we have, («), But, , (7), , so that on multiplying by mk and summing from k — 1 to n,, , where we have used (4) and (6). Integrating this with respect to t from *t to tz which represent, the times at the endpoints of the path C, we have, (*), , since 8rfc = 0 at tj and t2. If the force field is conservative there is a potential V so that W = —V., Thus (8) becomes
Page 398 :
[CHAP. 16, , CALCULUS OF VARIATIONS, , 388, , which is Hamilton's principle, i.e. a system moves from time ti to time tz in such a way that, , is an extremum, actually a minimum., , 16.18. Derive Lagrange's equations from Hamilton's principle., If the generalized coordinates specifying the position of a body are given by qltq%, ...,q$,, then the position vector rfc of each particle of the body is a function of qit q%, .. •, q^ so that its, velocity rfc = dik/dt is a function of qi,qz, • • • » q r w »nd q\,kz> • • • > t f w Thus the kinetic energy T, is a function of qlt qz,..., qN, glf qz, .. •, ?N- Also the potential energy, which we shall assume, depends only on position, is a function of <?i, q%, qN., Now by Hamilton's principle the body moves so that, , is an extremum. Thus by Euler's equations we have on writing L = T — V, called the Lagrangian,, k = l,...,N, which are often also referred to as Lagrange's equations., , 16.19. A mass m, suspended at the end of a vertical spring which has spring constant «, and negligible mass, is set into vertical vibration [see Fig. 16-9]. Find the equation, of motion of the mass., If we let x be the displacement of m from the equilibrium position,, then by Hooke's law the force is given by, , M, where i is a unit vector in the downward direction. Since, (2), where V is the potential energy, we have from (1) and (2), , taking the arbitrary constant as zero., , Fig. 16-9, , The kinetic energy of the mass is, , Thus the Lagrangian is, Then Lagrange's equation describing the motion of the mass is, or, , which agrees with the result obtained from Newton's laws.
Page 399 :
389, , CALCULUS OF VARIATIONS, , CHAP. 16], , 16.20. A particle of mass ra moves in the xy plane under the influence of a force of attraction to the origin O of magnitude F(P) > 0 where p is the distance of the mass from, O. Set up the equations describing the motion., Use polar coordinates (p, <f>) to locate the position, of m, [Pig. 16-10]. Since the rectangular coordinates (x, y) of m, are related to the polar coordinates by, the position vector is given by, where i and j are unit vectors in the x and y directions, respectively. Thus, , Fig. 16-10, , so that the kinetic energy is, , since the force is given by, where pt is a unit vector in the direction of p and since, , we have, Then the Lagrangian is, Thus the Lagrange equations, , become, , (1), , From the second equation of (1) we have, , (2), 2, , where K is a constant. Then using $ = /e/p in the first of equations (1), we obtain, (*), , These equations can be used to describe the motion if F(p) is known., This problem is useful in discussing the motion of the planets around the sun., , 16.21. Use Hamilton's principle to find the equation for the, small vibrations of a flexible stretched string of, length I and tension T fixed at its endpoints., In the vibration [see Fig. 16-11] an element of length dx, of the string is stretched into an element of length ds where, , approximately, using the binomial theorem and the fact that, (dy/dx)2 is small compared with 1., , Fig. 16-11
Page 400 :
390, , CALCULUS OF VARIATIONS, , [CHAP. 16, , The work done per unit length against the force of tension T is then given by, , Thus the total work done for the whole string [which by definition is the potential energy] is given by, U), , The velocity of the string is dy/dt and if the density [mass per unit length] is n then the total, kinetic energy of the string is, (2), , Then Hamilton's principle states that, (S), , or equivalently, , (4), , Thus by Problem 16.52 with y = t, z — Y we have, (5), , where F is the integrand in (4) given by, («), , Using (6) in (5) we obtain, , (7), , or if T and /t are constants and o2 = T//I we have the required equation, (*), , STURM-LIOUVILLE SYSTEMS AND RAYLEIGH-RITZ METHODS, 16.22. Show that t h e extremals o, , f, , s, , u, , b, , j, , e, , c, , t t o t h e constraint, , are solutions of the Sturm-Liouville equation, , We shall find it convenient to use the Lagrange multiplier —X rather than X. Then proceeding, as in Problem 16.8, we must make, , an extremum. Denoting the integrand by F and using the Euler equation, , we obtain the required Sturm-Liouville equation.
Page 401 :
CHAP. 16], , CALCULUS OF VARIATIONS, , 391, , 16.23. Show that the extremals of, , where the denominator need not be equal to 1 as in Problem 16.22, are also solutions, of the same Sturm-Liouville system of Problem 16.22., The ratio of the integrals can be denoted by, X = IIJ, Then we have if X is an extremal,, , using the fact that Sy = 0 at xl and x2. Since Sy is arbitrary we are again led to Sturm-Liouville, equation, , (py'Y + qy + \ry = 0, , 16.24. Show how to use Problem 16.23 to find eigenvalues and eigenfunctions of the SturmLiouville system, ai#(a) + a2y'(a) = 0,, , biy(b) + b2y'(b) = 0, , Problem 16.23 with x^ = a, x2 = b shows that we can find the eigenfunctions of the SturmLiouville equation by determining the non-trivial functions y(x) which satisfy the conditions, «i2/(«) + <*22/'(a) — 0, bjy(b) + b2y'(b) = 0 and which make the ratio X = 113 an extremum. The, corresponding values of X are then the eigenvalues. That this is true can be seen by noting that if, an eigenfunction and eigenvalue are given by y\(x) and Xj respectively, then for this eigenfunction j/t, , (1), On integrating the numerator by parts, we have, , using the boundary conditions to show that, , and the fact that, , Thus it follows from (1) that X = Xt., It can be proved that the smallest eigenvalue is the minimum value of the ratio in (1) for all, possible functions satisfying the boundary conditions in the Sturm-Liouville problem. This is often, called Rayleigh's principle.
Page 402 :
392, , CALCULUS OF VARIATIONS, , [CHAP. 16, , 16.25. (a) Find approximately the smallest eigenvalue of the Sturm-Liouville system, y" + \y = 0,, , y(0) = 0,, , »(I) = 0, , and (6) obtain the corresponding approximate eigenfunction., (a) This is a special case of Problem 16.24 where p = 1, q = 0, r = 1, a = 0, 6 = 1, ax = 1,, «2 = o, 6j = 1, 62 = 0. To find the smallest eigenvalue, we consider, , V), , Assume that, , y — A0 + Atx + A2xz + Asx3, , (2), , To satisfy j/(0) = 0, j/(l) = 0 we require A 0 = 0, A 3 = —(Aj + A2) so that (2) becomes, y = A !<* - x*) + A2(*2 - *»), , (5), , Substituting into (1) we find, (4), , on dividing numerator and denominator by A§ and writing B = Aj/A2. Then X is a minimum, when d\/dB — 0, which leads to, 168B2 - 1792B - 637 = 0, (5), with solutions, , B - 11.01102, -.3443521, , (6), , The second value leads to a negative value for X which is clearly impossible from the form of (1)., The first value leads to X = 10.5289. As is easily found, the true eigenvalue is »r2 = 9.8696, approximately so that the error is less than 7%. Note that instead of (3) we could have, assumed directly that, y = cc(l-a;)(a 0 + 64*), which automatically satisfied the boundary conditions and then proceeded as above. Better, approximations could then be arrived at by choosing a convenient number of terms in the series, y = *(1 — «)(ao + ctiX + o2*2 + • • •), (6) Since B = Aj/A 2 = 11.01102, we find from (5), y = 11.01102A2(a; - x3) + A2(a;2 - a;3) = A2(11.01102a; + x2- 12.01102a;8)
Page 403 :
CHAP. 16], , CALCULUS OF VARIATIONS, , 393, , Supplementary Problems, EULER'S EQUATION AND APPLICATIONS, 16.26. If F(x, y, y') does not involve y explicitly, show that the extremals of, , are solutions of, 16.27. Find the extremals of, , and thus show that the shortest distance between two, , points in a plane is a straight line., 16.28. (a) Show that the shortest distance between two points expressed in polar coordinates (p, 0) by, (pi»0i) and (p2, ^2) ig, , (6) By minimizing the integral in (a) obtain the equation of a straight line in polar coordinates., 16.29. Work Problem 16.7, page 381, without using the result of Problem 16.4, showing that the differential, equation to be solved is 1 + y'2 — 2yy" = 0., 16.30. Show that if F is a function of y' alone, then the extremals of, 16.31., , are straight lines., , The shortest distance between two points on any surface is called a geodesic of the surface. Show, that the geodesies on the surface of a sphere of radius a are the arcs of great circles. [Hint:, First show that the element of arc length on the sphere is given by ds2 — a2(de2 + sin2 e d<f>2), where e and 0 are spherical coordinates.], , 16.32. Find the geodesies for (a) a right circular cylinder, (6) a right circular cone., 16.33. Find the extremals of, , such that, , 16.34. According to Fermat's principle, a light ray travels in a medium from one point to another so, that the time of travel given by, where s is arc length and v is velocity, is a minimum. Show that the path of travel is given by, , 16.35. Work Problem 16.34 if (a) v — y, (b) v = \/y,, curves are obtained if we use x in place of y., , (c) v = 1/VJ7 and show that the same types of, , THE VARIATIONAL NOTATION, 16.36. Prove that (a), 16.37. If F is a function of x,y,y', we have, , (b), , (c), so that by analogy we, , would expect t o h a v e D e t e r m i n e whether this analogy i s correct, or not by comparing with the result in equation (16), page 376.
Page 404 :
394, , CALCULUS OF VARIATIONS, , [CHAP. 16, , 16.38. Use the variational notation to justify the method of Lagrange multipliers given on page 376., 16.39. Work (a) Problem 16.5, (6) Problem 16.7, (c) Problem 16.33 by direct use of the variational symbol., CONSTRAINTS, 16.40. If the curve o of Problem 16.5 is required to have a given length I, what is the shape of c so that, when it is revolved about the * axis the surface generated will be a minimum? Compare the result, with that of Problem 16.5 and explain., 16.41. A curve c of given length I connects points (xlt j/j) and (xz, y2). Find the shape of o so that the, area bounded by c, the lines x = xlt x = x2 and the * axis will be a maximum., 16.42. A rope of length I suspended vertically from two fixed points hangs so that its center of gravity, is as low as possible. Prove that the curve in which the rope hangs is a catenary. [The principle, involved here is equivalent to the principle of minimum potential energy in mechanics.], dx is an extremum if, , 16.43. Find a function y(x) for whichis an extremum ifand j/(0) = 0,, »« = I-, , and, , 16.44. Of all curves enclosing a region of given area A which one will have the minimum length? Discuss, the relationship of this problem with Problem 16.8., GENERALIZATIONS, 16.45. Use the results of equations (20) and (21), page 377, to find extremals ofand, interpret the results., 16.46. Find extremals of, , and interpret the results., , 16.47. Show that a necessary condition for, subject to the two constraint conditions, , to be a maximum or minimum is, , where X1; \2 are Lagrange multipliers and G = Ft + \1F1 + \2F2., 16.48. Show that a necessary condition for, , to be an extremum is, and explain how the result can be generalized., 16.49. Find a function y(x) such t h a t w h i c h m, VW = 0, »"(0) = 0, y"(r) = 0., , a, , k, , e, , s, , a minimum i f j/(0) = 0 ,, , 16.50. If G = G(x, y, z) in a closed region ^, show that a necessary condition for, , to be an extremum [in this case a minimum] is that V2G = 0 in %
Page 405 :
395, , CALCULUS OF VARIATIONS, , CHAP. 16], , 16.51. (a) Show that in orthogonal curvilinear coordinates %, u2, u3, , where the element of arc length is, (6) Use Problem 16.51 and the fact that dx dy dz = h^hg dut du2 dus, equation in these curvilinear coordinates is, , to show that Laplace's, , 16.52. A surface S having equation z = z(x, y) is bounded by, a closed curve C whose projection on the xy plane is C', which forms the boundary of a region ^ [see Fig. 16-12]., Show that a necessary condition for z = z(x, y) to be, an extremal of, , where p = dz/dx,, , q — dz/dy,, , is, , 16.53. (a) Show that the surface area of S in Problem 16.52 is, , (6) Show that the surface with minimum surface area must be a solution of the equation, , which can be written as, where, (c) Show that, , is an exact differential., , 16.54. Generalize Problem 16.52 by showing that a necessary condition for z — z(x, y) to be an extremal of, , where r, s, t are defined in Problem 16.53, is, , 16.55. Generalize Problem 16.52 to the case where F — F(x,y,z,u,zx, zv,ux,uy)., 16.56. (a) Find a necessary condition for z = z(x, y) to be an extremal of, , if, (6) Give a geometric interpretation to part (a)., , (a constant)
Page 406 :
[CHAP. 16, , CALCULUS OF VARIATIONS, , 396, , 16.57. Show that the problem of finding geodesies on a surface <j>(x, y, z) = 0 joining points (xlt ylt zt), and (»2» 2/2' zz> can be f oun d by obtaining the minimum of, , and thus show that these geodesies can be found by solving simultaneously the equations, , 16.58. Assuming that the line x = x2 in Problem 16.16 is replaced by a curve g(x, y) = 0 in the vertical, plane, show that the shape of the wire should be a cycloid which intersects the curve at right, angles., 16.59. Investigate the problem of the brachistochrone (a) from a curve to a fixed point and (6) between, two curves., HAMILTON'S PRINCIPLE AND LAGRANGE'S EQUATIONS, 16.60. In a simple pendulum a mass m is suspended from a rod of negligible mass and length I and the, system can vibrate in a plane [see Fig. 16-13]., (a) Show that the potential energy of the mass is V = mgl(l — cos e) apart from an additive, constant., (6) Show that the kinetic energy is, (c) Use Lagrange's equations to show that '0 + (g/l) sin e — 0., (d) Show that if e is small so that sin e = e very nearly, then the period of vibrations is, Zir't/l/g approximately., , Fig. 16-13, , Fig. 16-14, , Fig. 16-15, , 16.61. Two masses ml and m2 are suspended vertically from an inextensible string which in turn passes, around a fixed pulley as shown in Fig. 16-14. Use Lagrange's equations to show that the acceleration of either mass is given numerically by, , 16.62. A system consisting of masses mllm2 connected to massless springs having stiffness factors, K1,K2 [Fig. 16-15] is free to vibrate vertically under the influence of gravity., (a) If xl and x2 are the displacements of m,l and m2 respectively from their equilibrium positions,, show that the potential energy of the system, is, (6) Write the equations of motion of the masses and solve., 16.63. (a) Show that for small angles elt 02 the equations of motion of the double pendulum of Fig. 16-3,, page 378, are given by, , and (6) find the natural frequencies of the motion.
Page 407 :
CHAP. 16], , CALCULUS OF VARIATIONS, , 397, , 16.64. Use Lagrange's equations to determine the motion of a mass m sliding without friction down an, inclined plane of angle a., 16.65. Show that the equations of motion of a particle of mass m in spherical coordinates (r, e, <t>) if the, potential is V(r, e, <t>) are given by, , 16.66. Use Hamilton's principle to prove that a system which is in equilibrium has minimum potential, energy., 16.67. A particle of mass m resting on top of a fixed sphere of radius a is given a slight displacement, so that it slides down the sphere. Assuming no friction, determine the place where it leaves the, sphere., 16.68. (a) A membrane situated in the xy plane executes small transverse vibrations denoted by, z(x, y, t). Assuming that the density and tension are given by n and r respectively, show that, the membrane vibrates so that, , *\., , where ^ is the region in the xy plane occupied by the membrane., (6) Use part (a) to show that, where a2 = T//I., STURM-LIOUVILLE SYSTEMS AND RAYLEIGH-RITZ METHODS, 16.69. Compare the approximate and exact eigenfunctions of Problem 16.25(6) by first normalizing them, so that, , and then plotting their graphs on the same set of axes., , 16.70. Express in variational form the problem of solving y" + \y = 0, y(0) = 0, y'(i) = 0 and find the, lowest eigenvalue and the corresponding eigenfunction. Compare with the exact solution., 16.71. Obtain an approximate solution for the extremals of, , and compare with the exact solution., 16.72. Show that y - 6 t sin vx + 62 sm 2jr» satisfies the boundary conditions in Problem 16.25 and, use this to find the approximate eigenvalue and eigenfunction. Discuss the significance of your, findings., 16.73. Use an appropriate trigonometric expansion to obtain an approximate solution to Problem 16.71., 16.74. The equation for a vibrating spring [see Problem 16.19] is m x + KX — 0. Assuming the boundary, conditions x(0) = 0, i(0) = v0, express the problem in variational form and show how to find the, period approximately by a Rayleigh-Ritz method., 16.75. Express the equation (1 - x2)y" - 2xy' + 12y = 0, j/(0) = 0, yd) = 1 in variational form and, obtain approximate solutions. Compare with the exact solution., 16.76. Is it possible to solve Problem 16.25 by finding extremals ofof the form, y = x(l — x)(a0 + 04;*;)? Justify your answer., , dx of the form
Page 408 :
398, , CALCULUS OF VARIATIONS, , [CHAP. 16, , 16.77. Show that eigenvalues and eigenfunctions for the fourth order Sturm-Liouville type differential, equation, can be obtained by considering the extremals of, , 16.78. In the periodic transverse vibrations y(», t) = v(x) sin at of a beam which is simply supported, at its endpoints x = 0, x = I, the boundary-value problem which arises is given by, V(0), , -, , v(l), , = V"(Q) = v"(l) = o, , (a) Express this in variational form using Problem 16.77 and (6) obtain the lowest "critical", frequency for the beam., , Answers to Supplementary Problems, 16.28. (6) p cos (<t> + Cj) = c2, 16.33. y = Bn sin nx, n = 1,2,3, ..., 16.35. (a) Circles (x - ctf + (y - c2)2 = e\, , (6) Catenaries, , (c) Parabolas, 16.43. y = £(1 - cos x) + £(2 - TT) sin *, 16.49. y = Bn sin nx, n = 1,2,3, ..., 16.70. Exact solution y = B sin ~, exact eigenvalue ^~, &, , 16.71. y = 2jr cos * — 2 sin * + 2x, 16.75. y = $(5x*-3x), , 4
Page 409 :
INDEX, Beams, 43, bending of, 43, 61, semi-infinite, 336, 337, Bending moment, 43, Ber and bei functions, 226, 235, 236, Bernoulli's equation, 40, 52, Bessel functions, 224-241, of first kind and order n, 224, 229-232, functions related to, 235, 236, graphs of, 224, half odd integer order, 224, 229, 230, integral representation of, 233, integrals of, 230, 231, modified, 226, orthogonality of, 227, 236, 237, of second kind and order n, 224, 225, 233-235, series of, 227, 228, 237, 238, solutions of partial differential equations, using, 260, 272-274, zeros of, 227, Bessel's differential equation, 224, equations transformed into, 226, 236, general solution of, 224, 229, 233-235, Beta function, 211, 212, 215-218, relation of to gamma function, 211, Bilinear transformation, 292, Binomial coefficients, 2, Binomial theorem, 2, Binomial vector, unit, 144, Box product, 124, Boundary-value problems, 41, 42, 46-48, 258, 262, [see also Differential equations], methods of solving, 260, 261, 268, 271, 272, solution of by complex inversion formula,, 335-338, Brachistochrone problem, 382, 386, 387, Branches, 287, Branch line, 310, Branch points, 289, 310, inverse Laplace transforms of functions with,, 330, 331, Bromwich contour, 324, 325, 333, Bromwich's integral formula, 324, , Abel's identity, 80, Absolute value, 3, of complex numbers, 12, Absolutely convergent series, 6, 7, 21, complex, 301, rearrangement of terms in, 6, Acceleration, 42,126,137, centripetal, 137, in cylindrical and spherical coordinates, 145, due to gravity, 42, along a space curve, 137, tangential and normal components of, 137, Amperes, 42, Amplitude, 57, of a complex number, 12, 29, Analytic continuation, 210, Analytic functions, 286, 296, condition for, 287, Analytic part of a Laurent series, 289, Anti-derivative, 5, Applications of differential equations, 41, 42,, 77, 102, 115, 116, to chemistry, 43, 60, to deflection of beams, 43, to electric circuits, 42, 92, 93, 105,116, to mechanics, 42, 90-92, 115, 116, to miscellaneous problems, 43, to orthogonal trajectories, 43, Approximate integration, 19, 20, 23, Approximate solutions to differential, equations, 43, 44, Arbitrary constant, 38, Arc length, 126, in cylindrical and spherical coordinates,, 140, 141, Area magnification factor, 291, Area under curve, 5,19, 20, Argand diagram, 12, Argument of a complex number, 12, 29, Associated Legendre functions, 243, 244, 251, orthogonality of, 244, Associative law, for addition and, multiplication of numbers, 1, for addition and multiplication of vectors,, 122, 130, for convolutions, 102, failure of for cross products, 124,135, Asymptotic formula, for Bessel functions, 227, for gamma function, 211, Asymptotic series or expansions, 212, 213,, 220, 221, Auxiliary equation, 73, Axioms, 2, , Calculus of variations, 375-398, Canonical form, of a linear second order, differential equation, 76, 77, 87, of a quadratic form, 349, Capacitance, 42, Capacitor or condenser, 42, Catenary, 381, Cauchy or Euler equation, 76, 86, Cauchy-Riemann equations, 287, 295-297, in polar coordinates, 318, Cauchy's inequality, 318, Cauchy's integral formulas, 288, , Base of logarithms, 2, Battery, 42, , 399
Page 410 :
400, , Cauchy's theorem, 287, 288, 297-300, proof of, 298, 299, Cayley-Hamilton theorem, 349, 366, Centimeter, 42, Centrifugal force, 92, Centroid, 157, of a surface, 168, C. G. S. units, 42, Characteristic equation, 73, 348, Characteristic functions, 245, Characteristic values, 245, of a matrix, 348, Characteristic vector, 348, Chemistry, applications of differential, equations to, 43, 60, Clairaut's equation, 41, 53, Collection, 1, Commutatitive law, for addition and, multiplication of numbers, 1, 12, for addition and multiplication of vectors, 122, for convolutions, 102, for dot products, 123, Complementary equation, 71, solution of, 72, 73, Complete equation, 71, 81-83, Completeness, 194, Complex inversion formula for Laplace, transforms, 324-341, applications of to boundary-value problems,, 325, Complex notation for Fourier series, 184, Complex numbers, 1, 11, 12, 28-30, commutative, associative and distributive, laws for, 12, operations with, 11, 28, 29, polar form of, 12, 29, 30, Complex plane, 12, Complex variables, 286-323, methods for solving partial differential, equations by, 261, Components, of a vector, 123, Component vectors, 123, Condenser or capacitor, 42, Conformal mapping, 291, 311-317, definition of, 291, of half plane on to a circle, 292, solutions of Laplace's equation by, 293,, 314-317, Conjugate of a complex number, 12, Conservative force field, 152, 176, 377, Constant coefficients, differential equations, with, 71, 73-76, Constant of integration, 38, Constraints, 11, in calculus of variations problems, 376,, 382, 383, Continuity, 4, 14, 15, definition of, 4, for functions of a complex variable, 286,, 294, 295, for functions of two or more variables, 8, in an interval, 4, piecewise, 99,105, , INDEX, Continuity (cont.), relationship of with derivatives, 4, 16, of vector functions, 125, Contour, 290, Convergence, of sequences, 6, absolute [see Absolutely convergent series], of series, 6, uniform [see Uniform convergence of series], Convolution theorem, for Fourier transforms,, 203, 206, for Laplace transforms, 102, Coordinate curve, 128, Cosine integral, 212, Coulombs, 42, Cramer's rule, 10, 25, 26, proof of, 359, 360, used in solutions of differential equations, 89, Critical points, 311, Critically damped motion, 91, Cross-cut, 164, Cross or vector products, 124, distributive law for, 124, 132,133, expressed as determinants, 129,133, failure of associative and commutative, laws for, 124,132, Curl, 127, 137-140, of the gradient, 127, in orthogonal curvilinear coordinates, 129, Curvature, 137,144, Curve, 2, simple closed, 151, Curvilinear coordinates, 128,140-142, special, 129, used in integrations, 148,149, Cycloid, 382, 387, Cylinder, temperature in, 337, 338, Cylindrical coordinates, 129, Damped oscillatory motion, 91, Damping force, 91, De Moivre's theorem, 12, 29, 30, Definite integrals, 5, evaluation of by complex variables, 290, Deflection curve, 43, Deflection of beams, 43, Degree, of a polynomial, 2, Degrees of freedom, 378, Del (V), 126, formulas involving, 127, Density, 259, 263, Density function, 185, Dependent variable, 2, 8, 261, Derivatives, 4, 15-18, definition of, 4, of functions of a complex variable, 286,, 295-297, higher order, 4, Laplace transforms of, 106,107, relationship of with continuity, 4, 16, theorems on, 15-18, of vector functions, 125, Determinants, 9,10, 344-346, 352-357, cofactors of, 345
Page 411 :
401, , INDEX, Determinants (cont.), Cramer's rule for [see Cramer's rule], Laplace expansion of, 345, 353, minors of, 344, 345, of second order, 9, solutions of differential equations using, 89, theorems on, 345, 346, of third order, 10, value of, 345, 352, 355, Dextral system, 123, Difference, of matrices, 342, of vectors, 122, Differentiable functions, 4, Differential, 4, 9,18, 24, 25, exact, 25, of vector, 125,136,137, Differential equations, 38-70, 258-285, classification of, 38, 44, 45, definition of, 38, of families of curves, 39, 46, 47, first order, 39, 40, 45, 46, general solutions of, 39, 45-49, of higher order than first, 41, 53, 54, Laplace transform solutions of, 102,114,115, linear, 71-97, order of, 38, 258, ordinary, 38-70, partial, 258-285, [see also Partial differential equations], particular solutions of, 39, 45, 46, 48, 258, 262, singular solutions of, 39, 46, 59, 258, solutions of, 38, 45, 46,102,114,115, solvable for one unknown, 41, 53, Differential operators, 71, Differentiation, 4, formulas, 4, 5, Diffusivity, 259, 263, Dirac delta function, 119, Directional derivative, 140, Dirichlet conditions, 183,188,189, 201, Dirichlet integrals, 212, 219, 220, Dirichlet problem, 293, Discharge coefficient, 59, Discontinuity, 15,186,189 [see also Continuity], of sum of series, 21, Distributive law, for convolutions, 102, for dot products, 123,131, for numbers, 1,12, for vectors, 122, Divergence, of the curl, 127, 139, in cylindrical coordinates, 129, 141, in orthogonal curvilinear coordinates, 128, of a sequence or series, 6, in spherical coordinates, 129, of a vector, 126, 137-140, Divergence theorem, 154, 169-172, proof of, 169,170, Division, 1, Dot product, 123, 124, 131, 132, Double integrals, 147,155,156, in curvilinear coordinates, 158, transformation of, 158, Drum, vibrations of, 273, 274, , Duplication formula for gamma function,, 211, 218, Dynes, 42, Eigenvalues and eigenfunctions, 245, 378, 379,, 391, 392, Eigenvalues and eigenvectors, 348, 349, 363-365, of a Hermitian matrix, 365, 366, theorems on, 349, 365-368, Elastic curve, 43, Electric and magnetic field vectors, 145, Electric circuits, 42, 56, 57, Electric potential, 259, Electricity, principles of, 42, 259, Electromagnetic theory, 145, Electromotive force, 42, Element, 2, Elementary functions of a complex variable, 286, E.m.f., 42, Envelope, 46, 59, Equality, of complex numbers, 12, of matrices, 342, of vectors, 121, Error function, 212, 221, 272, 331, 332, Essential arbitrary constants, 38, Essential arbitrary functions, 258, Essential singularity, 289, 302, 303, Euler's constant, 222, 225, Euler's equation, 76, 86, 375, 379-382, Euler's formulas or identities, 12, 30,184, Euler's method for differential equations, 43,, 61-63, Even and odd functions, 30,183,190-193, 201, Even extension, 193, Exact differential, 25, 39, 152, Exact Differential equation, 39, 40, 49-51, 77, 87, Existence and uniqueness theorems, 41, 71, Exponential function, 2, Exponential integral, 212, Exponential order, 99,105, Exponents, 2, Extremal, 375, Extremum, 27, 375, necesary condition for integral to be,, 375, 377, 379, 380, 384-386, Factorial function, 210, [see also Gamma function], Factorial n, 2, Factorization of operators, method of, 77, 87, 88, Farads, 42, Format's principle, 393, Flexural rigidity, 43, Flow problems, 59, Force, 42, field, 152, Fourier coefficients, 182, 188, generalized, 199, Fourier cosine series, 183, 184, Fourier expansion [see Fourier series], Fourier integrals, 201-210, methods of solving partial differential, equations by, 260, 271, 272
Page 412 :
402, Fourier series, 182-200, approximation to functions by, 194, complex notation for, 184, convergence of, 195-197, definition of, 182,183, differentiation and integration of, 184,195, half range, 183, 184, 190-193, methods for solving partial differential, equations by, 260, 267-271, Fourier sine series, 183, 184, 267, Fourier's integral theorem, 201, [see also Fourier integrals], connection of with complex inversion, formula, 326, proof of, 206-208, Fourier transforms, 202, 203-205, symmetric form for, 202, Fractional transformation, 292, Frenet-Serret formulas, 144, Frequency, 57, critical, 398, of vibration of a membrane, 274, of vibration of a string, 269, Fresnel sine or cosine integrals, 212, Frobenius, method of, 77, 88, 89, solution of Bessel's equation, 228, 229, solution of Legendre's equation, 246, Frobenius series, 77, Functional, 375, Functions, 2,13,14, of a complex variable, 286, 294, 295, [see also Complex variables], multiple-valued, 286, single-valued, 286, of two or more variables, 8, 23-25, Fundamental theorem of calculus, 5, Gamma function, 99,103, 210-223, Bessel functions in terms of, 224, Beta function in terms of, 211, recursion formula for, 99, Generalized coordinates, 378, General solutions, of ordinary differential, equations, 39, 45-49, of partial differential equations, 260, Generating functions, for Bessel functions,, 225, 232, 233, for Hermite polynomials, 244, 252, for Legendre polynomials, 247, Generator, 42, Geometry problems, 58, 59, Gradient, 126, 127,137-140, in cylindrical coordinates, 129, 141, in orthogonal curvilinear coordinates, 128, Gram, 42, Gravitation, theory of, 259, Gravitational potential, 259, Gravity, acceleration due to, 42, Greater than, 1, Green's theorem in space, 154, Green's theorem in the plane, 151,152,162-165, proof of for multiply-connected regions,, 163,164, , INDEX, Green's theorem in the plane (cont.), proof of for simply-connected regions, 162,163, Half plane, mapping of on to the unit circle,, 292, 293, 312, 316, Half range Fourier series, 183, 184, 190-193, Hamilton-Cayley theorem, [see Cayley-Hamilton theorem], Hamilton's principle, 377,378, 387, 388, derivation of Lagrange's equations from, 388, proof of, 387, 388, Hankel functions of first and second kinds,, 226, 235, Harmonic functions, 287, 314, 315, Heat conduction equation, 259, derivation of, 263, Heat flow, 43, 267, Heaviside's unit step function, [see Unit step function], Henries, 42, Hermite polynomials, 244, 252, 253, generating function for, 244, 252, orthogonality of, 252, 253, Rodrigue's formula for, 244, Hermitian matrix, 344, 249, eigenvalues of, 365, 366, Homogeneous differential equations, 40, 51, 52, Homogeneous equation, 71, 81, 82, solution, 72, 73, Hooke's law, 90, Hyperbolic functions, 3, Hypocycloid, 59, Infinite series, 6, convergence and divergence of, 6, theorems on, 6, 7, Initial point, of a vector, 121, Integers, 1, Integral equations, 113, 204, 206, Integral test, 7, 20, 21, Integrals, 5, 18-20, of functions of a complex variable, 287,, 297-300, line [see Line integrals], maximum or minimum of, 375, multiple [see Multiple integrals], Integrating factors, 40, 50, 51, depending on only one variable, 50, 51, Integration, 5, formulas, 5, 6, interchange of order of, 148, by parts, 5,19, by substitution, 6,19, Interval of convergence, 8, Inverse, of addition, 1, of logarithmic functions, 2, of a matrix, 346, 357, 358, 360, 361, of multiplication, 1, Inverse Fourier transform, 202, Inverse hyperbolic functions, 3, expressed as logarithms, 3,14, Inverse Laplace transforms, 99,100, complex inversion formula for, 324-341
Page 413 :
INDEX, Inverse Laplace transforms (cent.), elementary, 106, of functions with branch points, 325, 330-332, of functions with infinitely many, singularities, 332-334, theorems for finding, 111-113, uniqueness of, 100, Inverse operators, 75, 76, Inverse trigonometric functions, 3, Inversion, 292, Irrational number, 1, proof that -\/2 is an, 13, Isogonal, 291, Isolated singularity, 288, Isoperimetric problem, 376, 383, Iterated integrals, 147, 148, Jacobian, 127, 128, 142, for analytic function, 312, of a transformation, 128, used in transforming integrals, 149, Kilogram, 42, Kinetic energy, 377, 378, 387, Kirchhoff's laws, 42, 56, 57, 93, Kronecker's symbol, 185, Kutta-Runge method, 44, 64, 65, Lagrange multipliers, method of, 11, 26-28, proof of, 27, 28, used in calculus of variations, 376, Lagrange's equations, 378, 388, Lagrangian, 378, 388, Laguerre polynomials, 244, 245, 253, orthogonality of, 253, Laplace inversion formulas, 102, complex, 102, Laplace transform operator, 98, linearity of, 98, 105, Laplace transforms, 98-120, convolution theorem for, 102, definition of, 98, of elementary functions, 98,103-105, existence of, 98, 105, integration of, 101, inverse [see Inverse Laplace transforms], methods of solving partial differential, equations by, 260, 277, special theorems on, 101,102, 108-111, Laplace's equation, 259, 263, 287, Laplacian, 127, in cylindrical coordinates, 129,142, in orthogonal curvilinear coordinates, 129, Laurent's series, 289, 302, 303, Laurent's theorem, 319, Law of the mean, 8, Left and right hand limits, 183, Legendre functions, 242-251, [see also Legendre polynomials], associated [see Associated Legendre functions], of the second kind, 242, 243, 248, 249, Legendre polynomials, 242, 246, 247, [see also Legendre functions], , 403, , Legendre polynomials (cont.), generating function for, 242, orthogonality of, 243, 249, 250, recurrence formulas for, 242, series of, 243, 250, 251, solutions of partial differential equations, using, 275, 276, Legendre's differential equation, 242,243, general solution of, 242, Leibnitz's rule for differentiating an integral,, 11,28, Length or magnitude of a vector, 121, Lerch's theorem, 100, Less than, 1, Limits, 3, 4,14,15, definition of, 3, for functions of a complex variable,, 286, 294, 295, for functions of two or more variables, 8, theorems on, 4,15, of vector functions, 125, Linear dependence and independence, 72, 73,, 79,80, connection of with Wronskians, [see Wronskians], Linear differential equations, 40, 51, 71-97, 258, first order, 40, 51, general, 71, homogeneous, 40, 51, 52, 258, solution of by determinants, 89, Linear equations, 9,10, 25, 26, determinant of system of, 348, homogeneous, 347, inconsistent, 361, linearly dependent, 348, 362, matrix form of, 347, non-homogeneous, 347, solution of by Cramer's rule, [see Cramer's rule], systems of, 347, 359-362, trivial solutions of, 348, Line integrals, 149-152,160-162, connection of with complex integrals, 287, evaluation of, 150,151, Green's theorem and [see Green's theorem], independence of path of, 152, properties of, 151, vector notation for, 150, Logarithmic functions, 2, rules of, 13, Magnetic and eelctric field vectors, 145, Magnification factor, 291, Magnitude or length of a vector, 121, Mapping, conformal [see Conformal mapping], Mass, 42, 167, 259, Mathematical formulation of physical, problems, 42, Mathematical models, 42, Matrices, 342-374, addition of, 342, associative and distributive laws for, 343, 350, column, 342
Page 414 :
404, , Matrices (cent.), complex, 342, complex conjugates of, 344, conformable, 343, definition of, 342, determinants of [see Determinants], diagonal, 344, elements of, 342, equality of, 342, failure of commutative law for, 343, 350, Hermitian, 344, 349, 351, 365, 366, inverses of, 346, 357, 358, 360, 361, linear dependence of, 346, 356, multiplication of, 343, null or zero, 344, operations with, 350-352, as operators, 349, orthogonal, 346, 347, 358, 359, 365, 366, powers of, 343, principal or main diagonal of, 344, real, 342, 343, reduction of to diagonal form, 349, row, 342, singular, 346, skew-Hermitian, 344, 349, skew-symmetric, 343, 351, square, 342, subtraction of, 342, symmetric, 343, 351, trace of, 344, transform of, 349, transpose of, 343, unitary, 346, 347, 349, 358, 359, zero or null, 344, Maxima and minima, 11, 26-28, [see also Extremum], for integrals, 375, Lagrange multipliers for finding, 11, 26-28, relative, 11, Maxwell's equations, 145, Mean, 183, Mean square error, 194, 200, Mean-value theorem, 8, Mechanics, 42, 126, problems, 54-56, Member, 2, Membrane, vibrations of, 273, 274, 397, Meter, 42, M.K.S. system, 42, Modulus, of complex number, 12, 29, of elasticity, 259, 260, Young's, 43, Moebius strip, 169, Moment of inertia, 43, 155, 167, 260, polar, 155-157, Moments, 157, 167, bending, 43, Momentum, 42, Multiple integrals, 147-149,155-160, evaluation of, 155-157, transformation of, 158-160, Multiple-valued functions, 295, Multiply-connected regions, 164, , INDEX, Musical tone, 269, 274, Natural base of logarithms, 2, Natural frequency, 91, Natural logarithms, 2, Natural modes, of a vibrating membrane, 274, of a vibrating string, 269, Natural numbers, 1, Negative numbers, 1, Negative sense for traversing a curve, 151,162, Neumann function, 224, Newtons, 42, Newton's law of cooling, 60, Newton's law of mechanics, 42, 54, derivation of Hamilton's principle from,, 387, 388, Noise, 274, Non-analytic functions, 295, 296, Non-homogeneous equation, 71, 347, Non-linear differential equation, 71, Normalized functions, 185, Normal line to a surface, 139 Normal modes of vibration [see Natural modes], Normal plane to a space curve, 139,140, Normal vector to a surface, 126,138, unit, 137, Null vector, 122, Numerical methods, for integration, 19, 20, 23, for obtaining eigenvalues, 379, 390-392, for solving differential equations, 43, 44, 61-65, Odd extension, 192, Odd functions [see Even and odd functions], Ohms, 42, Operator, linear, 71, 78, Operator notation, 71, Operator polynomials in D, 71, Operator techniques for solving differential, equations, 75, 76, 83-86, Orthogonal curvilinear coordinates, 127, 128, Orthogonal functions, 184, 185, 197, 198, Orthogonal trajectories, 43, 58, 297, Orthogonality, of Bessel functions, 227,236, 237, of Hermite polynomials, 244, of Laguerre polynomials, 244, of Legendre polynomials, 243, Orthonormal series, 185, Orthonormal set of functions, 185,197, as solutions of Sturm-Liouville systems, 245, Orthonormal vectors, 359, Parabolic cylindrical coordinates, 145, Parallel vectors, condition for, 124, Parallelepiped, volume of, 124, 134, Parallelogram, area of, 124,133, Parallelogram law, for complex numbers, 35, for vectors, 121, Parameters, 39, 43, 47, 48, Parseval's identity, 184,193,194, 241, for Fourier transforms, 202, 203, 205, Partial derivatives, 8, 9, 23-25, Partial differential equations, 258-285, classification of, 261, definition of, 258
Page 415 :
405, , INDEX, Partial differential equations (cont.), elliptic, 258, 261, existence and uniqueness theorems for, 258, general solution of, 258, 262, 263, 277, homogeneous, 258, hyperbolic, 258, 261, linear, 258, 261, non-homogeneous, 258, non-linear, 258, 261, order of, 258, parabolic, 258, 261, particular solution of, 258, 262, singular solutions of, 258, solutions of, 258, 262-264, sone important, 259, 260, 263, 264, Partial fractions, 84, used in finding inverse Laplace transforms,, 102,113,114, Partial sums, 6, Particular solution, 39, 45, 46, 258, 262, 263, of a linear differential equation, 72-74, 260, Pendulum, 378, Period, 57,182, Periodic functions, 2, 182, Laplace transforms of, 101, Perpendicular vectors, condition for, 124, Picard's method, 44, 63, 64, Piecewise continuity, 99,105, Plane, equation of, 135, Planets, motion of, 389, Points, on a line, 1, Polar coordinates, 158, 168, Polar form of complex numbers, 12, 29, 30, Polar moment of inertia, 155, Poles, 288, 289, Polynomials, 2, Position vector, 123, Positive integers, 1, Positive normal, 154, Positive numbers, 1, Positive sense for traversing a curve, 151,162, Postulates, 2, Potential, 42, 377, 378, 387, equation, 259, function, 377, 378, of a hollow sphere, 276, 277, Potential energy, principle of minimum, 394, Power series, 8, uniform convergence of, 8, Primitive, 39, Principal branch, 295, Principal normal vector to a space curve,, 137, 144, Principal part of a Laurent series, 289, Principal value of logarithms, 295, Product, cross [see Cross product], dot [see Dot product], of matrices, 343, Quadratic equation, 29, Quadratic form, 349, 367, 368, 370, in n variables, 370, , Quadratic form (cont.), reduction of to canonical form, 349, 368, symmetric, 370, Quotient, 1, Radioactivity, 43, Radium, decay of, 60, Radius of curvature, 137,144, Radius of torsion, 144, Radius vector, 123, Ratio test, 7, Rational numbers, 1, Rayleigh-Ritz methods, 379, 390-392, Rayleigh's principle, 391, Real numbers, 1,13, Real part, of a complex number, 12, 286, Rectangular component vectors, 123, Rectangular coordinate system, right handed,, 122, Rectangular coordinates, 128, Rectified sine wave, 111, Recurrence or recursion formulas, 77, for Bessel functions, 225, 230, for gamma function, 210, for Hermite polynomials, 244, for Laguerre polynomials, 245, for Legendre polynomials, 248, for modified Bessel functions, 235, Reduced equation, 71, 80, 81, Reduction of order, method of, 75, 77, 81, Region of convergence of series, 7, Relative maxima and minima, 11, [see also Maxima and minima], Remainder, in Taylor series, 8, Removable singularity, 289, 303, Residue theorem, 289, 290, 304-306, evaluation of integrals by, 306-311, proof of, 304, 305, use of in finding inverse Laplace transforms,, 324, 328-338, Residues, 289, 304-306 [see also Residue Theorem], Resistance, 42, Resistor, 42, Resonance, 91, Resultant of vectors, 121, Riemann-Cauchy equations, [see Cauchy-Riemann equations], Riemann's mapping theorem, 291, 292, Riemann's theorem, 195, 207, Right and left hand limits, 183, Right handed coordinate system, 123, Rodrigue's formula, for Hermite polynomials,, 244, for Laguerre polynomials, 244, for Lagendre polynomials, 242, 247, Roots, of complex numbers, 12, of polynomial equations, 2, Rotation, 292, expressed in matrix form, 350, Rules of algebra, 1, Runge-Kutta method, 44, 64, 65, Saturated solutions, 60, Scalar field, 125
Page 416 :
406, , Scalar product [see Dot product], of column vectors, 347, Scalar triple product, 124, expressed as a determinant, 134, Scalars, 121, Scale factors, 128, 140, 141, Schwarz-Christoffel transformation, 293,, 312, 313, proof of, 312, 313, Semi-infinite solid, temperature in, 335, Separation of variables, for first order, differential equations, 39, 48, 49, for partial differential equations, 260, 266, 267, Sequence, 6, 7, 20, 21, definition of, 6, limit of, 6, 20, Series, 6, 7, absolutely convergent, [see Absolutely convergent series], complex, 301-304, infinite, 6, methods, 77, 88, 89, power [see Power series], Set, 1, Signs, rules of, 2, Simple closed curves, 151, Simple harmonic motion, 91, 92, Simple pole, 288, 303, Simply-connected regions, 151,164, Green's theorem for, 162, 163, Simpson's rule, 32, Simultaneous differential equations, 77, 89, Sine integral, 212, Singular points, 288, [see also Essential singularity], defined from Laurent series, 289, Singular solution, 39, 46, 59, 258, Slope, 4, 11, Soap films, 381, Space curve, 126, normal plane to, 139, 140, parametric equations for, 126, tangent line to, 139,140, Special functions, 212, 220, 221, Specific heat, 259, 263, Speed, 126, Spherical coordinates, 129, Spring, weight suspended from, 90, 91, 388, Stationary value, 375, Steady-state current, 56, 92, Steady-state heat flow, 263, Steady-state temperature, 259, 268, of a plate, 270, 271, of a semi-infinite slab, 317, Step by step or Euler method, 43, Stirling's asymptotic formula and series,, 211, 219, Stokes' theorem, 154, 172-176, proof of, 172, 173, Strain, 259, Stream function, 297, Stress, 259, Stretching, 292, , INDEX, Sturm-Liouville systems, 245, 254, 255, 378,, 379, 390-392, formulated in terms of calculus of variations,, 378, 379, orthogonality of solutions of, 255, reality of eigenvalues of, 254, 255, Submatrices, 370, Subset, 1, Subtraction, 1, Sum, of infinite series, 6, of matrices, 342, of vectors, 121, Superposition principle, 72,260, 267, Surface area, 167,168, Surface integrals, 153, 166, 167, minimum, 380, 381, Surfaces, 8, one sided, 169, two sided, 153, vectors perpendicular to, 138, Tangent line, 4, to a space curve, 139,140, Tangent plane to a surface, 139, Tangent vector, unit, 126, 137, 144, Taylor series, 8, 23, for functions of a complex variable, 288,302, for functions of two or more variables, 9, 25, method for solving differential equations,, 44, 63, 65, Taylor's theorem, 302, Temperature, 43, in a bar, 266, 267, 271, 272, in a circular plate, 272, 273, problem on, 60, in a solid, 259, 263, steady-state [see Steady-state temperature], Tension, 259, 264, Terminal point, of a vector, 121, Thermal conductivity, 259, 263, Thermodynamics, 145, Torsion, 144, Transformation equations, 127, Transformations, 6, general, 292, integration using, 6,158-160, involving complex variables, 310, involving matrices, 349, 350, of variables, 76, 77, Transient current, 56, 92, Translation, 292, Translation theorems, for Laplace transforms,, 101, Transverse vibrations, of a beam, 260, 336, 337, of a string, 259, 336, Trapezoidal rule, 20, 32, Trial solution, 73, 74, 81, 82, Triangle, area of, 134, Trigonometric functions, 2, 3, Triple integrals, 148, 156,157, in cylindrical coordinates, 160, in curvilinear coordinates, 159, transformation of, 159
Page 417 :
INDEX, Triple products, 124, 125, 134, 135, Trivial solution, 362, Undetermined coefficients method, 73, 74, 81, 82, for solving partial differential equations,, 265, 266, Uniform convergence of series, 7, 21-23, with complex terms, 301, connection of with Fourier series, 184,188,193, continuity and, 22, and termwise integration, 22, theorems on, 7, Weierstrass M test for, 7, 22, Uniqueness and existence, of Laplace, transforms, 100, of solutions of differential equations, 41, 258, Unit circle or disk, 291, Unit step function, 100,107,108, Laplace transform of, 100, 107, Unit vectors, 122, 185, rectangular, 122, 123, Units, 42, Value of a function, 2, Variable coefficients, differential equations with,, 71, 76, 77, Variation of parameters, 74, 77, 82, Variational notation, 376, 377, 383, 384, Vector [see also Vectors], algebra, 121,122, 130,131, analysis, 121-146, column, 342, field, 125, functions, 125, multiplication of by a scalar, 122, product, 124, row, 342, triple product, 124, , 407, , Vectors, 121, geometric interpretation for derivatives of,, 126, orthogonal, 346, Velocity, 42,126,137, of light, 145, along a space curve, 137, Velocity potential, 297, Vibrating spring, 90, 91, 388, Vibrating string equation, 259, 268, 269,, 336, 389, 390, derivation of, 264, 389, 390, Vibrations of a beam, longitudinal, 259, transverse, 260, Vibrations of a membrane, 273, 274, 397, Virtual displacement, 387, Voltage, 42, Volts, 42, Volume element, in cylindrical and spherical, coordinates, 129, in curvilinear coordinates, 128, Weierstrass M test, 7, 301, for integrals, 33, 208, Weight function, 185, Work, 150, 152,162, 176, 387, Wronskians, 72-74, 79, 80, of Bessel functions, 232, xy coordinate system, 2, Young's modulus, 43, Zero, 1, matrix, 344, vector, 122
