iQ JERROLD E. MARSDER A) ANTHONY J. TROMBs# PM ee ee NESE eh F EL: Ce ae TO ANE ATG 4 er are a AUREUS: = - RI y < Bete?DERIVATIVES dau du w Ty od diu + ») ae dws) 3 way ix 9. Ho ogy a7, PF oer wi 10. ates a) 2. 4 = 4 = —{sech uj(tanh u) # Py dee a q ty 5 i —(esch u)(éoth ») “ 12, 48H osu dcosu 5. ue 2 yi 6, 1H tan secu 1 £884 = (cot tose darcsin w 1 du 18,INTEGRALS (AW ARBITRARY CONSTANT MAY BE ADDED TO. \ EACH INTEGRAL) 1. forage (n#-1) a +1 a. [ fés ton foe 3 fe dxmet , 6 sting do a) 6 fea J, 8. [sin xdx = —cos x bs, mnet® | 8 [cos xdx = sin x f 7, fran xdx = —log feos x| Q L 8. fot xdx = log |sin x} Joec x dx = log sec x + tan x] = log ftan Gx +42) 10, [ose xdx = log fose x — cot x| = log jtan 43| arcsin = + Ja? =x fa>0) 12, [rtecosz dx - warscos -JF—F (a>o} 13. freon * dx = xarctan : = Stop (a? + 2%) {a>} “4. fs 15. foor mxdx = fim + sin mx cos mx) 1. fossa x ax= a mx dx = 5 (ma ~ sin mx cos mx) 16. sec? xdx = tan x a, fese? xdx = —cot x 18. fot as 19. (Continued at the back of the book)BASIC MULTIVARIABLE CALCULUS Tyuuers (ite (CSB pointe (998BASIC MULTIVARIABLE CALCULUS Jerrold E. Marsden University of California—Berkeley Anthony J. Tromba University of Munich and University of California—Santa Cruz Alan Weinstein University of Colifornia—Berkeley 6) Springer-Verlag LH W.H. Freeman and CompanyCover photograph courtesy of H. Armstrong Roberts, Inc., New York, NY. Text itlustracons from Veesor Caleulus used with permission of W.H. Freeman and Company. Figure 6.4.4 used wich permission of Cordon Arr B.V.. Baarn, The Nethedlands Library of Congres Marsden, Jeni E, Basic mubivariabie caeulus/Jereld E, Maden, Anthony |. Teoma, Alan Weinstein pcm. Indudes index ISBN 3-540.97976-X (Springer ISBN 0-7167-2443-X (WH. Freeman) 1. Calculus, 1. Tromba, Anchony. IL Weinstein, Alan, 1943: ualoging-in-Publication Data QA303.M336 1993 515—de20 92-38049 Printed on acid-fee paper. ©1993 Springer-Verlag New York, fc All rights reserved. This work may not be tensiared ar copied in whole or in pare without dhe written pet- ‘mission of the publisher (Springes-Verlag New York, Inc. 175 Fit Avenue, New Yor‘, NY 10010, USA}, except for bri excerpts in connection wich reviews or scholasly analysis. Use in connection with any form fof information storage and retrieval, dectcorie adaptation, computer sofewase, of by similar or distimlar _methodalogy now known or hereafer developed is forbidden, The ure of general descriptive names, rade names, trademarks, ec. ip this publication, even if the former are not especially identified, ie nat co be taken as a sign that such names, as understood by the Trade Marks and Merchandise Mass Act, may accordingly be used freely by anyone, ‘Unéer the eo-publishing agreement hecween Springer-Vecag and W.H. Freeman and Campany, the rex: is available in Noth America exclusively from W.H. Freeman and Company and ovrside North America ‘exclusively from Springer-Verlag. Phovocomposed copy from the authors les uting LaTeX. Printed and bound by Hemilion Printing Co., Reasselaes, NY. rinsed in the United Sraces of America 98765432 ISBN 3-540-97976-X Springer-Verlag Berlin Heidelberg New York ISBN 0-7167-2443-X W.H. Feeema and Company New York‘To Barbara, Inge, Margo, and the memory of Murray Weinstein.Preface ‘This text is intended for a one-semester sophomore-level course in the catculus of functions of several variables, including vector analysis. Such a course is sometimes preceded by a beginning course in linear algebra, but this is not an essential prerequisite. We use only the rudiments of matrix algebra, and the necessary concepts are developed in the text, We do assume a knowicdge of the fundamentals of one-variahle calculus—differentiation and integration of the standard functions. Computational skills and intuitive understanding are, for many students, more important than theory at this level, a need we have tried to meet by making the book as concrete and student-oriented as possible, We do this in two ways. First, we include a large number of physical illustrations from such areas as flu mechanics, gravitation, and electromagnetic theory, although prior knowledge of these subjects is not presumed. Second, we endeavor to present the material as siraighforwardly and simply as possible. For example, although we formulate the definition of the multidimensional derivative properly, we do so in terms of matrices of partial derivatives rather than abstract linear transformations. ‘This device alone can save one or two weeks of teaching time and avoids the necessity of a linear aigebra prerequisite. We have isolated many important items in boxes, which should help the student identify key concepts. A few of the really important results are in shaded boxes. We have also inciuded some historical and other notes for the student. This text is a synthesis of our books Vector Galewius (W.H. Freeman) and Cateutus IMF Springer-Verlag). The former gives a more thorough treatment ata higher level, whereas the latter is multivariable and vector calculus in the comtext of a standard calculus course,vill Preface The Student Guide and the Instructor's Guide The Role of the Computer Acknowledgments The student guide by K. Pao and F. Soon that accompanies this text is com- mercially available for student use, It contains solutions to every other odd- numbered problem, sample exams, and helpful hints for the student, The ne structor's guide is available to teachers from W.H. Freeman. It contains extra examples worked out for classroom use as well as additional sample exams. The computer is becoming essential as a tool for doing mathematics, including calculus and its applications. Computers are becoming cheaper, and software for doing symbolic catculations Ge, with formulas rather than numbers) and drawing graphs is becoming easier to use. The availability of this software may make you wonder why it is necessary to learn the theory of calculus and to do calculations and graphing “by hand.” There are several reasons for this. © In doing a problem by hand, you get more than the answer. You may learn something from the intermediate steps which gives you insight into the problem. * You need to know how to set up problems in a form the computer can accept. © Computer programs make mistakes. By knowing the theory, and having done numerous examples by hand, you are more likely to be able to recognize an unreasonable answer from the computer. © Problem-solving ability improves with practice. Furthermore, many prob- Jems require simple computations which are still easier to do by hand. © What you lear in doing problems by hand helps you solve other prob- lems. Many universities have computer labs that can provide very valuable supple- ments for vector calculus. We especially recommend the supplement “Labora tory Manual for Multivariable and Vector Calculus” by B. Felsager and B.H. West (Comell University), which was written in consultation with the authors of this book. Additional computer-related material to accompany this text is in preparation, We thank all the users of the previous books Vectar Calculus and Calculus JET on which this book is based, who took the trouble fo tell us of corrections and gave us suggestions for improvements. We especially thank Fred Soon, Karen Pao, Mike Hoffman, Frederick Hoffman, Martin Shim, Takashi Toriguchi, Kin Lee, Alex Baptista, Joanne Seitz, Asha Weinstein, Beverly West, Zhang-ju Lin, Stephen Miller, Kok-Wui Cheong, who assisted us in various aspects. We also thank the students in our classes who generously informed us of errors.Corrections Prerequisites and Notation Preface ix We are very grateful to Barbara Marsden for her superb preparation of the original BX file. Finally, we thank all the people at Springer-Verlag and W.H. Freeman who worked in such a dedicated way on all aspects of this project. Despite our effors and extensive class testing to eliminate errors from this hook, some will invariably remain. Please send your corrections and other re- marks to: Jerrold Marsden (email address is marsden@math_berkeley.edu) or Alan Weinstein (alanw@math.berkeley,edu) at the Department of Mathematics, University of California, Berkeley, CA 94720 or Anthony Tromba at the De- partment of Mathematics, University of California, Santa Cruz, CA 95064, or Mathematisches Institut, Theresien Strasse 39, 8000 Munchen 1, Germany. Jerrold E. Marsden Anthony J. Tromba Alan Weinstein ‘This section summarizes some concepis and notation used throughout the book, Students can read through it quickly now, then refer back later if the need arises. The collection of all real numbers is denoted R. Thus, includes the integers, that is, the collection ...,—3,-2,~-1,0,1,2, the rational numbers p/4, where p and q are integers (g # 0}; and the irrational numbers, such as V2, x, and ¢. When we write a € R, we mean that a is a member (or element) of the set R, in other words, that a is a real number. Members of B may be visualized as points on the real-numier line, as shown in Figure 0.0.1. Given two real numbers @ and 6 with a < b (i.e, with @ less than 8), we can form the closed interval [a,d], consisting of all such that a < 2 0 and —a if @ < 0, For example, [3] = 3,| - 3] = 3,|0| = 0, and (6] = 6. The ervangle “1 0 1 2 3 FIGURE 0.0.1. The geometric representation of points on the real-number line.x Preface a 6 ce ¢ e f closed open alt open FIGURE 0.0.2. The geometric representation of the intervals [a,b], (c,d), and {e, f). inequality states that the following always holds; |a + 6| < Jal + |b|. The distance from ato bis a — b|. Thus, the distance from 6 to 10 is 4 and from 610 3 isd For two collections of objects (ie, sets) A and B, A C B means that A is a subset of By that is, every member of A is also a member of B. For example, if B = RB, the set of integers {...,-3,—2,-1,0,1,2,3,...} is a subset of B, as is the set Q of rational numbers or all of R, but the set € of complex numbers is not. ‘The symbol AUB means the fon of A and B, the collection whose members are members of either A or B. Thus, t Similarly, AN B means the intersection of A and B, i.e, this set consists of those members of A and B that are in both A and B. The intersection of the two sets above is {-1, 0}. -8,—2,-1,0}U{-1,0,1,2,..J={.., , 2, -1,0, 1,2, We write A\B for the set of those members of A that are not in B. Thus, { —2,-1,0}\{-1,9,1,2,. 3} ‘We can also specify sets a5 in the following examples: {oER|aisan integer} = {. -1,0,1,2, {a€R|aisan even integer} = {...,—-2,0,2,4,...} {reRjasesd} = jad). A function f : A — B is a cule that assigns to each a € A one specific member f{a) of B. We call A the domain of f and B the target of f. The set {f(z) | 2 € A} consisting of all the values of f(x) is called the range of f. Denoted by f(A), the range is a subset of the target B. (It may be all of B, in which case f is said to be onto B.) The fact that the function f sends a to fla) is denoted by a + f(a), For example, the function f(z) = 2°/(1 - 2) that assigns the number «/(1— 2) to each 2 # 1 in R can also be defined by the rule 2 + 2/(1— 2). Functions are also called mappings, maps, or transformations. The notation f : A CR +R means that A is a subset of R and that f assigns a value f(z) in R to each x € A. The graph of f consists of all the points (x, f()) in the plane (Figure 0.0.3),Note FIGURE 0.0.3. The graph of a function with the half-open interval A as its domain, The notation Jy7Ly a; means ay +-+-4+aq, Where a1,..., aq are given numbers. For instance, the sum of the first » integers is _ mnt) ae 142teotns Pi The derivative of a function f(x) is denoted J"(x) or, if y = f(z), dy de Occasionally, we will commit what is cailed an “abuse of notation” and write the derivative as df/dz or y'(x). The definite integral of f from a to b is written f? f(z) dz. ‘We assume that the reader is famitiar with the functions used in one-variable calculus, such as sinz, cos, expx = e*, arcsing = sin—'z, and logz (we write toga for the naturat logarithm, which is sometimes denoted In « or log «). Students are expected to know, or to review as the course proceeds, the basic rules of differentiation and integration for these Functions of one variable, as well as the chain rule, the quotient rule, integration by pants, and so forth, Note especially the use of exp and log, For example: exp (2+ y) means et; log (3) means in (3x),Short tables of derivatives and integrals, adequate for the needs of this text, are printed at the front and back of the book along with other tables useful in mutivariable and vector calculus. ‘The end of a proof is denoted by the symbol ll, whereas the end of an example or remark is denoted by the symbol ¢. A reference such as §3.2 means Section 2 of Chapter 3.Contents Preface vii 1 Algebra and Geometry of Euclidean Space 1 Ll Vectors in the Plane and Space... eee eee eee _ 2 1.2. The Inner Product and Distance... 022 eee eee 2 13 2x 2and 3 x 3 Matrices and Determinants... 2-2... 39 1.4 The Cross Product and Planes 0 eee 46 15 n-Dimensional Euclidean Space 2.216 ee eee 60 16 Curves in the Plane and in Space. 660 o eee eee ee B Review Exercises. 00 00 eee eee 83 2. Differentiation 1 2.1 Graphs and Level Surfaces... 6c eee eee 92 2.2 Partial Derivatives and Continuity... 2-22... 200s eee 109 2.3 Differentiability, the Derivative Matrix, and Tangent Planes... 124 24 The ChainRule....:... eee eee 133 25 Gradients and Directional Derivatives 0.0 e eee 146 2.6 Implicit Differentiation ©. ee eee 160 Review Exercises 200 ee ee 166xiv Contents Higher Derivatives and Extrema 171 3.1 Higher Order Partial Derivatives ©... 0-2... e se eee 172, 52 Taylors Thetten qxawererey ¢ panes anne eee 182 33 Maximaand Minima 6... eee eee eee eee 190 3.4 Second Derivative Test... ee eee ee eee 201 3.5 Constrained Extrema and Lagrange Mulipliers ....-..... 21 RAVIEWEBRETCIBES ex cccccwiemcim om wesc enadesacens 221 ‘Vector-Valued Functions 227 diy snceeletatlintencsescwonaneuncenere OKRA HEN REAR 228 HB Kroon srcoroihSeodbdddheedS $8948 Sanase Byes 235 A9 NERC REE accumu owe ea nas saan 824 44 Divergence and Gul oe ee eee 249 Review Exercises 6. eee eee eee eee 263 Multiple Integrals 269 5.1 Volume and Cavalier’s Principle 6... eee eee ee 270 5.2. The Double Integral Over a Rectangle... . 0... e eee 280 5.3 The Double Integral Over Regions... 6... 21 54 Triple Imegrals 22.0... e eee ee 5.5 Change of Variables, Cylindrical and Spherica) Coordinates . . . 318 5.6 Applications of Multiple Integrals 6... 66. 339 REVS BERS wana yee ee RH 350 Integrals Over Curves and Surfaces 355 NG MOE BARE GRRLE ceceescornencnanerscasronueaamcs # Rane SO 356 6.2. Parametrized Surfaces . . . 374 63° Area of a Surface. - 382 64 Surface Integrals Review Exercises. 41dContents xv 7 The Integral Theorems of Vector Analysis 415 Wa Gremn's Theorem «eee eee eee es mn 416 7.2 Stokes’ Theorem... 006.0 429 73 Gauss’ Theorem 0 eee 446 74 Path Independence and the Fundamental Theorems of Calculus 458 Review Exercises oe eee eee eee eee eee 473 Epilogue 479 Practice Examination 1 481 Practice Examination 2 485 Answers to Odd-Numbered Exercises 489 Index 5211 Algebra and Geometry of Euclidean Space Corresponding to amy system of vector analysis is its practical unity. This was Gtbb's point of view in building up bis system. He used it in his courses on electricity and magnetism and on the electromagnetic theory of light. E.B. Wilson (from Vector Analysis, 1901, the first vector calculus text for students)2° 4. Algebra and Geometry of Euclidean Space 1.1 Vectors in the Plane and Space In this chapter we explain the basic operations performed on vectors in three- dimensional space: vector addition, scalar multiplication, and the dot and cross products. In §1.5 we generalize some of these notions to Euclidean n-space, and in §1.6 we study curves and their tangent vectors. Points P in the plane are represented by ordered pairs of real numbers (a1,@2); the numbers a1 and ag are called the Cartesian coordinates of P. We draw two perpendicular lines, label them as the x and y axes, and then drop per- pendiculars from P to these axes, as in Figure 1.1.1. After designating the inter- section of the x and y axes as the origin and choosing units on these axes, we produce two directed distances a; and az as shown in the figure; a; is called the © component of P, and az is called the y component, y FIGURE 1.1.1. Cartesian coordinates in the plane Points in space may be similarly represented as ordered triples of real numbers. To construct such a representation, we choose three mutually perpendicular lines that meet at a point in space. These lines are called the x axis, y axis, and z axis, and the point at which they meet is called the origin (this is our reference point). We choose a scale on these axes, as shown in Figure 1.1.2 The triple (0,0,0) corresponds to the origin of the coordinate system, and the arrows on the axes indicate the positive directions. For example, the triple (2,4,4) represents a point 2 units from the origin in the positive direction along the 2 axis, 4 units in the positive direction along the y axis, and 4 units in the positive direction along the z axis (Figure 1.1.3). Because we can associate points in space with ordered triples in this way, we often use the expression “the point (a;,42, a3)" instead of che longer phrase “the point P that corresponds to the triple (a;,a,03)." We say that a; is the coordinate (or first coordinate), a2 is the y coordinate (or second coordi- nate), and ag is the z coordinate (or third coordinate) of P. It is also common to denote points in space with the letters x,y, and z in place of a1, a3, and as. ‘Thus the triple (¢,y, 2) represents @ point whose first coordinate is , second coordinate is y, and third coordinate is 2.LA Vectors in the Plane and Space 3 1123 2 FIGURE 1.1.2. Cartesian coordinates in space. FIGURE 1.1.3. Geometric representation of the point (2, 4, 4) in Cartesian coordinates. ‘We employ the following notation for the line, the plane, and three-dimensional space. @ The real number line is denoted R* or simply R. GD The set of all ordered pairs (2, y) of real numbers is denoted R?, Gai) The set of all ordered triptes (x, y, 2) of real numbers is denoted R® ‘When speaking of R',R?, and R® simultaneously, we write R", where n = 1,2, or 3; or R™, where m = 1,2,3. Staning in §1.5 we will also study R* for n=4,5,6,..., but the cases n = 1,2,3 are closest to our geometric intuition and will be stressed throughout the book.4 1. Algebra and Geometry of Euclidean Space The operation of addition can be extended from R to R? and R*. For R®, this is done as follows. Given the two triples (a1, 42,43) and (by, b2, bs), we define their sum to be (1,023) + (61,52, bs) = (ar + 1,02 + ba, 03 + Ba). Example 1 {1,1,1) + (2,-3.4) = (8,-2,5), {z,y,2) + (0,0,0) (ty 2), (1.7.3) +4 (abc) = (L4+a,7+b,34c). + The element (0,0,0) is cailed the zero element (or just zero) of R*. The element (a1, —a2,—a9) is the additive inverse (or negative) of (a1,42,03), and we write (a1, a2, 03) ~ (61, ba, bs) for (a1, a2, 03) + (—b1, be, ~ba), ‘There are several important product operations that we will define on R8. One of these, called the fner product, assigns a real number to each pair of elements of R°. We shall discuss it in detail in §1.2. Another product operation for R° is called scalar multiplication (the word “scalar” is a synonym for "real number” ). This product combines scalars (real numbers) and elements of BR? (ordered triples) to yield elements of BR? as follows: given a scalar a and a triple (a1, a2, a3), we define the scalar multiple by (24,42, 43) = (a, aa2, aa). Example 2 Adel) = (2+4,2-e,2-1) = (8,2e,2), 6(1,1,1) = (6,6,6), Uujvw) = (u,9,w), O(p.a.7) = (0,0,0). As a consequence of their definitions, addition and scalar multiplics satisfy the following identities: @ = (@8}(a1, 62,03) = a[f(a1,42,43)] (associativity) Gi) (+ 8)(a1, a2,03) (distributivity) = (a1, 07,03) + Har, 02, 03) i) a[(ay, 02,43) + (01,2, ba)] Gdistributivity) = (ar, 02, 03) + ob, bas ba) (iv) @(0,0,0) = (0,0,0) (property of zero) (0,0,0) (property of zero) (2, a2, 43) (property of the unit element) @) O(a1,42,03) Gi) 11,42, 03)1.1 Vectors in the Plane and Space 5 The identities are proven directly from the definitions of addition and scalar mukiplication, For instance, (a+ B)(a1, 42,43) {{a + Bjar, (a + A)a2, (a+ B)as) {any + Bar, a2 + Baz, oas + fag) = o(ay,42,03) + A(a1,a2,43)- For R?, addition and scalar multiplication ave defined just as in R°, with the third component of each vector dropped off. All the properties above stil! hold. Let us turn to the geometry of these operations in IR? and R°. For the moment, we define a vector to be a directed line segment beginning at the origin, that is, a line segment with specified magnitude and direction, and initial point at the origin. Figure 1.14 shows several vectors, drawn as arrows beginning at the origin. In print, vectors are usually denoted by boldface letters: a. By hand, we usually write them as d or simply as a, possibly with a line or wavy line under it. FIGURE 1.1.4. Geometrically, vectors are thought of as arrows emanating from the origin. Using this definition of a vector, we may associate with each vector a the point (a1, 03,3) where a terminates, and conversely, with each point (a1, 42,43) in space we can associate a vector a. Thus, we shall identify a with (a1,22,43) and write a = (a1,02,03). For this reason, the elements of R? not only are ordered triples of real numbers, but are also regarded as vectors. The triple (0,0,0) is denoted 0, We call a1,a2, and ay the components of a, or when we think of a as a point, its coordinates. ‘Two vectors a = (a), a2, a3) and b = (b1,b2,63) are equal if and only if a) = 1,2 = be and a3 = bs. Geometrically this means that a and b have the same direction and the same length (or “magnitude” ). Geometrically, we define vector addition as follows. In the plane containing the vectors a = (a1,a2,a3) and b = (b,b2, bs) (see Figure 1.1.5), form the parallelogram having a as one side and b as its adjacent side. The sum a+6 1. Algebra and Geometry of Euclidean Space b is the directed line segment along the diagonal of the parallelogram. This geometric view of veetor addition is useful in many physical situations, as we shall see later. = a fey & FIGURE 1.1.5. The geometry of vector addition. To show that our geometric definition of addition is consistent with our alge- braic definition, we demonstrate that a+b = (a +4), a2-+b2, 43+). We shall prove this result in the plane and leave the proof in three-dimensional space to the reader, Thus, we wish to show that if a = (a1, a2) and b = (bj, 2), then a+b = (a; + bya +2) FIGURE 1.1.6. The construction used to prove that (ait) + (aabo) = (ax + Bs,02 +62). In Figure 1.1.6 let a = (a,a2) be the vector ending at the point A, and let b = (61, be) be the vector ending at point B. By definition, the vector a+b ends at the vertex C of parailelogram OBCA. To verify that a+b = (a1 +51, 42 + ba), it suffices to show that the coordinates of C are (ay + by, a + ba). The sides of the triangles OAD and BCG are parailel, and the sides OA and BC have ‘equal lengths, which we write as QA = BC. The triangles are congruent, 50 BG = OD; since BGFE is a rectangle, EF = BG. Furthermore, OD = a and OE = by. Hence, EF = BG = OD = a). Since OF = EF + OE, it follows that OF = 01 +4. This shows that the # coordinate of a+b is a1 +0). The proof that the ¥y coordinate is ay + be is analogous. This argument assumes A and B to be in the first quadrant, but similar arguments hold for the other quadrants.LI Vectors in the Plane and Space 7 Figure 1.1.7(a) illustrates another way of looking at vector addition; in terms of triangles rather than parallelograms. That is, we translate (without rotation) the directed line segment representing the vector b so that it begins at the end of the vector a. The endpoint of the resulting directed segment is the endpoint of the vector a+b. We note that when a and b are coilinear, the triangle collapses to a line segment, as in Figure t.1.7¢b). y prams @ ro) FIGURE 1.1.7. (a) Vector adcition may be visualized in terms of triangles as well as par allelograms. (b) The triangle collapses to a kine segment when a and Ib are collinear. In Figure 1.1.7 we have placed a and b head to tail. That is, the tail of bis placed at the head of a, and the vector a+b goes from the tail of a to the head of b. If we do it in the other order, b+ a, we get the same vector by going around the parallelogram the other way. This figure teaches us something useful — it is a good idea to let vectors “glide” or “slide,” keeping the same magnitude and direction, We want, in fact, to regard ewo vectors as the same if they have the same magnitude and direction, When we insist on vectors beginning at the origin, we will say that we have bound vectors. If we allow vectors to begin at other points, we will speak ol free vectors or just vectors. Vectors Vectors (also called free vectors) are directed line segments in {the plane or] space represented by directed line seaments with a beginning (tail) and an end (head). Line segments obtained from each other by translation (but not rotation) represent the same vector. The components (a1, @2,43) of a are the (signed) lengths of the projec- Gons of a along the three coordinate axes; equivalently, they are defined by placing the tail of a at the origin and letting the head be the point (01,42, a3). We write a = (a1, @2, 03). Two vectors are added by placing them head to tail and drawing the ‘vector from the tail of the first ta the head of the second, as in Figure 1.1,7.8 1. Algebra and Geometry of Euclidean Space Example 3 FIGURE 1.1.8, Some scalar multiples of a vector a. Scalar multiplication of vectors also has a geometric interpretation. If a is a scalar and aa vector, we define aa to be the vector that is {al times as long as a, with the same direction as a if @ > 0, but with the opposite direction if a < 0. Figure 1.1.8 illustrates several examples. Using an argument based on similar triangles, one finds that if a= (a, 42,43), aa = (2a;, 002,043). ‘that is, the geometric definition coincides with the algebraic one. Given two vectors a and b, how do we represent the vector b~a geometrically, ie, what is the geometry of vector subtraction? Since a+ (b—a) =b, we see that b — a is the vector that one adds to a to get b. In view of this, we may conclude that b — a is the vector parallel to, and with the same magnitude as, the directed line segment beginning at the endpoint of a and terminating at the endpoint of b when a and b begin at the same point (see Figure 1.1.9}. In Figure 1.1.10, which vector is (a) w+ _v? (b) 3u? jwExample 4 1.1 Vectors in the Plane and Space 9 FIGURE 1.1.10, Find w+ v,3u and —v. Solution, (@) To constrict u+v, we represent u and ¥ by directed line segments 80 that the head of the first coincides with the tail of the second, We fill in the third side of the triangle to obtain wu + v (see Figure 1.1.11. Comparing with Figure 1.1.10 we find that u+¥ = w. (b) 3u = q (see Figure 1.1.12). (© —v=(-l)v= re (see Figure 1.1.12). @ Let v be the vector with components (3,2, —2) and let w be the vector from the point (2,1,8) to the point (~1,0,—1). Find v+-w, Mlustrate with a sketch. (-3,-1,-4), (0,1, —6), as Solution Since w has components (—1,0,-1) — (2,1,3) -we find that v + w has components (3,2, —2) + (-3,-1,—4) illustrated in Figure 1.1.13. @ To describe vectors in space, it is convenient to introduce three special vectors along the x,y, and z axes:10 1. Algebra and Geometry of Euclidean Space oo ym FIGURE 1.1.11. The geometric construction of u + v. FIGURE 1.1.12, To find 3u, draw a vector in the same direction as u, three times as long: —Viis a vector having the same length as v, pointing in the opposite direction i: the vector with components (1,0,0) fs the vector with components (0, 1,0) k: the vector with components (0, 0,1). ‘These standard basts vectors are illustrated in Figure 1.1.14, In the plane ‘one has, analogously, 4 and j with components (1,0) and (0, 1). Let a be any vector, and let (a1, a2, a3) be its components. Then a= a+ agj+agk since the right-hand side is given in components by ,(1, 0,0) +.22(0, 1,0) +a5(0,0, 1) (21,0,0) + (0,a2,0) + (0,0,¢3) = (a:,02,¢3). ‘Thus we can express every vector as a sum of scalar multiples of i,j, and k. The Standard Basis Vectors 1. The vectors i,j, and kc are unit vectors along the three coordinate axes, as shown in Figure 1.1.14. 2. If a has components (a1,42,a3), then axl tag) + ask,1.1 Vectors in the Plane and Space 11 FIGURE 1.1.13. Adding v = (3, 2, ~2) tow, the vecter from (2, 1, 3} t0 (-1, 0, Toa.o om, FIGURE 1.1.14. The standard basis vectors, ‘We have used the term “standard basis" because it is commonly used in linear algebra. For the moment, do not worry abour the precise meaning of the tere “basis.” The vector (2,3,2) is 21+ 3 + 2k, and the vector (0,-1.4) is —j + 4k. Fig- ure 1.1.15 shows 21+-3+2k; the student shouid drawy in the vector -j+4k. Addition and scalar mukiplication may be written in terms of the standard basis vectors as follows: (ari + aaj + ask) + (1+ baj + O3k) = (ay + by)i+ (a2 + bo)j + (aa + bak and (ari + a9j + ask) = (aay)i + (cag)j + (oas)k.12 1, Algebra and Geometry of Euclidean Space Example 6 FIGURE 1.1.15. Representation of (2,3, 2) in terms of the standard basis vectors, i,j, and k. To apply vectors to geometric problems, it is useful 10 assign a vector to a pair of points in the plane or in space, as follows. Given two points P and P/ we can draw the vector ¥ with tail P and head P’, as in Figure 1.1.16, where we write = PP’ fos v. Et F — FIGURE 1.1.16, The vector from P to P’ is denoted PP WP = (x,y,z) and P = (2',y',2’), then the vectors from the origin to P and Pi area =clt+y)+zkand al = 2/i+y'j+2'k, respectively, so the vector PP’ is the difference a! —a= (2! —x)i-+ (y/— yj + (2’ — 2)k. Gee Figure 1.1.17.) The Vector Joining Two Points If the point P has coordinates (4,2) and P’ has coordinates (2/,y/',2'), then the vector PP’ has components (2/ — 2, y! — y, 2" ~ z). (a) Find the components of the vector from (3,5) to (4,7). (b) Add the vector v from (—1,0) to (2,—8) and the vector w from (2,0) to (1,1). (©) Multiply the vector v in (b) by 8. If this vector is represented by the directed line segment from (5,6) to Q, what is Q?1,1 Vectors in the Plane and Space 13 FIGURE 1.1.17, PP’ =OP’ — OP. Solution (@) As in the preceding box, we subtract the ordered pairs (4,7) — (3,5) = (1,2). Thus the required components are (1,2). (b) The vector v has components (2,—8) - (-1,0) = (3-8) and w has components (1,1) — (2,0) = (-1,1). Therefore, the vector ¥ +w has components (3, —3) + (-1,1) = (2,-2) (©) The vector 8v has components 8{3,—3) = (24, —24). If this vector is represented by the directed line segment from (5,6) to Q, and Q has coordinates (x,y), then (x, y) — (5,6) = (24,24), so (2, y) = (5,6) + (24, -24) = (29,-18). Many of the theorems of plane geometry can be proved by vector methods. Here is one example. Example 7 Use vectors to prove that the diagonals of a parallelogram bisect each other. Solution Let OPRQ be the parallelogram, with two adjacent sides repre- sented by the vectors a = OP and b =OQ. Let M be the midpoint of the diagonal OR, N the midpoint of the other, PQ. (See Figure 1.1.18.) FIGURE 1.1.18, The midpoints M and N coincide, so the diagonals OR and PQ bisect each other.14 1. Algebra and Geometry of Euctidean Space = a+b, by the parallelogram mule for vector addition, PG =09 - OF 0 is a number, then there is a triangle with sides parallel to those of PQR and side lengths b times those of PQR. 36. The medians of a triangle intersect at a point, and this point divides each median in a ratio of 2:1, Problems 37 and 38 require some knowledge of chemical notation, 37. Write the chemical equation CO + HaO = Hy + CO2 as an equation in ordered triples (21,72, Z3) where 21,22,23 are the number of carbon, hydrogen, and oxygen atoms, respectively, in each molecule. 38.€a) Write the chemical equation pC3H4O3 + gO2 = rCOz + sH20 as an ‘equation in ordered triples with unknown coefficients p, 9,7, and 8. (b) Find the smallest positive integer solution for p,q,7, and s.22 1, Algebra and Geometry of Euclidean Space 1.2 The Inner Product and Distance Example 1 In this chapter we discuss two products of vectors: the inner product and the cross product. These products are useful in physical applications and have interesting geometric imerpretations. ‘We begin in this section with the inner product (the names dot product and scalar product ase often used instead), giving an algebraic definition of the inner product and then showing how it is related to the geometric concepts of length and angle. Let a = ard + aaj + a3k and b = byi + ba} + dgk. We define the inner product of a and b, written a-b, to be the real number ab = aiby + aabe + ashy. Note that the inner product of two vectors is a scalar. (a) Ia = 3i4+j— 2k and b=i—j+k, calculate a-b, (b) Calculate (21+j — k) - (3k — J). Solution (@) a-b=3-141- (1) 4 (-2)-1=3-1-2=0 (b) (21-44 —k) - (8k — 2) = (28+ J —k) - (01 2) + 3k) =2-0-1-2-1-3=-5. @ The inner product has several important algebraic properties, If a,b and ¢ are vectors in R? and a and fare real numbers, then @ a-a>0;a-a=0 if and only if a= (i) (aa) -b = afa-b) and a- fb = fia-b); Gil) a-(b+¢)=a-bta-c and (at+b)-c=a-e+b-c; and (iv) a-b=b-a. To prove the first of these properties, observe that if a= a1i + a2j + agk, then a-a= af +a} +3. Since @1, a2, and ag are real numbers, we know that a?,a3 and a2 are non-negative; hence, so is their sum. Moreover, if a? +a} + a3 = 0, then @ = a2 = a3 = 0; therefore a = 0 (zero vector). The proofs of the other properties of the inner product are also easily obtained from the formula that defines it, It follows from the Pythagorean theorem that the length of the vector a = aul + aoj + agk is \/ay + a} + a3 (see Figure 1.2.1). This length is denoted by all and is sometimes called the norm of a. Since a-a = a? +03+a3, it follows that llal| = @-ajtExample 2 1.2 The Inner Product and Distance 23 ‘Vectors with norm 1 are called unit vectors. For example, the basis vectors 4,j,k discussed in §1-1 are unit vectors. Observe that for any nonzero vector a, the vector a/|lal| is a unit vector. When we divide a by all, we say that we have normalized a. x Naf ai FIGURE 1.2.1, The length of the vector a = (a1,a2,4) is given by the ?ythagorean formula: |lal| = (a? + a3 + a3)2 (a) Normalize v = 21+ 3j— $i. () Find unit vectors a,b, and ¢ in the plane such that b+ ¢ =a, Solution (a) We have |[vl] = 2? +3? +(/2)? = (1/2) V53, so the normalization of vis 1 4 6 1 Wi" yaa’ al (b) Since all three vectors are to have length 1, a triangle with sides a,b, and © must be equilateral as in Figure 1.2.2. Orienting the triangle as in the figure, we take a = i, then b = $1+(¥3/2)j, and ¢ = hi—(8/2)j. You should check that llal| = jbl] — ell =1andthab+e-a ¢ 1 MB Be FIGURE 1.2.2, The vectors a,b, and ¢ are represented by the sides of an equilateral trangle. If a and b are vectors based at the same point, we have seen that the vector b ~ ais parallel to and has the same magnitude as the directed line segment24 1, Algebra and Geometry of Euclidean Space from the endpoint of ato the endpoint of b. it follows that the distance from the endpoint P of a to the endpoint Q of b is [ball Cor ]PQ|) (see Figure 1.2.3). x FIGURE 1.2.3. The distance between the tips of a and b is Ib — al Inner Product, Length, and Distance Letting a = aii + ao] + ask and b = bii+ boj + bak, a-b = ayby + agby + agb3, lal = fot +03 + a3. To normalize a vector a, form the vector a er The distance between the endpoints of a and b is ja — bl|, and the distance between P and Qs ||PQ||. Example 3 Find the distance from the endpoint of the vector i to the endpoint of the vector i Solution Wi- i] = (O17 40-07 +0 0PF = v2. @ Example 4 Let P; = ¢(1,1,1) (a) What is the distance from P; to (3,0,0)? (b) For what value of t is the distance shortest?1.2 The Inner Product and Distance 25, ©) What is the shortest distance? Solution (a) By the distance formula, the distance is vit—3 + (t- 0? +(-0? = Ve—t+o+ eae = V3 648. () The distance is shortest when its square, namely 3¢” — 6¢ + 9, is least, that is, when (d/dt}(3#? — 6 + 9} = 6 -6=0, ort = 1, (c) For t= 1, the distance in (a) is VB. ‘Trigonomeuy telis us that we can find the angles of a triangle if we know the lengths of its sicles. Since lengths are given by the dot product, we should also be able to use the dot product to compute angles, The relationship is given as follows. Angles and the Inner Product Let a and b be two vectors in R3 and let @ be the angle between them, where 0 < @ < 7m (see Figure 1.2.4). Then a-b= |al||[bl| cose. Thus we may express the angle between nonzero vectors a and b as eo (a): FIGURE 1.2.4. The vectors a and b and the angle @ between them,6 1. Algebra and Geometry of Euclidean Space To establish this, we recall the law of cosines from trigonometry. Tt states that for the triangle with adjacent sides determined by the vectors a and b, we have {|b — al? = [jal]? + |[b||? — 2 fal - Ib] cos 8. Since [[b — all? = (b—a) - (b —a), jjalj? = aa, and bl? = b+ b, we can rewrite the preceding equation as (b—a)-(b—a)=a-a+b-b—2|[al] [bl] cos. By the distributive law for the dot product, (b—a)-(b-a) = b-(b-a)—a:(b—a) b-b-b-a-a-hya-a = a-atb-b-2a-b. ‘Thus, a-atb-b—2a-b=a-atb-b—2lfall |b|| cosd. That is, a'b= |jal||[bl| cosé, proving our assertion. Example 5 _ Find the angie between the vectors ij +k and i+ jk (see Figure 1.2.5). ' ‘ isj-\! FIGURE 1.2.5. Finding the ang'e between a =i+j+kandb=i+j—k Solution Using the preceding box, we have (4 p+k)-(+j—k) = |li+j+kllli+j—kllcose -Example 6 Example 7 1.2 The Inner Product and Distance 27 and so 1+1-1 = (73)(73) cos, Hence, cos @ = 1/3. That is, @ = cos ~*(1/3) 1,23 radians (= 71°). © Notice that if a and b are nonzero, then a-b = 0 if and only if cas@ = 0. Hence we get: Perpendicular Vectors The inner product of two nonzero vectors is zero if'and only if the vectors are perpendicular. Often we say that perpendicular vectors are orthogo- nat. ‘The standard basis vectors i,j, and k are mutually orthogonal and of length 1, any such system is called orthonormal. The vectors ig = (cos #)1 + (sin 8)j and jp = —( sin 8)i+ (cos 6)j are orthonor- mal, since lg je = —cos@sind + sincos6 =0, \jloll = Vcos?6-+ sin?0 = 1, Wall = sind cos = (see Figure 1.2.6). @ FIGURE 1.2.6. The vectors ig and jg are orthogonal. Find a unit vector in the zy plane that is orthogonal to v = i— 3}. Solution If w = ai+ }j is perpendicular to i— 3), then 0 = v-w = a—3b so @ = 3b. A solution is 34 +}, but this is not a unit vector. Dividing by the length Vi'+1? = V0, we find the solution w = (31+ §)/V10. Another solution is — (314 j)/V10 Gee Figure 1.27). @28 1. Algebra and Geometry of Euclidern Space Example 8 ai + i) hate) vio ind FIGURE 1.2.7. There are two unit vectors orthogonal to 4 — 3j ‘The inner product of two vectors is the product of their lengths times the cosine of the angle between them. This refalionship is often of value in geometric problems. [1 also leads to the following important inequality, Cauchy—Schwarz Inequality For any two vectors a and b, ja-by < [all [bi Equality holds if and only if either a is a scalar multiple of b, or one of a or bis 0. To prove this, we use the formula a-b = [all |bl|cos@ to give Ja- bl = Hall bt|| cos] < jal] [ll (since ~1 < cos < 1). Tf either a or bis zero, then. both sides are zero. Otherwise, equality holds in |a-b] < |[al{||b|| exactly when [cos] = 1; te, @ = 0 or x, which means that a is a (positive or negative) scalar multiple of b. Verify the Cauchy—Schwarz. inequality for a i+j+kandb=3itk. Solution The dot product is a-b = —3 +0 +1 = 250 |a-b| = 2. Also, \lall = VIF 141 = V3 and [bl] = YOF1 = V0, and it is true that 2 V3-V3=322 © The inner product heips us compute the projection of a vector in a given direction. If v is a vector, and Lis the line through the origin in the direction of a vector a, then the orthogonal projection of von a is the vector whose tip is obtained by dropping a perpendicular tine to t from the tip of v, as in Figure 1.2.8.Example 9 1.2 The Inner Product and Distance 29. FIGURE 1.2.8. pis the orthogonal projection of von a. Referring to the figure, we see that p is a multiple of a and that v is the sum of p and a vector that is perpendicular to a, Thus we can write the equation veata where p = ca and a-q = 0. Taking the dot product of a with both sides of v= catg, we find a-v = ca-a, so ¢= (a: y)/(a-a), and hence Lay, Pe fale ‘The length of p is a-vl lat| ipl = Tare |! =a = Wileosé- Orthogonal Projection ‘The orthogonal projection of v on a is the vector ay p= a. Find the orthogonal projection of i+j on 1 3 Solution Witha= ~ 3j and v = 14}, the orthogonal projection of v on a is (see Figure 1.29. A useful consequence of the Cauchy—Schwarz inequality, called the triangle inequality, relates the lengths of vectors a and b and of their sum a+b,301, Algebra and Geometry of Euclidean Space FIGURE 1.2.10. This geometry shows that [OQ] < ||OR|| + {[RQl]. oF in vector nota- ‘tion, that lla + bl] < ||all + [Ib||, which is the triangle inequality. Geometrically, it says that the length of any side of a triangle is no greater than the sum of the lengths of the other two (see Figure 1.2.10) Triangle Inequality For vectors a and b in space, lla+bj] < lal +b. To demonstrate the triangle inequality, we consider the square of the left-hand, side: {a +bl? =(a+b)- (a+b) = fall? +2a- b+ |ibI2. By the Cauchy~Schwarz inequality, we have [fal]? + 2a-b + [fb]? << ffall? + 2 lal bl] + [BI = (tall + Ibl)?. Thus, \la+b]? < (lal + Ib):Example 10 Example 11 1.2 The Inner Product and Distance 31 taking square roots proves the triangle inequality. (2) Verify the triangle inequality for a @) Prove that ju — vf] < lu —wil + [lw — vl] for any vectors u,¥, and Ww. Illustrate with a figure in which u,v, and w are based at the same base point. ind b= 245 +k Solution (a) We have a + b = 31+ 2+, so la + bl] = VOT 441 = V4. On the other hand, |lal| = v2 and |[bli = V6, so the triangle inequality asserts that Vid < v2 + V6. The numbers bear us out: V1 = 3.74, while V2 + V6 = 141+ 2.45 = 3.86. (b) We find that u—v = (u — w) + (w—v}, so the result follows from the triangle inequality with a replaced by u — w and b replaced by w — v. Geometrically, we are considering the shaded triangle in Figure 1.2.11. FIGURE 1.2.11. Illustrating the inequality || — vl] < [ju —wl| + ||w— ¥\ ‘We now turn to some physical applications of vectors, beginning with an appli- cation to navigation. Suppose that, on a part of the earth’s surface smatl enough to be considered flat, we introduce coordinates so that the z axis points east, the y axis points north, and the unit of length is the kilometer. If we are at a point P and wish (o get to a point Q, the displacement vector d = PQ joining P to Q tells us the direction and distance we have to travel. If «and y are the components of this vector, the displacement of P to Q is “x kilometers east, y kilometers north” Suppose that two navigators who cannot see each other but can communicate by radio wish to determine the relative position of their ships. Explain how they can do this if they can determine their displacement vectors to the same lighthouse. Solution Let Py and P» be the positions of the ships, and let Q be the position of the lighthouse. The displacement of the lighthouse froma the ith ship is the vector d; joining P; to Q. The displacement of the second ship from the first32 1. Algebra and Geometry of Euclidean Space is the vector d joining P; to Pz. We have d + dz = dy (Figure 1.2.12), and so d = d} — dy. That is, the displacement from one ship to the other is the difference berween the displacements from the ships to the lighthouse. FIGURE 1.2.12. Vector methods can be used to locate objects. ‘We can also represent the velocity of a moving object as a vector. For the mo- ment, we will consider only objects maving at uniform speed along straight lines. Suppose, for example, that a boat is steaming across a lake at 10 kilo- meters per hour (km/h) in the northeast direction. After 1 hour of travel, the displacement is (10/72, 10/73) ~ (7.07, 7.07); see Figure 1.2.13. postion after S ai ‘ ‘ t a if nits position FIGURE 1.2.13. If an abject moves northeast at 10 kirvh, its velocity vector has compo- nents (7.07, 7.07). The vector whose components are (10/V2, 10/2) is called the velocity vector of the boat. In general, if an object is moving uniformly along a straight line, its velocity vector is the displacement vector from the position at any moment Jo the position 1 unit of time later. Displacement and Velocity If an object has a (constant) velocity vector ¥, then in ¢ units of time the resulting displacement vector of the object is d= tv; see Figure 1.2.14, Returning to our boat on the lake, note that if a current appears, moving due eastward at 2 km/h, and the boat continues to point in the same direction with its engine running at the same rate, its displacement after 1 hour will have1.2 The Inner Product and Distance 33 displacement in ume £ FIGURE 1.2.14. Displacement = time x velocity, components given by (10/v2+ 2, 10/2}; see Figure 1.2.15, The new velocity vector, therefore, has components (10/7 + 2, 10/2). We note that this is the sum of the original velocity vector (10/2, 10/ V2) of the boat and the velocity vector (2,0) of the current. displacement due to current displacement, due to engine, ‘otal displacement FIGURE 1.2.15. The totai displacement is the sum of the displacements due to the engine and the current. Similarly, consider a seagull that flies in calm air with velocity vector v. If a wind comes up with velocity w and the seagull continues flying the same way, its actual velocity will be v + w. One can see the direction of the vector ¥ because it points along the axis of the seagull. By comparing the direction of actual motion with the direction of v, you can get an idea of te wind direction (see Figure 1.2.16). Sy Bow FIGURE 1.2.16, The velocity w of the wind can be estimated by comparing the “wingflap" velocity v with the actual velocity v + -w. Another example comes from medicine. An electrocardiograph detects the flow of electricity in the heart; both its magnitude and its direction are important. The net flow can be represented at every instant by a vector called the cardiac34 1. Algebra and Geometry of Euclidean Space tector. The motion of this vector (see Figure 1,217) gives physicians useful information about the heart's function. Tip of ‘cardiac vector moving in space Coraline vector at one moment FIGURE 1.2.17. The magnitude and direction of electrical flow in the heart are indicated by the cardiac vector. Example 12 4 bird is flying in a straight line with velocity vector 101+ 6j +k (in kilometers per hour). Suppose that (zy) are its coordinates on the ground and z is its height above the ground. (a) If the bird is at position (1,2,3) at a certain moment, where is it 1 hour later? 1 minute later? (b) How many seconds does it take the bird to climb 10 meters? Solution. {a} The displacement yector from (1,2,3} after 1 hour is 101+ 6j + ks, so the new position is (1,2, 3) + (10,6,1) = (11,8,4). After 1 minute, the displacement vector from (1, 2,3) is Dai. 4 1,1 Bot 4K) = Git I+ Gok and so the new position is 111 7 21 181 0.2.9)+ (Boras) > (Faery (b) Aftert seconds (= #/3600 hours), the displacement vector from (1, 2,3) is (t/3600}(101 + 6) + k) = (¢/360}i + (¢/600)j + (¢/3600)k. The in- crease in altitude is the z-component ¢/3600. This will equal 10 m (= (1/100)km) when £/3600 = 1/100, that is, when t = 36 seconds, ¢ Example 13 Physical forces have magnitude and direction and may thus be represented by vectors. If several forces act at once on an object, the resultant force is represented by the sum of the individual force vectors. Suppose that forcesExercises for §1.2 1.2 The Inner Product and Distance 35 i+ and j +k are acting on a body. What third force must we impose to counteract the wo, i.e, 16 make the total force equal to zero? Solution The force F should be chosen so that (1+ k) + (j +k) +F = 0; that is, F= -(i+) — G+) = -1—j — 2k. (Recall that 0 is the zero tector, the vector whose components are all zero.) Real-World Problems vs. Made-Up Problems You have probably wondered, afier taking a year of one-variable catculus, and possibly other mathematics courses, how realistic are the problems and examples? You may have asked: “Will I really be able to solve some significant problems in physics, engineering, ecology, of biology just a5 easily?” ‘The answer to this question is somewhat negative, but these is a positive side too. The negative news is that truly realistic problems are normally (but not always) more complex than the ones presented in elementary texts, and one cannot solve them without seriously studying how the problem is modelled and bringing to bear more knowledge of the other field and of mathematics. To even understand such problems properly and to proceed 10 reliabie solutions usually requires more time than one has available in the classroom. For example, consider the problem “How much solar en- ergy is received in a day at a point on the earth Ggnoring cloud cover) as a function of the day of the year and the latitude?” This is one of the “cleanest” but still realistic problems that we know of, yet it requires ak most all the tools of calculus and muhivariable calculus to solve, and it takes some time to understand. (The solution to this probiem is given in the instructor's guide for this book — your instructor can provide it to you if you are interested.) Other sources, such as the UMAP Modules: Tools for Teaching, published by the Undergraduate Mathematics and Its Appli- cations project, contain extended applications. On the positive side, knowing nuultivariable and vector calculus well and the “made-up” or simplified types of problems presented, can form a basis of knowledge, upon which one can build, eventually enabling one to solve the often complex physical problems presented by the natural world In Exercises 1—4, compute [jul], ||vij, and u--v for the given vectors in R3, ‘Lo w= Li 34 dew = 14 3j—k 2. u=Y-ive 3. w= 2+ 10j— 12k,v = 31+ 4k36. Algebra and Geometry of Euclidean Space 4, w= V3i+ aj + ck, v = 4i—j —k, where c is a constant. Normalize the vectors given for Exercises S—8. 5. The vector u in Exercise 1 6, The vector v in Exercise 1 7. The vector w in Exercise 2 8, The vector v in Exercise 2 In Exercises 9-12, find the angle between the given vectors. If necessary, express your answer in terms of cos, 9. The vectors in Exercise 1 10. The vectors in Exercise 2 11, The vectors in Exercise 3 12. The vectors in Exercise 4 In Exercises 13-16, verify the Cauchy—Schwarz inequality for the given pair of vectors, 13. The vectors in Exercise + 14, The vectors in Exercise 2 15. The vectors in Exercise 3 16. The vectors in Exercise 4 17. Find two nonparallel vectors, both orthogonal to (1, 1,1). 18, What restrictions must be made on b sa that the vector 24+-bj is orthogonal 0 @) -314+94+Kk OK 19. Tell without calculating whether |[8i— 12k -]6j-+4|| — (8i— 12k) -(6j +1) is equal to zero. Explain. 20. Find the line through (3, 1,—2) that intersects and is perpendicular to the line = ly = —2+4,2 =—1+¢. [Hint: If (co, yp, 20) is the point of intersection, find its coordinates.] In Exercises 21-24, find the orthogonal projection of u onto ¥.1.2 The Inner Product and Distance 37, 2u= itjtkv=2 22, w= 214+ j—3k,v 23, w= 34 4j-Sk,v 24. wi t+j+kv=3i+4j—5k 25. Use the formula (i-+j +k) -4= 1 to find the angle between the diagonal of a cube and one of its edges. Sketch. 26.2) If [ful] = |v], and wand w are not parallel, show that u+v and w—¥ are perpendicular. (b) Use the result of part (a) to prove that any triangle inscribed ina circle, with one side of the triangle as a diameter, is a right triangle. 27. A ship at position (1,0) on a nautical chant (with nosh in the positive y-direction) sights a rock at position (2,4). What is the vector joining the ship to the rock? What angie does the vector make with due north? (This is called the Bearing of the rock from the ship.) 28, Suppose that the ship in Exercise 27 is pointing due north and sraveling ata speed of 4 knots relative to the water, There is a current flowing due east at 1 knot, The units of the chart are nautical miles; 1 knot = 1 nautical mile per hour. (a) If there were no current, what vector w would represent the velocity of the ship relative to the sea bottom? {b) If the ship were just drifiing with the current, what vector v would represent its velocity relative to the sea bottom? (©) What vector w represents the total velocity of the ship? (d) Where would the ship be after 1 hous? (e) Should the captain change course? (D) What if the rock were an iceberg? 29, The wind velocity vy is 40 miles per hour (mi/h) from east to west while an airplane travels with air speed v2 of 100 mish due north, The speed of the airplane relative ( the ground is the vector sum vy + v2. (a) Find v; + ¥2. (b) Draw a figure to scale. 30. An airplane is located at position (3,4,5) at noon and traveling with ve- locity 4004 + 500} — k kilometers per hour. The pilot spots an airport at position (23,28, 0).38 1. Algebra and Geometry of Euclidean Space (@) At what time will the plane pass directly over the airport? (Assume that the earth is flat and that the vector kc points straight up.) (b) How high above the airport wili the plane be when it passes? 31. A boat whose top speed in still water is 12 knots, points nosth and steams at full power. If there is an eastward current of 5 knots, what is the speed of the boat? 32. A 1L-kilogram mass focated at the origin is suspended by ropes attached to the points (1,1,1) and (-1,~1,1). If the force of gravity is pointing in the direction of the vector —k, what is the vector describing the force along each rope? (Hint: Use the symmetry of the problem. A L-kilogram mass weighs 9:8 newtons (N)] 33. A force of 6 N (newtons) points to the upper right, making an angle of 1/4 radians with the y axis. The force acts on an object that moves along the line segment from (1,2) to (5,4). (a) Find a formula for the force vector F. (b) Find the angle @ between the displacement direction D=(5—2)i + (4—2)j and the force direction F. (© The work done is F -D, or equivalently, |F| ||| cos. Compute the work from both formulas and compare. 34, Imagine that you look to the side as you walk on a windless, rainy day. Now you stop swalking, (a) How does the (apparent) direction of the falling rain change? {b) Explain the observation in (a) in terms of vectors. (c) Suppose that you know your walking speed, How could you determine the speed at which the rain is falling? Exercises 35—38 form a unit. 35. Suppose that € and ez are perpendicular unit vectors in the plane, and let v be an arbitrary vector. Show that v = (v-e:)e + (v-€2)e2. The numbers v-e; and v-€2 are called the components of v in the directions of e; and e. This expression of v as a sum of vectors pointing in the directions of e; and e2 is called the orthogonal decomposition of v relative to e, and e. 36. Consider the vectors e, = (1/¥/2)(i + j) and e2 = (1/V2){i— j) in the plane. Check that e; and e3 are unit veetors perpendicular to each other, and express each of the following vectors in the form v = ae: + are (that is, as a Hnear combination of e, and €2):1.3 2x 2and 3 x 3 Matrices and Determinants 1.32 x 2and 3 x 3 Matrices and Determinants 39 @v=i @) v=j © v=%4) (@) v=-21-j 37. Suppose that a force F (for example, gravity) is acting vertically downward on an object sitting on a plane that is inclined at an angle of 45° to the horizontal, Express this force as a sum of a force acting parallel to the plane and one acting perpendicular to it. 38. Suppose that an object moving in direction i + j is acted on by a force given by the vector F = 21+. Express this force as a sum of a force in the direction of motion and a force perpendicular to the direction of motion, 39. Show that if a is a nonzero vector and r is a nonzero scalar, then the orthogonal projection of a vector on ra is the same as its orthogonal projection on a. Vectors can be thought of as “ists of numbers, whereas matrices are arrays ‘of numbers. In this section we study some basic properties of matrices in preparation for the next section on the cross product of vectors. AQx 2matrix is an amay a1 on of four scalars. For example, 21) f-10 o4asl ad are 2.2 matrices. The determinant an a3 O22 13,7 wo [ST] a2 a2 of such a matrix is the umber defined by the equation an 21 12 a4 2 one @13022 ~ @izaar- 56 |} 3|-0-2=—1. Algebra and Geometry of Euclidean Space Example 2 A3 x 3 matrix is an array of scalars an a2 as 92 22 a23 a3, daz a3 where aij denotes the entry in the array that is in the ith row and the jth column, We define the determinant of a 3 x 3 matrix by the rule an 42 as 921 G22 423 ay 32 G33 x2 293) ~ 412 @y2 O33 421 23 ag O35 an 42 an 2 ag, Os Without some mnemonic device, this formula would be difficult to memorize, The rule to learn is that you move along the first row, multiplying aj by the determinant of the 2 x 2 matrix obtained by crossing out the first row and the Sth column, and then you add these up, remembering to put a minus in front Of the ayz term, For example, the determinant multiplied by the middle term of the preceding formula, namely, on a3 a3 a33 |" is obtained by crossing out the first row and the second column of the given 3.x 3 matrix, Schematically, the rule is illustrated in Figure 1.3.1, 3% 3 determinant = - x FIGURE 1.3.1. One way to remember how to evaluate a 8 x 3 determinant. o1 ao -34+12-9=0 @132% 2and 3 x 3 Matrices and Determinants 41 An important property of determinants is that interchanging two rows or two columns results in a change of sign. For 2 x 2 determinants. this is verified as follows: For rows, we have a1 a9 A122 ~ arte 221 a2 Gn Gap = (ona aer)=—|y0 ae and for columns, an ay oy an = (012021 — 211022) = : fm dyg| 7 CML UM) = | a Similarly, in the 3x3 case, the interchange of any two rows or any two columns changes the sign of the determinant, We also nove that the determinani of a matrix with two identical rows is zero, Simple observations like this will lead to improved methods for evaluating determinants ab ac ‘The transpose of a matrix | 7 9 | is defined to be the matix | 5 obtained by reflection across the main (upper left to lower right) diagonal. For abe ad g a3xdmatix| de f |,thecansposeis| 6 e h |. ghi efdé 158 Example 3 (a) Find the transpose of | _5 9 |. >) Show that the determinant of a 2x 2 matrix is equal to the determinant a b|_la « of its transpose: |? 3 /=|5 (c) Check (b) for the matrix in (a). Solution G@ The tanspose of | sJe[5 4): ® |¢ él ad — be;| 5 4 | = ad — eb. Since be = eb, they are equal. t 2 1 -3 ely g[=2415=an ao, |b psatis=17 *421. Algebra and Geometry of Euclidean Space Example 4 Another fundamental property of determinants is that we can factor scalars out of any row or column, For 2 x 2 determinants, this means that ear, aye ay aay] _ [aun aia a2. axe 2 az a2 ax = [aan are an ag an am] [aes cary Similarly, for 3 x 3 determinants we have am, Ad ans az. azn aay ag1 93233 M1 O12 aay a Om Oa ag, 432 G33 41 aay. aig @21 aa72 aay 43, 032 53 and so on, These results follow from the definitions. In particular, if any row or column consists of zeros, then the value of the determinant is zero. ‘A third fundamental fact about determinants is that if we change a row (or column) by adding any multiple of another row (or column) to it, the value of the determinant remains the same. For the 2 x 2 case this means that for any number k, Sy Ae: dbp ay ba + kag For the 3 x 8 case, a a2 a3| [a +hb) ag+kby ag + kby a +kaz a2 a3 db be bs) =] bs b3 bi +hb, by bs 1 cr % ce crtke o 03 and so on. Again, this property can be proved using the definition of the determinant (see Exercise 9) Evaluate 21 1 31 1). 40 -1 Solution Subtract the second row from the first: 211 -10 0 31 1 B31 1 40 -1 40 -1 |1.32 x 2and 3 x 3 Matrices and Determinants 43 Determinants seem to have been invented and first used by Gottfried Wilhelm Leibniz, Historical Note in 1693 in connection with solutions of linear equations. Colin Maclaurin and Gabriel Cramer developed their properties between 1729 and 1750; in particular, they showed that the solution of the system of equations @ui01 + a12ta + aiats = bi, enix; + azata + asta = ba, ema; +a59¢2 + asae3 = bo, is y fa a2 an as m= qb an ars |, tras], bs asa ass bs ass. and ay by ab], ayg by where This is known as Cramer's rule, Gabriel Cramer (1704—1752) published this formula in his book Fuiroduction a FAnalyse des Lignes Courbes Algebriques (17503; however, this result was probably already known to Colin Maclaurin. Later, Alexandre-Theophile Vandermonde (1772) and August Cauchy (1812), treating determinants as a separate topic worhy of special attention, developed the field more systematically, with con- tributions by Laplace, Jacobi, and others. Formulas for volumes of parallelepipeds in terms of determinants are due to Joseph Louis Lagrange (1775). We shall study these in the next section, For the full history up to 1900, see T. Muis, The Theory of Determinants in the Historical Order of Development, reprinted by Dover, New York, 1960. Example $ Suppose that the rows a, b, and ¢ of a 3 x 3 matrix satisly asabt fe; fe, (01,02, 03) = a{by, bo,bs) + B(crs¢2,03). (We say that a is a Hnear combination of b and ¢.) Show that a a2 43 dy by bg ee es“4 1. Algebra and Geometry of Euclidean Space Example 6 Solution The determinant in question is ab, + Be, aby + Beg abs + ficy a by bs |e ice c os Subtracting @ times the second row and then times the third row from the first row gives o 0 0 by ba bg ee e2 which is equal to. # ‘The 2 x 2 determinants in the formula for the 3 x 3 determinant are called mmtnors, and the formula itself is called the expansion by minors of the first row. It turns out that a determinant can be evaluated by expanding in minors of any row or column, To do the expansion, multiply each entry in a given sow or column by the 2 x 2 determinant obtained by crossing out the row and column of the given entry. Signs are assigned to the products according to the checkerboard pattern: +--+ ~ + -}. +a + Find the value of 10 3 21-2). 5 oO 4 Solution It is efficient to expand along the row or down the column with the most zeros, Here we expand the determinant by minors of the second column; the only nonzero term is iF < aw —1L, Since there is a + sign at the place in the checkerboard corresponding to the 1, —11 is the value of the determinant. @Exercises for gu3 & 3. 7. 10, un 12. 13. 41.32 X 2and 3 x 3 Matrices and Determinants 45 Evaluate the determinants in Exercises 1-8. ri 1d at elie 6 5 «4 [9 9 12 10 3 17 101 110 o19 6 Jo 1d 101 oli 2-10 111 4°32 8 |2 2 2 3°01 333 . Show that adding three times the second row of a 3 x 3 matrix to the first row leaves the determinant unchanged. Show that the determinant of any 3 x 3 matrix is equal to the determinant of its transpose. Find a 2x 2 matrix with determinant 27, all of whose entries are negative, Find a 3 3 matrix with determinant 31, all of whose entries are negative. Show that if then the solutions to the equations az + by cx + dy are given by the formulas eb ae fa ef a bl * "Tab cal [sa ‘This result is an instance of Cramer's rule.46 1, Algebra and Geometry of Euclidean Space 1.4 The Cross Product and Planes 14, Verify Cramer's rule for 3 x 3 systems: suppose that the determinant a bo az be te a3 83 D= is unequal to 2e10, Then the solution of the equations arthy tae =a, ona + boy + o22 = da, aga + bgy + 032 = dg, is given by the formulas dob oa by dy dy by ca be dy ra bs ds D DO 15, Use the 2 x 2 Cramer mule (Exercise 13) to solve the system of equations 4 + 3y = 2,22 - by = 1. 16, Use the 3 x 3 Cramer rule (Exercise 14) to solve the system of equations -t+y=Meetyte Barty +52 =—1 In §1.2 we studied a scalar ab obtained from two vectors a and b. In this section we define a vectora x b, called the cross product of a and b. This new vector will have the pleasing geometric property that it is perpendicular to the plane spanned by a and b. Our definition is based on determinants, a concept presented in the preceding section. Definition of the Cross Product Let a = ai-+a2j+agk and b = byi+ bo) + bgk be vectors in R°. The cross product or vector product of a and b, denoted a x b, is defined to be the vector az as|, |r as a a axba ls asl! [a e+ bh be or, symbolically, ijk axb=/a@, a2 a3]. bon bsExample 1 Example 2 1.4 The Cross Product and Planes 47 Even though we only defined determinants for arrays of scalars, the formal expression involving vectors is a useful memory aid for the cross product, Find (3i—J +) x (1+ 9) —). Solution ij k (Bi-j4k)x@49—-m) = [3-11 102-1 _ fa ay fa ak fs at “|e a -[ al +i al = -ibjtth 4 Certain algebraic properties of the cross product follow from the definition. If a,b, and ¢ are vectors and a, 8, and 7 are scalars, thers (a) ax b= ~(b xa); ®) ax (6b + ye) = Bla xb) +4[a xe), (aa + Ab) xc = afa xc) + Alb xe) Note that a x a = —(a x a), by propery (a). Thus, a x a= 0. In particular, ixi=0, jxj=0, kxk=0. Also, ixjek, jxk=i, kxixj, which can be remembered by writing i,j, and kx ina circle like this: CO ae (a) Compute (3i + 2j — kk) x Gj — k) by using the properties of the cross product. (b) Find i x (ix j) and (ix 4) xj, Are they equal?48 1. Algebra and Geometry of Euclidean Space Solution (a) We use the products i xj =k, etc., and the algebraic rules as follows: GIF 9-k)xG-W) = GitA-k)xj-Gi+9-k) xk = Bxp+%xjokxj-3ixk ~xktkxk = Sk+O+i+ 3-240 = + 3f4 3k. This can be checked using the definition in the preceding box. (b) We find that i x (ix j) = ix k = —j, whereas (ix i) xj = 0 xj = 0, so the two expressions are not equal. This example means that the cross, product is not associative, that is, one cannot move parentheses as in ordinary multiplication. @ To give a geometric interpretation of the cross product, we first introduce the triple product. Given three vectors a,b, and , the real number (axb)-c is called the triple product of a,b, and ¢ (in chat order), To obtain a formula for it, et a = a1 + a9) + ask, b = 64+ baj + bsk, and ¢ = cii+caj + cak. Then = (loz as], Jar aa], far ap wxme = ((2 Bl: rare ale zs) “(cud + ea} + e3k) = [a2 af, lar as a a2 = le bs a bi by cat bbe C3. ‘This is the expansion by minors of the third cow of the determinant, so ay a2 a9 (axb)-c=|b ba bs]. a a ca If ¢ is a vector in the plane spanned by the vectors a and b, then the third row in the determinant expression for (a x b)-¢ is a linear combination of the first and second rows and therefore (a x b) -¢ = 0. (Compare Example $ in §1.3.) In other words, tbe vector a x b is orthogonal to any vector in the plane spanned by a. and b, in particular to both a and b. Next, we calculate the length of a xb. Note that 2 2 2 a2 a3 ay ay a ay da bg by bs bb = (axbg — agb2)? + (arbs ~ bras)? + (arb2 ~ bra)? laxbl? =Example 3 1.4 The Cross Product and Planes: Aa If we square the terms in the last expression, we can recollect them to give (az + a3 + a3 )(0F + 03 + 03) — (arbi + abe + abs)”, which equals shall? [b|? — 1 +b)? = [lal |[b||? — |\al|?|b|? cos7@ tal? Ub)? sin#e where @ is the angle between a and b,0 < @ < x. Taking square roots and using VE? = |k|, we find that |\a x bl] = {lal [bl] | sin 4). Combining our results, we conclude that a x b is a vector perpendicular to the plane spanned bya and b with length |\al b| | sin 0]. We see from Figure 1.4.1 that this length is also the area of the parallelogram (with base |al| and height \[bsin@}) spanned by a and b, There are still two possible vectors that satisfy these conditions, because there are two choices of direction that are perpen- dicular (or normal) to the plane P spanned by a and b. This is clear from Figure 1.4.1, which shows the two choices m and —m perpendicular to P, with [Iau] = [| ~ |] = [fall [Ib [sin 8}. FIGURE 1.4.1. m1 and mp are the two possible vectors orthogonal te both a and b, and with norm |fal| bj] | sin | Which vector represents a xb, m or —m,? The answer is m1. Try a few cases such as k = i x j to verify this. The following “right-hand rule” determines the direction of a x b in general. Take your right hand and place it so your fingers curl from a towards b through the acne angle 9, as in Figure 1.4.2. Then your thumb points in the direction of a x b. Find the area of the parallelogram spanned by the vectors a = i+ 2) + 3k and b=-i-k Solution We calculate the cross product of a and b by appl nent or determinant formula, with a1 = 1,a2 = 2,4 = 3,61 wel axb = [(2)(-1) — (3) + (3)(-1) - (1-1) + (00) — 2-1)50 1. Algebra and Geometry of Euclidean Space = -%-+2k, Thus the area is llax bl = /(-2)? +27 + QP =2Vv3. « FIGURE 1.4.2. The right-hand rule for determining in which of the two possible directions ax b points, length = 1b 1 Ia 61 FIGURE 1.4.3, The length of a x b is the area of the parallelogram formed by a and b. Given two vectors in space, the problem of finding a third vector that is or thogonal to both of them can be tumed into a pair of linear equations if we use the dot product. The cross product, though, gives the solution directly, as in the following example. Example 4 Find a unit vector orthogonal to the vectors i+ jandj +k.