Chapter 1 Vector Analysis LI Vector Algebra 1.1.1. Vector Operations If you walk 4 miles due north and then 3 miles due east (Fig. 1.1), you will have gone a total of 7 miles, but you're not 7 miles from where you set out you're only 5. We need an arithmetic to describe quantities like this, which evidently do not add in the ordinary way. ‘The reason they don’t, of course, is that displacements (straight line segments going from fone point wo another) have direction as well as magnitude (length), aud itis essential to take both into account when you combine them. Such objects are called veetors: velocity, acceleration, force and momentum are other examples. By contrast, quantities that have magnitude but no direction are called scalars: examples include mass. charge, density, and. temperature, I shall use boldface (A, B, and so on) for vectors and ordinary type for scalars. The magnitude of a vector A is written |A| or, more simply, A. In diagrams, vectors are denoted by arrows: the length of the arrow is proportional to the magnitude of the vector, and the arrowhead indicates its direction. Minus A (—A) is a vector with the Figure 1.1 Figure 1.22 CHAPTER I. VECTOR ANALYSIS same magnitude as A but of opposite direction (Fig. 1.2). Note that vectors have magnitude and direction but not location: a displacement of 4 miles due north from Washington is, represented by the same vector asa displacement 4 miles north from Baltimore (neglecting, of course, the curvaturé of the earth). On h tiagram, therefore, you can slide the arrow around at will. as long as vou don’t chang its length or direction ‘We define four vector operations: addition and three kinds of multiplication. (® Addition of two vectors. Place the tail of B at the head of A; the sum, A +B, is the vertor from the tail nf A to the head of B (Eig. 1.3). (This rule generalizes the obvious procedure for combining two displacements.) Addition is commutative ASB=By A; 3 miles east followed by 4 miles north gets you to the same place as 4 miles north followed by 3 miles east. Addition is alse associative: (A+B) + A+(B+O) To subtract a vector (Fig. 1.4), add its opposite: A-B=A4(-B), , A (B+A) Lk of yo ZS Figure 1,3 Figure 14 (ii) Multiplication by a scalar, Multiplication of a vector by a positive scalar a mul- fiplies the magnitude bust leaves the direct Fig. 1.5). Uf a is negative, the direction is reversed.) Scalar multiplication is disiriburive om unchang AA 4B) =A + 0B. (ii) Dot product of two vectors. The dot product of tivo vectors is defined by A-B= ABeosé, ay where @ is the angle they form when placed tail-to-tail (Fig. 1.6). Note that A - Bis itself a scalar (hence the altemative name sealar product). Te dot product is commutative, A-B=B-A,1.1, VECTOR ALGEBRA 3 Figure 1.5 Figure 1.6 and distributive A-B+C Geometrically, A - Bis the product of A times the projection of B along A (or the product of B times the projection of A along B). If the two vectors are parallel, then A - B= AB. In particular, for any vector A, ABLALC a2) AA <3) If A and B are perpendicular, then A -B = 0. B (Fig. 1.7), and calculate the dot product of C with itself ©-C=(A—B)-(A-B)=A-A-A-B-B-AGB-B, 2ABeos0. ‘hus 1 the law of cosines. (iv) Cross product of two veetors. The cross product of two vectors is defined by Ax B= ABsingi, aay ‘where fis a unit vector (vcctor of length 1) pointing perpendicular to the plane of A and B. (I shall use a hat (*) to designate unit vectors.) Of course, there are two directions perpendicular to any plane: “in” and “out.” The ambiguity is resolved by the right-hand Tre: let your fingers point inthe direction of the first vectorand curl around (via the smaller angle) toward the second; then your thumb indicates the direction of fi. (In Fig. 1.8 A x B points into the page: B x A points our of the page.) Note that A Bis itself a veetor (hence the alternative name veetor product). The cross product is distributive Ax (B+O) = (A xB) + (A x ©), as)4 CHAPTER I. VECTOR ANALYSIS Figure 1.7 but not commutative. tn fact, (Bx A)=-(A xB) 6) Geometrically, |A x B| is the area of the parallelogram generated by A and B (Big. 1.8). If {wo vectors are parallel, their cross product is zero. In particular, AxA=0 for any vector A Problem 1.1 Using the definitions in Fas. 1-1 and 1.4. and appropriate diagrams, show that the dot product and eross product are distributive, 2) when the three vectors are coplanar; ) in he general case, Problem 1.2 Is the cross product associative? (AxB)xC2Ax Bx), Ifso, prove it if not, provide a counterexample. 1.1.2 Vector Algebra: Component Form In the previous section I defined the four vector operations (addition, scalar multiplication, dot product, and cross product) in “abstract” form—that is, without reference to any partic~ tular coordinate system. In practice, itis often easier to set up Cartesian coordinates x, y, z and work with vector “components.” Let &, §, and Z be unit vectors parallel to the x, y,and z axes, respectively (Fig, 1.9a)). An arbitrary vector A can be expanded in terms of these basis vectors (Fig. 1.9(b)): AR AREAS + Ah.1.1. VECTOR ALGEBRA 5 Figure 1.9 ‘The numbers Ax, Ay, and Az, are called components of A; geometrically, they are the projections of A along the three coordinate axes. We can now reformulate each of the four vector operations as a rule for manipulating components: A | BH(AR+ AVS + Aci) + (Br + BG + 82) (Act Bok + (Ay + Bye + (As +B: a7 ( Rute: To add vectors, add like components GA = (aAy)R + (aAy)¥ + (ADE. (1.8) Gi) Rule: To multiply by a scalar, multiply each component. Because X, §, and Z are mutually perpendicular unit vectors, RRHG Fae &-f=k-2=§9-2=0. (9) Accordingly, AB = (Axk+Ayh + AB)- (BX + By + BB) = ANB | AyBy | ADD. (1.10) (iii) Rule: To calculate the dot product, multiply like components, and add. In particular, AAS AT+AR+ As fatale’ quip (this i, if you like, the three-dimensional generalization ofthe Pythagorean theorem.) Note that the dot product of A with any unit vector is the component of A along that direction (thus A-& = Ay, A-§ = Ay. and A-2= Az6 CHAPTER I. VECTOR ANALYSIS Similarly,! kxk= $x = tx7=0, RxGa-oxk =f fxisaxg = & ax Kxd b (1.12) ‘Therefore, ARB = (AR + ASS + Ad) x (Bk + By + BD) a3) = (AyB: ~ A, By)X + (A: By — AyBz)¥ + (AyBy — Ay By). This cumbersome expression can be written more neatly as a determinant: gop k AxXB ir Ay Az (day | B, By iv) Rules To calculate he cross product, form the determinant whose frst os is X,Y. whose second row is A (in component form), and whose third row is B. Example Find the angle between the face diagonals of a cube Solution: We might» well use a eube of side [, and place iss shown in Fig. 1.10, with one ‘commer atthe origin. The face diagonals A and Bare ASIEHOS 41% R=ORHIV4IE 4/11, 0,0) Figure 1.10 "hese signs pertain oa right handed coordinate sytem (x-axis out ofthe page, y-axis tothe righ, z-axis up, ‘orany rotated version thereof). Ina lef-handed system (sans daw) he signs are veversed: 8 § =~ an so (on, We shall use right-handed sysiems exclusively1.1. VECTOR ALGEBRA 7 So, in component form, ABal-O40-141-121 ‘On the other hand, in “abstract” form, A.B = ABoosi = ¥2V2cos6 = 2058, ‘Therefore, cos = 1/2, or 6 = 60". ‘Ut course, you can get tne answer more easily by drawing in a diagonal across the top of the ‘cube, completing the equilateral triangle. But in cases where the geomelty is not so simple, this device of comparing the abstract and component forms of the dot product can be a very efficient means of finding angles. Problem 1.3 Find the angle between the body diagonals of a cube, Problem 1 Use the ors prot to find the esmponents ofthe anit vector i perpen tothe pln shown in Fig. LIT 1.1.3 Triple Products Since the cross product of two vectors is itself a vector, it can be dotted or crossed with a third vector to form a tripte product. @ Scalar triple product: A - (B x C). Geometrically, |A - (B x C)| is the volume of the parallelepiped generated by A, B, and C, since |B x C| is the area of the base, and |A.cosd) is the altitude (Fig. 1.12). Evidently, A. (Bx) =B-(Cx A) for they all correspond to the same figure. Note that view of By. 1.6, the “noialphabetival” Uiple products, A-(C x B)=B-(AxC)=C-(Bx A), (Ax B), is) Iphabetical” order is preserved—in Figure 1.11 Figure 1.128 CHAPTER 1. VECTOR ANALYSIS have the opposite sign. In component form, Ar Ay A, AWBxC)=| B By Be (116) Cr Gy C, Note that the dot and cross can be interchanged A Bx O)~ (As B)-C (this follows immediately from Ey. 1.15); however, the placement of the parentheses is critical: (A. B) x C is a meaningless expression—you can't make a eruss product from a scalar and a vector. i) Vector triple product: A x (B x C). The vector triple product can be simplified by the so-called BAC-CAB rule: Ax (Bx C) = B(A-€) ~ CAB). a7) Notice that (Ax B) x C= -€ x (A x B) = -A(B-C) + BAC) is an entirely different vector, Incidentally, all higher vector products can be similarly reduced, often by repeated application of Eq. 1.17. so it is never necessary for an expression to contain more than one crass product in any term, For instance, (AxB)-(CxD) = (A-OB-D)~(A- DVB-C); Ax (Bx(CxD)) — BA (C xD) ~(A-BYC x Dy (ts) Prahlom 1§ Prove the RAC.CAR rule hy siting out both cides ia component foun Problem 1.6 Prove that [A x (Bx ©) + Bx (Cx A+ (Cx (A xB] =0. Under what conditions does A x (B x ©) = (A x B) x C? 1.1.4 Position, Displacement, and Separation Vectors ‘The location of a point in three dimensions can be described by listing its Cartesian coor dinates (x, y, 2). The vector to that point from the origin (Fig. 1.13) is called the position vector: ska ype (119)1.1. VECTOR ALGEBRA 9 source point 2 Figure 1.13 Figure 1.14 1 will reserve the letter r for this purpose, throughout the book. Its magnitude. 4.20) is the distance from the origin, and r_oxktypte ro fete yte a2 is a unit vector pointing radially outward. The infinitesimal displacement vector, from (x. y.2)t0 (e+ dx, y + dy, 2 + ds), is dl =dxi+dy§ +dzi. (1.22) (We could call this dr, since that’s what itis, but itis useful to reserve a special letter for infinitesimal displacements.) Inclectrodynamics one frequently encounter prublems involving évv puinls—typivally source point, r’, where an electric charge is located, and a field point, r, at which you are calculating the electric or magnetic field (Fig. 1.14). It pays to adopt right from the start some short-hand notation tor the separation vector trom the source pomt to the held point. I shall use for this purpose the script letter: (1.23) Its magnitude is . (1.24) and a unit vector in the direction from ¥” to ris — (125)lo CHAPTER I. VECTOR ANALYSIS In Cartesian coordinates, (1.26) 1.27) (1.28) (from which you can begin to appreciate the advantage of the script-2 notation). Problem 1.7 Find the separation vector from she source point (2,8,7) tothe field point 4.68). Determine its magnitude (2), and construct the unit weetor 1.1.5 How Vectors Transform ‘The definition of a vector as “a quantity with a magnitude and direction" is not altogether satisfactory: What precisely doce “directjon” mean? ‘This may seem a pedantic question, but we shall shortly encounter a species of derivative that Jooks rather like a vector, and we'll want to know for sure whether it fs one, You might be inclined to say that a vector is anything that has three components that combine properly under addition. Well, how about this: We have a barrel of fruit that contains Ny pears, Ny apples, and N, bananas. IsN = N,& + Ny¥ +N, vector? It has three components, and when you add another barrel with M, pears, My apples, and M, bananas the result is (N+ My) pears, (Ny -+ My) apples, (N. + Mz) bananas, So it does add like a vector. Yet it’s obviously not a vector, in the physicist’s sense of the word, because it doesn’t really have a direction. What exactly is wrong with it? The answer is that N does not transform property when you change coordinates, The coordinate frame we use to describe positions in space is of course entirely arbitrary, but there is a specific geometrical transformation law for converting vector components from ‘one frame to another, Suppose, for instance, the F, ¥, Zsystem is rotated by angle @, relative ma —¥ anes, Pua Fig. 1.15, to x,y,z, about the cos Ay = Acosé, Az = Asin8, while = Acos® = Acos(? ~ $) = A(cos4 cos@ + sind sing) = cos@ay +sin ga, Asin = Asin(6 ~ 8) = A(sin6 cos $ — cos 8 sing) = sing, +cospA:. his section ean be skipped without lose of continuity.1.1. VECTOR ALGEBRA " Figure 1.15 ‘We might express this conclusion in matrix notation: Ay cos sing \ / Ay (3)-(288 8) (2) we More generally, for rotation about an arbitrary axis in three dimensions, the transfor- ‘mation law takes the form (2)-(E RE E)G) om 2 Deas daly i where the index | stands for x, 2 for y, and 3 for z. The elements of the matrix R can be ‘ascertained, for a given Totation, by the same sort of geometrical arguments as we used for a rotation about the x axis Now: Do the components of N transform in this way? Of course not—~it doesn’t matter ‘what coordinates you use 10 represent positions in space, there is still the same number of apples in the barrel, You can't convert a pear into a banana by choosing a different set of axes, But you cant turn A, into Ay. Formally, then, a vector is any set of three components ‘hat transforms in the same manier as a displacement when you change coordinates. As always, displacement is the model for the behavior of all vectors. By the way, a second-rank) tensor is a quantity with nine components, Ty. Trys Tres Tx, o++5 Tez, which transforms with no factors of R: ‘or, more compactly, Tay = Rey (ResTex + RayTey + RaeTee) Ray (RexTyx + ReyTry + Reels) + Rec(RecToe + RayTey + Re12 CHAPTER |. VECTOR ANALYSIS or, more compactly, 33 Ty = Oo Ra RT (1.32) lu general, un mh-rank tensor has m indices and. 5” components, and transtorms with 1 factors of R. In this hierarchy, a vector is a tensor of rank 1, and a scalar is a tensor of rank ato. Problem 1.8 (a) Prove that the two-dimensional rotation matrix (1.29) preserves dot products. (That is, show that Ay By + A-B; = AyBy + A:B:.) (b) What constraints must the elements (R;,) of the three-dimensional rotation matrix (1.30) satisfy in order o preserve the length of A (forall vectors A)? Problem 1.9 Find the transformation matrix R that describes a rotation by 120° about an axis from the origin through the point (1, 1, 1). The rotation is clockwise as you look down the axis toward the origi. Problem 1.10 (2) How do the components ofa veetor transform under a translation of coordinates (F Fy —a,2—2,Fig. 1.164)? (b) How do the components ofa vector transform under an inversion of coordinates (F y=-y ig. 1.160)? (6) How does the cross product (1.13) of two vectors transform under inversion? [The cross- product of two vectors is properly called a pseudovector because of this “anomalous” be- havior] Is the cross product of two pseudovectors a vector. or a pseudovector? Name two pseudovector quantities in classical mechanics, (@) How does the scalar triple product of three vectors transform under inversions? (Such an cchject ia called a peeudocealan) 1/2 «@ Figure 1.161.2 1.2. DIFFERENTIAL CALCULUS 13 Differential Calculus 1.2.1 “Ordinary” Derivatives Qucstion. Suppose we have a function uf une vatiable. fix). What does the derivative, df /dx, do for us? Answer: Ittells us how rapidly the function f(x) varies when we change the argument x by a tiny amount, dx: df a= (Zh) ax (1.33) Inwords: Ifwe change x by an amount dx, then f changes by an amount df the derivative is the proportionality factor. For example, in Fig, 1.17(@), the function varies slowly with x, and the derivative is correspondingly small. In Fig. 1.17(b), increases rapidly with x. and the derivative i large, as you move away from x = 0. Geometrical Interpretation: The derivative df/dx isthe slope ofthe graph off versus fy tt / Co Le a * Figure 1.17 1.2.2 Gradient Suppose, now, that we have a function of shree variables—say, the temperature T(x, y,:) inaroom, (Start out in one comer, and set up a system of axes; then for each point (x, y, 2) 1m the room, 7 gives the temperature at that spot.) We want to generalize the notion of “derivative” to functions like T, which depend not on one but on three variables. Now a derivative is supposed to tell us how fast the function varies, if we move a little distance. But this time the situation is more complicated, because it depends on what direction we move: If we go straight up, then the temperature will probably increase fairly rapidly, but if we move horizontally, it may not change much at all, In fact, the question “How fast does T vary?” has an infinite number of answers, one for each direction we might choose to explore. Fortunately, the problem is not as bad ay it looks. A theorem on partial derivatives states that . r= (2)are (Zar+(Z)a 34)14 CHAPTER 1, VECTOR ANALYSIS imal amounts This tells us how T changes when we alter all three variables by the infin dx. dy,dz. Notice that we do not require an infinite mimber of derivatives—rhree will suffice: the partial derivatives along each of the three coordinate directions. Equation 1.34 is reminiscent of a dot product: ) (dvi+dy9 + dza (1.33) where Terao Tae aT, ax” ay? is the gradient of T. VT is a vector quantity, with three components; i isthe generalized derivative we have been looking for. Equation 1.35 is the three-dimensional version of Ey, 1.33, Geometrical Interpretation of the Gradiem: Like any vector, the gradient has magnitude and direction. To determine its geometrical meaning, let's rewrite the dot product (1.35) in abstract form: (1.36) aT =VT d= {VT ijdljcose, 137) where @ is the angle between VT and dl. Now, if we fix the magnitude |dl and search around in various directions (that is, vary 6), the maximum change in T evidentally occurs, when @ = 0 (for then cos® = 1). That is, for a fixed distance |dll, dT is greatest when I move in the same direction as VT. ‘Thus: The gradient VT points in the direction of maximum increase of the function T, Moreover: The magnitude |VT| gives the slope (rate of increase) along this maximal direcsion Imagine you are standing on a hillside, Look all around you, and find the direction of steepest ascent. That is the direction of the giadict, Now measure the ofupe in dha direction (rise over run), That is the magnitude of the gradient. (Here the function we're talking about is the height of the hill, and the coordinates it depends on are positions— lucitude and longitude, say. This function depends on only two variables, not three, but the geometrical meaning of the gradient is easier to grasp in two dimensions.) Notice from Eq, 1.37 that the direction of maximum descent is opposite to the direction of maximum ascent, while at right angles (6 = 90°) the slope is zero (the gradient is perpendicular to the contour lines). You can conceive of surfaces that do not have these properties, but they always have “kinks” in them and correspond to nondifferentiable functions ‘What would it mean for the gradient to vanish? If VT" = O at (x, y. 2), then dT = 0 for small displacements about the point (x, y, 2). This is, then, a stationary point of the It could be a maximum (a summit), a minimum (a valley), a saddle function T(x, y1.2. DIFFERENTIAL CALCULUS 15 point (a pass), or a “shoulder.” This is analogous to the situation for functions of one variable, where a vanishing derivative signals a maximum, a minimum, or an inflection, In particular, if you want to locate the extrema of a function of three variables, set its gradient equal to zero, Example 1.3 Find (the magnitude of the position vector) v= ‘Does this make cence? Well, it says that the distance froma the origin ineresses most rapidly in the radial direction, and that its rate of increase in that direction is L...just what you'd expect, Problem 1.11 Find the gradients ofthe following functions: sy gare rec, ©) fly. 2) axe (6) FO 9,2) =e" siatypin(2), Problem 1.12 The height of certain hill (in fet) is given by Bx, ¥) = WOQry — 34? ~ y? — 18x + 28y + 12), where vis the distance fin mies) nod, «the distance east of South Hi (@) Whereis the top of the hill located? (b) How high isthe hil? (©) How steep is the slope (in feet per mile) ata point { mile north aad one mile east of South Hadley? tn what ditection is the slope steepest, at that point? Problem 1,13 Let be the separation vector froma fixed point (x, y', 2) the point (x. yz) and Jet be its length, Show that Va) =m vam (©) What isthe general formula for V2")? aelo CHAPTER I. VECTOR ANALYSIS Problem 1.14 Suppose that / is # function of two variables (y atx z) only. Show that the gradient Vf = (2//89 8 + (@f/8cye transforms asa vector unde rotations. Eq. 1.29. Hints (af/3¥) = (Af f@y)(@y 95) + (BF/02)(B2/ 99), and the analogous formula for A//IE. We cos +z sing and z = —y sing + 7 x xf? Figure 1.22 Figure 1.28 Example 1.7 - a Calculate the surface integra of v +(e +2)9 +y(c2 — 3) Zover tive sides (excluding the bottom) of the cubical box (side 2) in Fig. 1.23. Let “upward and outward” be the positive direction, as indicated by the arrows. Solution: Taking the sides one ata time: (x =2, da=dydei, v-da=2xedyd: fom (i) x =0, da =—dydzi, v-da= -2xzdy dy1.3. INTEGRAL CALCULUS a Gil) y= 2, da =drdz§, veda = (+ 2)drdz, 90 fr a= [rar fae (Wy 50, das —arazy, Veda =u + 2axdz, 90 frdn=- [odd [ode = ()2=2, da=dxdya, v-da = y(z? —3)dxdy = ydxdy, $0 fra fe ydy=4. ParaoeY t= 16404 12-1244 20. race Bvidontly the tora flux is (©) Volume Integrals. A volume integral is an expression of the form [re «st y where T is a scalar function and dz is an infinitesimal volume element, In Cartesian coordinates, dt =dxdydz sy For example, if T is the density of a substance (which might vary from point to point), then the yolume integral would give the total mass. Occasionally we shall encounter volume integrals of vector functions: frtcm fotenituads as fudrséfnarrafode ase because the unit vectors are constants, they come outside the integral Example 18 Caleulate the volume integral of T = xyz? over the prism in Fig. 1.24 Solution: You can do the thee integrals in any order. Let's do x first: iteuns from Oto (I~ y)s then y (it goes from 0 10 1); and finally ¢ (010 3): free fof ofa Pea fiav : ee } fee f'0- svat)28 CHAPTER I, VECTOR ANALYSIS Figure 1.24 Problem 1.28 Calculate the line integral of the function y = x? + 2yz + 5? from the ‘origin to the point (1,1,1) by three different routes: (@) 0,0,0) + (1,0,0) > 4, 1,0) > (1, Ds (©) (0.0,0) + (0,0,1) + 1, > QL Ds (6) The divcet stiaight line (@) What isthe line integral around the closed loop that goes our along path (a) and back along path (b)? Problem 1,29 Calculate the surface integral of the function in Ex. 1.7, over the bortom of the box. For consistency, let “upward” be the positive direction, Does the surface integral depend ‘only on the boundary fine for this function? What is th total flux over the closed surface of the box (including the bottom)? [Nore: For the closed surface the positive direction is “outward.” and hence “down.” for the bottom face.) Problem 1.30 Calculate the volume integral of the function ‘over the tetrahedron with corners at (0,0.0), (1.0.0), (0,1,0), and (0,0,1). 1.3.2. The Fundamental Theorem of Calculus Suppose f(x) isa function of one variable. The fundamental theorem of enlculus sates b Lax = $b) - fea) (1.54) dx In case this doesn’t look familiar, let's write it another way: . ff Fevac= 10) F0 where df/dx = F(x). The fundamental theorem tells you how to integrate F(x): you think up a function f(x) whose d2rivarive is equal to F1.3. INTEGRAL CALCULUS 29 Geomerrical Interpretation: According to Eg. 1.33, df = (df fdx)dx is the infnitesi- ‘mal change in f when you go from (x) to (x + dx). The fundamental theorem (1.54) says that if you chop the interval from a to b (Fig. 1.25) into many tiny pieces, dx, and add up the increments df from each little piece, the result is (not surprisingly) equal to the total coannge in f= f(b) — f(a). In oxher words, here are two ways {0 determine the total change in the function: either subtract the values at the ends or go step-by-step, adding up all the tiny increments as you go. You'll get the same answer either way. Notice the basic format of the fundamental theorem: the integral of a derivative over an interval is given by the value of the function at the end points (boundaries). In vector calculus there are three species of derivative (gradient, divergence, and curl), and each has its own “fundamental theorem,” with essentially the same format. I don’t plan to prove these theorems here; rather, I shall explain what they mean, and try to make them plausible. Proofs are given in Appendix A. | fe) f(b), - fa) pe Ce ns ade ee igure 1.25 Figure 1.26, 1.3.3. The Fundamental Theorem for Gradients Suppose we have a scalar function of three variables T(r, y. 2). Starting at point a, we move a small distance dy (Fig. 1.26). According to Eq, 1.37, the function T will change by an amount ar (VT) dh Now we move a little further, by an additional small displacement dlp; the incremental change in 7 will be (V7) -db. In this manner, proceeding by infinitesimal steps, we make the journey to point b. At each step we compute the gradient of T (at that point) and dot it into the displacement dl. ..this gives us the change in T’. Evidently the total change in T in going from a to b along she path selected is T(b) -T(a). (55)30 CHAPTER I. VECTOR ANALYSIS This is called the fundamental theorem for gradients; like the “ordinary” fundamental theorem, it says thatthe integral (here a ine integral) of a derivative (here the gradient) is iven by the value of the function at the boundaries (a and b). Geometrical Interpretation: Suppose you wanted to determine the height of the Eiffel ‘Tower You could climb the stairs, sing «ruler to measine the rise at each step, and adding them all up (that’s the left side of Eq. 1.55), or you could place altimeters at the top and the bottom, and subtract the two readings (that's the right side); you should get the same answer either way (that's the fundamental theorem) Incidentally, as we found in Ex. 1.6, line integrals ordinarily depend on the parh taken from a to b. But the right side of Eq, 1.55 makes no reference to the path—only to the cend points. Evidently, gradients have the special property that their line integrals are path independent: Corollary 1: f(T) - dlis independent ot path taken from a to b, Corollary 2: f(VT)- dl = 0, since the beginning and end points are identical, and hence T(b) — Ta) = 0 Fxample 1.9 Let T = xy2, and take point a to be the origin (0, 0,0) and b the point (2, 1,0). Check the Fonlamental theovean for yadients, Solution: Although the integral is independent of path, we must pick a specific path in order toevaluateit. Let's go out along the x axis (step i) and then up (step ii) (Fig. 1.27). As always, A= dri tdy3 dre 284 Oxvg (y=; dl=dei, VT -dl=y2dx =0,50 [or a-o (ux =2; dl=dyy, WE -dl=2xydy = aydy, so , 1 [era fossa hi ho Figure 1.271.3. INTEGRAL CALCULUS 31 [Evidently the total Line integral is 2. Is this consistent with the fundamental theorem? Yes: Tb) ~ Fa) =2-0=2. Now, just to convince you that the answer is independent of path let me calculate the same {ntegral along path ii (the straight line from a to): (i) y = bx. Problem 1.31 Check the fundamental theorem For gradients, using T= x2 4 4xy | 2y28.the points a = (0,0,.0), = (1, 1,1), and the three paths in Fig. 1.2 (2) (0,0.0) + 1.0,0) > .1,0) > UD: (6).0,0.0) + 0.0.1) + O11) + 1D: (€) the parabolic path 2 = x ieee 4 © 1.3.4 The Fundamental Theorem for ‘The fundamental thearem for divergences states that: fe de fe da, ergences (1.56) In honor, I suppose of its great importance, this theorem has at least three special names Gauss’s theorem, Green's theorem, or, simply, the divergence theorem. Like the other “fundamental theorems,” it says thatthe integral of a derivative (in this case the divergence) over a region (in this case a volume) is equal to the value of the function at the boundary32 CHAPTER 1. VECTOR ANALYSIS (in this ease the surface that bounds the volume). Notice that the boundary term is itself ‘an integral (specifically, a surface integral). This is reasonable: the “boundary” of a line is just two end points, but the boundary of a volume is a (closed) surface. Geometrical Interpretation: If v represents the flow of an incompressible ftwid, then the flax of v (the Fight side of Eg. 1.56) is the total arsount of fic passing ost through the surface, per unit time. Now, the divergence measures the “spreading out” of the vectors from a point—a place of high divergence is like a “faucet,” pouring out liquid. If we have tots of faucets in « region filled widh incompressible Guid, an equal amount of liquid will, be forced out through the boundaries of the region. In fact, there are nvo ways we could determine how much is being produced: (a) we could count up all the faucets, recording how much each puts out, or (b) we could go around the boundary, measuring the flow at each point, and add it all up. You gef the same answer either way: f (faucets within the volume) = ¢ (Bow out through the surface). This, in essence, is what the divergence theorem says. Example 1.10 ‘Check the divergence theorem using the function and the unit cube situated at the origin Fig. 1.29). Solution: In this case Vive drt), [rornaaaf’ [ floroan ferndcatey [iarve ue fide= and1.3. INTEGRAL CALCULUS 33 Evidently, f ‘So much for the left side ofthe divergence theorem. To evaluate the surface integral we must consider separately the six sides of the cube: 7A @ fv af f vadyde = 4 bb bo 1p “i fra Ef lb Jo A «in [ram [lors Prarae=§ by wy) o ow So the total flux is: Problem 1.32 Test the divergence theorem for the function v = (xy) + 2y2)# + Gexy 2. “Take as your volume the eube shown in Fig. 1.30, with sides of length 2.34 CHAPTER I. VECTOR ANALYSIS 71 Figure 1.30 1.3.5 The Fundamental Theorem for Curls The fundamental theorem for curls, which goes by the special name of Stokes’ theorem, states that wxyy-da= fy cs Lorre$ Asalways, the integral ofa derivative (here, the curl) over a region (here. a patch of surface) is equal to the value of the function at the boundary (hee, the perimeter of the patch). As in the case of the divergence theorein, the boundary term is itself an integral—specitically, ‘closed line integral. Geometrical Interpretation; Recall that the cur! measures the “twist” of the vectors va region of high curl is a whirlpool—if you put a tiny paddle wheel there, it will rotate, Now. the integral of the curl over some stirface (or, more precisely, the flux of the curl shrough that surface) represenis the “total amount of swirl.” and we can determine that switl just as well by going around the edge and finding how mich the flow is following the boundary ig. 1.31). You may find this a rather forced interpretation of Stokes’ theorem, but it's a helpfill mnemonic, if nothing else You might have noticed an apparent ambiguity in Stokes’ theorem: concerning the boundary line integral, which way are we supposed to go around (clockwise or counter- 1 eitut, The answer is 1.57) clockwise)? If we go the “wrung” way we'll pick up att uve that it doesn’t marter which way you go as long as you are consistent, for there is a com- pensating sign ambiguity in the surface integtal: Which way does da point? For a closed surface (as in the divergence theorem) da points in the direction of the ounward normal; but for an open surface, which way is “out?” Consistency in Stokes’ theorem (as in all such matters) is given by the right-Hand rule: If your fingers point in the direction of the line integral, then your thumb fixes the direction of da (Fig. 1.32). Now, there are plenty of surfaces (infinitely many) that share any given boundary line. ‘Twista paper clip into a loop and dip it in soapy water. The soap film constitutes a surface, with the wite loop as its boundary. If you blow on it, the soap film will expand, making a larger surface, with the same boundary. Ordinarily, a flux integral depends critically on what surface you integrate over, but evidently this is no? the case with curls, For Stokes"1.3. INTEGRAL CALCULUS 35 > Ss LY a Figure 1.32 theorem says that (Vx v) -da is equal to the line integral of v around the boundary, and. the latter makes no reference to the specific surface you choose, Corollary 1: (Vx v) «da depends only on the boundary line, not ‘on the particular surface used. Corollary 2: f(V x ¥)-da = 0 for any closed surface, since the boundary line, like the mouth of a balloon, shrinks down toa point, and hence the right side of Eq. 1.57 vanishes ‘These corollaries are analogous to those for the gradient theorem. We shall develop the paralle! further in due course. Example 1.11 Suppose v = (xz + 3y7)§ + (4y29}8. Check Stokes’ theorem for the square surface shown in Fig. 1.33, Solution: Here Vx vada — 20984228 wt 4a x Gre Figure 1.3336 CHAPTER I. VECTOR ANALYSIS (in saying shat a point inthe x direction, we are committing ourselves to a counterclockwise Jing imegral. We could as well write da = —dy dc%, but then we would be obliged to £0 clockwise.) Since 1 = O for this surface, Va Lhe lo by ‘Now, what about the line integral? We must break this up into four segments: vde = 4 dy, fv-dl= ff 3y2dy Oy dz, fvedl= ff de ds = sy2dy, fvedl =f) 3y2dy . fredis [Pde so fraai4g-140=$ Weherks, A point of strategy: notice how Ihandled step (i), There isa temptation to write dl = —dy here, since the path goes to the left. You can get away with this, if you insist, by running the integral from 0 > 1, Personally {prefer to say dl = dx 3+ dy§ + dzi always (never any ‘minus signs) and let the limits of the integral take care ofthe direction. Problem 1.33 Test Stokes’ theorem forthe function v = (xy) 8+ (2y2)$-+ (24) 3 sing the triangular shaded area of Fig. 134 Problem 1.34 Check Corollary | by using the same function and boundary Tine asin Ex. 1.11, but integrating over the five sides ofthe cube in Fig. 1.35, ‘The back of the cul 2 I i) Figure 1.34 Figure 1.351.3. INTEGRAL CALCULUS 37 1.3.6 Integration by Parts The technique known (awkwardly) as integration by parts exploits the product rule for derivatives: 7 at de (#)-*(Z) Integrating both sides, and invoking the fundamental theorem: [ Zoom=nl=[r(B)aefe(Da. [s(B)are [e(Dlarvse That's integration by parts. It pertains to the situation in which you are called upon to integrate the product of one function (f) and the derivative of another (g): it says you can transfer the derivative from g tof. at the cost of a minus sign and a boundary term. d. as (1.58) Fame 12 eed The exponential can be expressed as a derivative a eee eee We can exploit the product rut damental theorems, in exactly the same way. For example, of vector calculus, together with the appropriate fun teers VFA = FTA EAS) over a volume, and invoking the divergence theorem, yields fruaar fro Aydt + fa whar=f fa da, [a wash fa-da, (1.59) ly Is or [0m ,