\ Bea Vector Calculus Een 2:1 Introduction 182, 3.2. Derivatives of a scalar point function 199 3.3. Derivatives of a vector point function 203 3.4 Topics in integration 214 3.5 Engineering application: streamlines in fluid dynamics 248 3.8 Engineering application: heat transter 250 3.7 Review exercises (1-21) 254182. vEcToR catcutus ERY troduction Figure 34 Elemeniary vector algebra In many applications we use funetions of the space variable r= 14+.) + 2A as models ‘for quantities that vary stom point fo point in three-cimensional space. Thee are two {pes ‘of such funetions. There are sealar paint functtons, which mode! scalar quantities like the temperature ata point ina body, and yeetor paint functions, which model veetor ‘quantities like tbe velocity of the flow ata point ina Tiquid. We can express this more formally in the following way. For exch scalar point function fe have a rue «= 70). ‘which assigns to each point with coordinate rin the domain ofthe function a unique ‘eal number u. For vector point functions the mle w= Fr assigns io each 72 unique vector 1 inthe range of he function, Veotor calculus was designed to measure the variation of ‘such fanetions with espect tothe space variale . That development meade use of ths ideas bout vectors (components, addition, subtraction, scalar and vector products) described in Chapter 4 of Modem Engineering Marhemartes and summarized hete in Figure 3.1 compart: f In component form if a= (ay. dy, as) and b = (by, by 63) then. ab b= (Eda Eb ayt by) ab (ah babs+ah)= bra aoa Ja, a: | =-bxe lo by ty (bs ~ bts yas = abs. aby = bs)3.1 INTRODUCTION 183 3.11 Figure 3.2 [tel suraces off) Q.2 Der 4 8 The recent development of computer packages for the modelling of engineering problems involving vector quantities has relieved designers of much tedious analysis and computation. To be able io use those packages effectively, however, designers need good understanding ofthe mathematical ool they bring 1 thet tasks, iS on that basic understanding that this chapter focuses. Basic concepts We can picture a scalar point funetion i) by means of ts level surfaces f(r) = constant. For example, the level surfaces of (r) = 2 + 2y"~ 2 are planes parallel tothe plane 2e + 2y, as shown in Figure 3.2. On the level surface the function value docs not ‘hange, s0 the rate of change of the Funeion will be zero along any Fine drawn on the level surface. An alternative name Fora sealar point function is sealar ied. This isin ‘contast tothe vector point function (or vector feld). We picture a vostor field by its field (or flow) lines. field ine is a curve in space represented by the position vector ‘(such that at eoeh point ofthe curve its tangent is parallel tothe veetor field, Thus ‘ho fil lines of F(x) are givon by the eifferoatisl equation 46 FQ, whore ns) = SE =F), whore Xi) =r) and ry is the point on the ine conresponding fo 1 =. This vecior equation represents the three simultaneous ordinary differential equations where F =(P, 0.) ‘Modem computer algebra packages make it easier to draw both ihe level surfaces of scalar funetions and the field lines of veetor functions, but 1 underline the basic ideas wwe shall consider nyo simple examples.184 vecToR catcutus Example 3.1 Solution Figure 338 a) Level surfaces offt)=265 (6) field lines of | Fiy=(vrd) Sketeh (a) the level surices ofthe scalar point fanction fl») =20s (b) the field lines of the veetor point Furetion F(x) = (3, x51) (2) Consider the level surface given by f(r) = ¢, where c is a number. Then ze" =cand so: = ce", Foro, x and y all postive we can easily sketch part of the surface as shown in Figure 3.3(a), from which we ean deduce the appearance of the whole family of level sufices. « ® (0) For the function F()= (x. 1) the feld lines ate given by ar fecnny that is, by the ima neous differential equations & ar The general solution of these simultaneous equations is wi) =Acosr + Bsini, yu)=—Beost+ Asin’, 2()=14C where 4, # and C are arbitrary constants. Considering, in particular, the field line that passes through (1,0, 0), we determine the parametric equation Ge. (cost, sin 1) This represents a circular helix as shown in Figure 3.3(0), fom which we can deduce the appearance of the whoke family of low lines,3.1 INTRODUCTION 185 In MATLAR a level surface may he drawn using the osu function. Using the ‘Symbolic Math Toolbox the commands: syns xy 2 © fore = 1231 2 = ctenpxty); exsurt (2, (0, hold on, will precice three of the level surfaces of = = e* on the same set of axes. The surfaces may also be produced in MAPLE using the ©z>.r fimetion. The field lines may be plotted in MATLAB using the stream ine function ‘To investigate the properties of scalar and vector fields further we need to use the caleulus of several variables. Here we shall dseribe the basie ideas and definitions rveded for vector calculus, A flor treatment i given in Chapter 9 of Modern Engineer ing Mathematics Given 2 function flx) of a single variable x, we messure its rate of change (or _radient) by its derivative with respect to x. This is However, «function (x2) of three independent variables x, y and = does not have a unique rate of change, The value of the latter depends on the dicection in which i i measured. The rate of change of the funtion x,y, 2) in the x direction is given by is partial derivative with respect x, namely OE img Hus v2) = fix. v2 Ix ah a ‘Tis measures the rate of change off, 2) with respect to x when y and 2 are held constant. We can calculate such partial derivatives by differentiating Jr, x, 2) with respect 0x, treating. and = as constants, Similarly, define the partial derivatives of fx, 2) with respect to and = respectively.186 vEcTOR catcuLus Example 3.2 Solution For conciseness we Sometimes use a Sufix notation to denote partial devivaives, for example writing J for 2//9r. The rules for partial differentiation are essentially the ‘same as for onary differentiation, but it must sways be remembered which variables are being held constant. Higher-order partial derivatives may be defined in a similar manner, with, for example, Find the first partial derivatives ofthe fimotons fis, 9 (b).x(y +2) and (6) + Ve? +a). ‘vith formala (a) + 2p 2% (0) spy) = +27 +2. To obtain, we difereniat fx, y, 2) with repect 0 Kospingy and : constant. Thus f= |, sine the derivative of «constant (2y + with respoot tis zer. Similarly, = 2 and f= 3 ()) fis.ys2) =r 22). Howe we use the same idcar when we differentiate with respect to ne varable, wo teat tho osber two as constant, Thus Zpriy e294 LZ (°) = 2a 429) S02 2 = 04 29 Z (0) = 2a 4 23 Swirrmer2urasrnsr Swormergorw2=ra Bie a exdQyez! Sitve2ar diy +2) =P @=2 (9) fo.7.2)= 6+ NIE! x). Here we use the same idea, together with basic rules from ordinary differentiation: ye += C4 p= (quotient rule) ayy vy Se (chain rte)3.1 INTRODUCTION 187 Example 3.3 Solution ‘The partial derivatives /, and, ofthe Function fs»), with respect to rand y respee- tively, are given by the commands ware eLeeCE, 2 fy = @ifE(E, These commands can readily be exten to functions of more than two variables. [Also second-order partial derivatives can be obtained by suitably differentiating the firstorder partial derwavives already found. Thus in MATLAB the second-order partial derivatives of fr, y) are given by fax © GLEE(Ex,x) fey = dLEE(ERy) fyy = AlEE(fye”) fyx = alft ty.) Alterna tives can be oblained diccetly using the commands, Fuse = GLEE(E,,2) Evy = SLEECE,Y,2) ‘which eam be extended to higher-order partial derivatives, The eorresponding com ‘mands in MAPLE are fre ee dLEF[E x30 ly, the nonsmined deri fy en GLEE E x09) 7 ry as all Lys In Example 3.2 we used the chal (or composte-fencton) rule of ordinary Aitfrentinion af an dx dudx to otain the partial desivaive 7d. The malivaviable calculus form of the cain ale is alte more complete. ifthe variables u,v and ware define in terms of 9 and 2 inthe prt deivatve of ft ow) with respect ts adn. oe, ow Or duax” Wax” Aw Ox with similar expressions for 2/72» and 2 Find 67/@r and 67/@6 when. Tyee =arey" and ‘e080 and By the chain rule, ar _arae ari188 vEcToR catcutus Inthis example Example 3.4 nd di¥dr when (= sin(x= y) and x=20=3 ad y=$e=5e¢1 Solution We note that x and yare functions of r only, so thatthe chai rule Becomes Note the mixture of partial and ordinary derivatives. 1 i «function ofthe one variable 1, but its dependence is expressed through the two variables x andy. ‘Substituting for he derivatives involved, we have M4 = sfcos(3x— ye feostSx—y)]0- Att sfcos(ax— M4 feos@e— 9] - 5 = (I+ SyeosGr—y) = (1+ Syeos(? + 51~ 10) Example 3.5 A scalar point function (7) can be expressed in eri of rectangular cartesian coor ates (1, 2) orn terms of spherical polar coordinates (r, 6, 9), where sind cose,» sindsino, eos3.1 INTRODUCTION 189 Figure 34 Spherical, polar coords Example 3.6 Solution as shown in Figure 34, Find 2j/0s in terms of the parti derivatives of the function with respect tor, @and 0. Using the ebain rule, we have OH _ Har, HI, a0 ax” arex* 9095" [Link] From Figure 24,7? =27-4y? +24, tang = yleand tan@ = (x? + y°)!%, so that Thus 2 sino cove Su eorocost If ar sina aD ‘The Laplace equation in two dimensions is where « and y are rectangular cartesian coordinates. Show that expressed in polar co- ordinates 7, 6), where x= cos @ and y= rsin6, the Laplace equation may be weitten aw sno 22 Fine cos 6 Sitiarly 2 ern dM ax190 vEcTOR catcuLus Pe FH sig OH — Ht cos 04 2 Aaees Hence Lay, Lu Pu 94 ZU cowie 2TH sin a Bons tht =P sn'os Steer 6-224 sn Boas (@)_Trechain rue canbe reality nanced in both MATLAB and MAPLE. Considering ‘Example 3.3, in MATLAB the solution may be developed as falows ‘The commands OS = oxty 4 y*3y Tx = LEE(T os Ty = GiEECRA x = rtcos(theta); y = xtain (theta): we = AifE(xpel; xtheta = dife(nschetar; y cheta = diff (y,thetal: Te = tetae + Deter return = aiteueehs te = (3*x°2 ~ y)teos(tneta) + (-x + $*y"Z)*sinithetad ‘To substitute for cand y in terms of r and theta we make use of the eve function, with eval l™r) + pretty (ans)3.1 INTRODUCTION 191 ‘turing the answer (3r°cos(theta)* - rsin(theta))cos (theca) + (-eeos (theta) + 3:°sin(cheta)")sin(theta) ‘which readily reduces to the answer given in the solution. Similarly the commands Ttheta = Txtxtheta + tytythecal evel (ttheta); pretty lens) return the answor (-28"cos(thetal? + rsin(theta))rsin(theta) + (Creos (theta) + Sx"bin |theta)")rcos (theta) Which also reduces tothe answer given in the solution MAPLE solves this problem much more efficiently using the commarcs Tore beyl> x3 — xy + ¥73: Aire (r(ercos (theta), r*sin(cheta)), ri @LfE(T(e*cos (theta), r*sia(rheta)), cheteds collect (293 returning the answer (-3e09(8}%sia{0) + 3sin(0)%eos (6))= + (sin(@)? ~ c09 (6) 3.1.2 Exercises Cheek your answers wsing MATLAB or MAPLE whenever possible Sketch the concurs (in two dimensions) ofthe 5 Finda he fist and second pial derivatives of Scalar funtiens te functions () fasy)=InGe? +=) (b) fosy) = ty 09) fines Sketch the fo ines (i ho dimension) ofthe ier escleas 6 Find afar. where fa) Fine ate (oe =4ay7 (fe) ne oP nantes () Fisypayis (ha ere anc are unit vectors in the dieetion of the wand y aes respectively. 7 Find a//a» and af/dz in terms ofthe partial ecvatives off with respect to spherical polar coordinates (ro (S08 Example 3.5), ©) flp=sy2, and =e ysereesnzer ‘Sketcl the lve sures ofthe fnetiens 8 Show tha fue) =o) where Ps? 44s192 _vecToR catcutus then 2a rer ar sates the diferent equation ‘ence find the weneral form for (7). 9 Show tha: 5 ara ae ay de 10 Venty that Ys, 2)=sinbs coed cosh Se sais the dere equation a dy we 3.13 ‘Transformations Example 3.3 may be viewed as an example of transformation of coordinates. For ‘example, consider the transformation or mapping from the (x,y) plane to the (, 1) plane defined by sea, ey) ‘Thon a function w= f(s, ») of x andy becomes a function ‘wansformation, and the partial der on (5,0) of ond r under the 62) In matix notation this becomes au] [as a] fau 2:| las ax|las on a) | as at} au | La olla “The determinant ofthe matrix of the vansformation is called the Jacobian of the tans- formation defined by @.1) and is aboreviated 9 HO oesimply to J any) oan seth a) ja oy alle alle ol pede 4 on ay” |as ae) \ae at av oy| lax ar “The mati iets refered vo asthe Jacobian matrix ands generally expressed in a a & 3 ; vet tec torn |? | us sci plays an point ot in vaio pplication of ee ph: a a3.1 INTRODUCTION 193 mathematics in engineering, particularly in implementing changes in variables in multi= pile integrals, as considered ater in this Chapter ‘As indicated earlier (3.1) define a transformation ofthe (x, ») plane tothe (s, 2 plane and give the coordinates of a point inthe (s ) plane corresponding to a point in the (0) plane. If we solve (3.1) for and y, we obtain BeMeth ye Yoo as) which represent a transformation of the (5, ) plane into the (x, ») plane. This is called the inverse transformation of the transformation defined by (3.1), and, analogously to (8.2), we cam relate the partial derivatives by 6 where the suffix notation has beea used to denote the partial derivatives. Provided F40, ibis always tue tbat J) = J" or ov) iso Hs, 6) Oe ¥ IJ then the variables sand r defined by (3.1) are functionally dependent; that i, a relationship of the form J, £) = 0 exists, This implies a nen-unigue correspondence ‘between points in the x,y) and (s, ) planes (x,y), = s,3) then using MuPAD in MATLAB the commands delete x. ¥ Gee caer Mtn pars return the Jacobian matrix |? aH ax Oy The same result may be obtained with the Symbolic Math Toolbox using the commands facobian (Is, tle 0k, ¥)) or in MAPLE using the commands with (Vectorcaleatus) : Jacobian is, t], (x, ¥)194 vecToR catcuLus Example 3.7 Solution Example 3.8 (2) Obtain the icobian J of the transformation sedxty, tex-2y (b) Determine the inverse transformation ofthe above transformation and obtain its Jacobian J, Confirm that J, =". @ Co) x=1Oc+n, (= 29 The Jacobian ofthis inverse transformation is aan |} 2X0) [ 3 confirming that J, =". ‘Show that the variables x and y given by x on are functionally dependent, and obtain the relationship f(s, The Jacobian of the transformation (3.7) is Aa | 2 AD |x, y, ‘Since J= 0, the variables x and y are functionally related Rearanging (3.7), we have velth yatel so that @-DO-D=3.1 INTRODUCTION 195 v 12 13 4 aiving the functional relationship as Gry =0 ‘The definition ofa Jacobian i not restricted to functions of two variables, and itis reaully extendable to functions of many variables. For example, for fanetions of three vtables if Ubey.2% represents transforma 14,8 w then the correspon. Vout = bs, 2) as) in three dimensions from the variables x 5 =to the variables. fing Jacobian is ‘Again, if = 0, it follows that there exists a functional relationship ftv, ) =O between ‘the variables vend yr defined by (3.8). 3.1.4 Exercises ‘Show tha if.+ y= wand y =o, then 15 1 valve ofthe constant fer whieh ~ we ke sgt a0) vate we due Ser eOny ave funtionaly related and oben the ‘somresponding elation 1G Show tha itu = x¢s,») and f= €* cosu and =e" sav, obtain the evo a Jocobians £ a 2D gpg AD ‘wherein each case aa ae je He andl Sri tht they are mal inverse. ee) Find the vals ofthe constant parte for wie the feneons ae inetomally dependont 17 Use the resus of Exersise 16 1 obtain the partial envatives cos xe08y~ Asinssiny awww Oe OF On cosy t Leosssiny tee196 vecToR catcutus 3.1.5 The total differential Figure 25 Iilsraen of est Gay Consider function w =x») of te variables v and y. Lot Ax and Ay be increments inthe values of and y. Then the corresponding ineroment in is given by Au= ile +A y+ AY) —Je9) We rewrite this as two terms: one showing the change in 1 due to the change in x, and the other showing the change in x de tothe change in y. Thus AU= Lhe +8, AN) 9 + OPI L1G + API Dividing the first bracketed term by Ax and the second by Av gives From the definition of the partial derivative, we may approximate this expression by ae Lace oy We define the differential dy by the equation w= Lars Lay au= Lax + Lay 69) By seuting (x, ») =f. 99 =x and fi, v1 = £0.99 tum in (3.9), we ee that a= Hiser ay <0 aad &y=ay ‘so that for the independent variables incremten and iiferentials are equal. For the dependent variable we have Yan Hav Fay 010) We see that the differential duis an approximation to the change Ar in u = fx 9) resulting fom sel changes Scand Ain the indgpendet variables andy: ha wx du=2ay+ Hay = Hare Lay vn tus Lava ay = Larva Ga ares ita in Figuee 35, (eran tanta) wnat < | Eg ®3.1 INTRODUCTION 197 Example 3.9 Solution This extends to functions of as many variables as we please, provided that the partial derivatives exist. For example, fora function of three variables (9,2) defined by us fls,y 2) we have um du= Lara Lays Laz dun du = Lacs ays Ha Lav Hays Lae Have Says Ha: ‘The differential ofa function of several variables is often called a total differential, ‘emphasizing that it shows the variation of the Function with respect to small changes in all the independent variables, Find the total differentia of (x, ») Taking pani derivatives we have be ayo eat Beinn ae me ye Hence, ang (3.10, du = yah Indy Differentials sometimes arise naturally when modelling practical problems, When this ‘curs, itis often possible to analyse the problem further by testing to see if the expres- sion in whieh the differenials occur isa total differential, Consider the equation Pox, ypu + OCs, nav ‘connecting x,y and ther differentials. The left-hand side of this equation is said 10 be an exact differential if there isa function fx) such that Af= Pix yds + 00rd Now we know that - fae +80 if fix, ») exists then ray Z and ona Zl For fanetions with continuous second der198 vEcTOR catcuLus Example 3.10 Solution Thus if Cx, 9) exists then 2 > 42) This gives us a test for the existence of f(x»), but does not tellus how to find itl The technique for finding fa, y) is shown in Example 3.10, Show thet (G+ 9p + Md + Ox—4y 4 )dy is an exact differential and find the relationship between y and + given dy bros and the condition y= I when x= 0. In this example PO) 260499411 and OCG, 9= 91-43 First we test whether the expression isan exact differential. In tis example # wy ‘so from (3.12), we have an exact differential. Thus we know that there is @ function “Fes 9) se that a a Integrating (3.13a) with respect 10 x, Keeping y constant (that i, reversing the pani differentiation process), we have fix, y)= 30+ Say + Tse) oy) "Note thatthe ‘constant of integration sa funtion of y. You can check that this expression for fx.) is correct by differentiating it partially with respect tox. But we zlso know from (3.139 the partial derivative off») with respect toy and this enables us to find ‘(Differentiating (3.14) partially with respect to y and equating it @.13b), we have and oo oe soyet and 2 mor ays geet and 2 aoe—ay43 G.13a, b) Baye ao -dye Ha one aon-aye3 (Note that since gis a funetion of y only we use de/dy rather tan dy /2:) Thus dey a3 0, on integrating, at *4y+C ‘Substituting back into (3.136) gives fix,” = 30° + Day + Lx =2y? #3y+ C3.2 DERIVATIVES OF A SCALAR POINT FUNCTION 199 Now we are given that dy __ 6x 9ye dy 9endyad ‘Which implies that (6x4 Oy + Ide + Or~ ay +3)dy=0 ‘Whiet in tur implies thet BP 494 Ly 2 e+ C=O ‘The arbitrary constant C is fixed by applying the given concition aiving C==1, Thus.x and y satisty the equation Br? 4 94 y= 29" +3 when 3.1.6 Exercises 18 Determine which ofthe following are exact {she eet diferent oF fanetion ts, Uiferenials ofa fueon, afin, where pproprite, the corespo fa 0° #257 (b) Qn? +3ycos30)dx+ Gry +sin3ndy 6 (0) Gxy—y des Dee! we 19) Find the vale ofthe const (yeoee + Aeosy) d+ besing + sins yay Byyde + (129! —3spdy Beets Find he jm,» that also sisbes the Dade +eu+rjer 20. Show thatthe diffrent Lay? + Gay + 6y")de 408 day + 1S?) is not xa, but that «const can be chosen so thet ens 7H )) is equal t de, tho exact diffrent! ofa funtion 8.9) ind fs.) Asicht 3.2.1 Derivatives of a scalar point function In many practical problems itis necessary to measure the rate of change of a scalar point function. For example, in heat transfer problems we need 10 know the rate of ‘change of temipersture from point fo point, because that determines the rate at which heat flows. Similarly, if we are investigating the electric field due to state charges, ve need fo know the variation ofthe electric potential ftom point to point. To deter- ‘mine such information, the ideas of caleulus were extended! to vecior quantities. The first development ofthis was the concept of the gradient of a sealar point function The gradient of a scalar point function We described in Section 3.1.1 how the gradient ofa scalar field depended om the ditec~ lion along which its rate of change was measured. We now explore this idea further200 vEcTOR caLcutus Consider the rate of ehange of the function ir) atthe point (x y=) inthe direction of the unit vector Um, To find this, we need to evaluate the limit, tine ‘where Ar is inthe direction of (m,n). In terms of coordinates, this means re Aart Anion) (e+Ax. yt Ay, 24-42) sso that A Neer “Ths we have fo consider the limit tim Het Hr, vb mr, 2 nr) fxn a0 ar We can rowrite this as tim [Me lar, vor mar, 2 nr) ~ fx, ve mar, 2 nde], a" Tar + fim ELAM BEI h Evaluating the limits, remembering that Ar = JA and so on, we find thatthe rate of ‘change of fir) in the diretion ofthe unit veetor (mm) is BLL (Z FD mn Shee (ESE me ect (Sa) ae seh ant of he ears tin i 2 is tear by Vf, where V is the vector operator : 2 v ig 2472. Ox 105 ‘whore fj and & are the usual triad of unit vectors “The Symbol Vis called “ge” or sometimes “nal. Then @22) ws rat) =0/ = Hi yo Thas we can caleutate the rate of change of x» 2) along any direction we please, If ‘isthe unit vector in tha dieetion then, (end f)-a3.2 DERIVATIVES OF A SCALAR POINT FUNCTION 201. Figure 3.6 (@) Adiacnt. evel surfaces of fr} bya facts rma 19 the surface le) = Example 3.11 Solution ives the required directional derivative, tht is the rate of change of fl) 2) in the direction of a, Remembering that ¢-b = |aliA|cos6, where @is the angle between the ‘io vectors, it follows that the rate of change of fla, y, 2) is zero along diections per- peradicular to gradfand is maximum along the direction parallel to pra: Furthermore, ‘radf acts along the normal direction to the level surface of fx, 2). We ean see this Dy considering the level surfaces of the funtion corvespending 10 ¢ and c+ Ae, a8 shown in Figure 3.6(2). In going from P on the surface f(r) = ¢ to any point Q on J(e)= c+ Ae, the inerease in is the sime whatever point Q is chosen, but the distance PQ willbe smallest, and hence the rate of change of fx, v=) greatest, when Q lies on the normal fi tothe surface at P. Thus graft P is in the direction ofthe outward normal ‘it the surface f(r) = u, and represents in magnitude and direction the greatest ate of increase of fx, 2) with disance (Figure 3.6.6). Is fequently written as = Xi ana f= a where dan is referred to asthe normal derivative 1 the surface fir) = c, wy Sir)=e+ be r fiv=e @ o Find grad/ fore) = 347-4 2p? + {a) the directional derivative of f() at (1, 2,3) in the dreetion ofthe unit vector 12,2, Ds (6) the maximum rate of change ofthe function at (1, 2,3) and its dreeion. the point (1, 2, 3). Hence caleulate (a) Since 9//2x= Gx, 9//9y= 4y and Bffaz = 22, we have from (3.15) that rad Gu + 4y)+ 228 At the point (1.2, 3) grad f= 04+ 99+ 64 Thus the directional derivative offi) at (12,3) inthe direction of the unit vector (61+ 874 64) b) The maxinum rae of change ofr) at(1, 2.3) occurs along the direction parallel vo grad/at (1, 2, 3; that is, paralet 10 (6, 8, ©). The unit vector in that direction | G.4, 3)/34 and the maximum rate of change of fir) is |grad f1= 2,34.202 VECTOR caLcuLus Example 3.12 Solution Figure 3.7 Tangent plane at paraboloid 22 3,5) the I surface in three dimensions is specified by the equation fx.» 2)=¢, oF equival- ently ft) =<, then grad fis a vector perpendicular to that surfie. This enables us to calculate the normal vector at any point on the surface, and consequently to find the {equation ofthe tangent plane at that point. ‘+9 Find the unit normal veetor to the the equation of the normal and the tangent A paraboloid of revolution has equation 2 ‘surface atthe point (1,3, 3). Hence obs plane to the surface at that point A vector normal t the surface 2 x? 4 ys given by grad(? +p! =22)= 201+ 2-24 Atthe point (1, 3,5) the vector has the value 27 + 6f ~ 2k. Thus the normal unit veetor atthe point (1,3, 5) fs + 3)~ A)/11. The equation ofthe Tine through (1, 3, 5) inthe rection of this normal and the equation ofthe tangent plane is (= + KY=3)4 CDE-S) which simplifies tox + 3 (see Figure 3.7) ‘The concept of the gradient of a scalar field oceurs in many applications. The simplest perhaps, is when fir) represents the potential in an electric field due to sta charges. Then the electric fore is i te direction ofthe greatest decrease Of the pote ‘al. ls magnitude is equal to that rate of decrease, so that the forces given by —grad Using the Symbolic Math Toolbox in MATLAB th gratin grid fo the cla fan tion fis,9.2sgivenby the o components separately (Consider the vector field 9). To find the flow around an axis in the x direction at the point r, we take an elementary rectangle surrounding r perpendicular tothe x direction, as shown in Figure 3.10 To measure the circulation around the point r about an axis parle! to the x direc- tion, we calculate the flow around the elementary rectangle ABCD and divide by its area, giving 6 met whe A) 2Ay) + e4n,9 + Ay, 2*\2A2) ay) ~ Hy ~ ay, 2)202))(4004 where 34 9 €(y — Ary + Ay) Roarranging, ws obtain 4 As) and v= od + oy + 0 “ns66, 824+ nus, *)— n(s,y~ Bes 2)V2AY) + [eis Ay, Proceeding 10 the limit as AyeA= —> 0, we ebxain the x component of this vector as a. e By similar arguments, ws obi tho yand = components as, aoa ee respectively The vector measuring the rotation slbout a point in the fluid is called the cust of 265, 3a, Oy, 26, dey PES 223) oun I may be writen formally as k a Z| Gis S Yos ‘or more compactly as culv= Vx208 VECTOR caLcuLUS Example 3.14 Find the cutl ofthe vestor » = Qe —y*, 32+ x4, 4 ) at the point (1, 2,3) Solution Here y= 2r—yny= 36+ 33, y)=4y i oj ok 2 2 2 & so that 2ar-y gor-r] S143) ~AO~0) + Ae +20) =I HME ‘Tins, a the point 1,2,3).V xv= (1,0, 6), 2 a 2arexy-2 r[Bos )-2e As# More generally, the component ofthe curl of veccor fel F(r) in the diecton ofthe ‘unit vector A ata pont Lis found by enclosing L by an elementary area AS thats perpen Aicularto sin Figure 3.11, ad caleulaing the ow around AS per unit area, Thus yy Nou A (url F)-A = ji HO Figure 311 Geeaion avund ner way of visualizing the meaning ofthe eur of a vector isto consider the " motion of a rigid body, We can describe such motion by specifying the angular velocity ‘ofthe body bout sn sxis OA, where O isa fied point nthe by, together with he teaslatona! nea veloiy# of set. Then at any point P inte Do the velocity uiszivenby uavtaxr a shown in Figure 3.12. Hore and ae independent of (2). Thus calla = cure + cut (@x 7) =0 teu (@%r) Figure 32 » Retain sf rete F ealsion velocity3.3 DERIVATIVES OF A VECTOR FOINT FUNCTION 209 The vector @ x ris given by WX P= (0, Oy 0) X59 ay¥ + (ay Op)j+ (@y— Oh and 1 J & -| 2 2 az curt@xr) =| 2 s = =2ani + 20,) +20,6=20 Thus cule that, beurla Hence when any rigid body isin motion, she curl of is linear velocity at any point is ‘nvice its angular velocity in magnitude and has the same direction, ‘Applying this result tothe motion of a fui, we can see by regarding particles of the fluid as miniature bodies that when the earl ofthe velocity is zero there is no rotation of the particle, and the metion is ssid to be eurlfree or irrotational. When the curtis ‘non-7era, the motion is ratationl Using MuPAD in MATLAB the command 1snaig +2 cori, x) computes the curl ofthe three-cimensional vector field» with respect to the three-dimensional vector x in cartesian coordinates. For example, the curl of the vector v=Qr-y3r48ay-2) considered in Example 3.14 is given by the commands delete x,y, 24 Linalg #4 curl ({2%* -y fe ys 21) Bee + x12, ty - 2°21, 1 | | ee TEMATEE oats mane ser210 VECTOR caLcuLus 3.3.4 Exercises 38 Find w= curl when o= Gre" +20) 43. Mo=-yitaj+ekisthevlociy vectorofa Maid, find the lel value ofthe angular velocity athe 39. A vector il isdefined by v= (rx 47). Show point 3, 2. that curio = 0, 44 tne vetocty ofa ida the oie, 2)sgiven by 40. Show tha y= (x +), 29 +24, 22+.) the em (art bot e(oxs day ‘Sta w= al 9 oc ak = ea Sind the conditions on the constants. hoard din oder that 41 By evaluating each vem separately, verity the idanity iv cut Verity ttt in tis ease Vx M=anVxvi HV Kv = Lg1ad(ac’ + 2h") for fir) =x? =pand a0 9, 45 (Spherval pols coowtinates) Using the nota 42. Find conan, bandc ih athe esr Bld inredoed a Evers 90 show tet sbinciby ° sen Father och at By Fk i ta Fay tr romion, Wi he vases of bad wLl2 a 2 dkcrmie scl icon 2) ch Finale 39 30 Pvp i fe, eh 3.3.5 Further properties of the vector operator V So fir we ave wed the vector operator in thee ways: Vp = weds = Mes ye Me, fie) scalar ld Viradv r= Fs BS For) avector field a Vx = cul F -(¢-% i+(B-B)(B Be. FOP) avector field ‘A Turter aplication iin determining the dirstional derivative ofa veetor Held = (ob rad ead )e «(Brera e(a A vnih nn (ora vad3.3 DERIVATIVES OF A VECTOR FOINT FUNCTION 211 ‘The ordinary rules of cifferentiation cary over to this vector differential operator, but ‘hey have 1 be applied with care, using the rules of vector algebra. For no-onthogonal ‘oerdinate systems specialist texthook should be consulted. Thus fer scalar fields f(r), g(r) and vector fields (7). (7) we have Vise) = VE fineir = (WA +/1NVetr) Vue) wir = 0 x0 %u) + ux (0x0): V+ Vw Wr fPu WyxurfVxu Ve luke) x ery] =0-(P xu) — we (Vxe) Vx (ade) > ofr] = (0 Vou 0(V-) — (#Vpo + Vw) Ve ag Higher-order derivatives can also be formed, giving the following: oe, af, a ivfarad r)]= VV L. beet ar af at Where W? is called the Laplacian operator (sometimes denoted by A) curl [rod fin)] = Vx Yfin) =0 oy ol Be ae when all second-order derivatives of f(7) ate continuous; divfourl o(r)] =¥-(W x0) =0 grad (div ») = WV») = (Big + 12 (Be Fe-(Z +2 Zotroiran curl feu any] = Vx (x0) = VV-0) = Vv G09) G9») 6.90, 6.19%) 6.490) G91) 6.19%) G20 aan a2 aay om a2,212 vecToR catcutus Example 3.15 Verify that Vx (V sv) = V(V-r) ~ Ve forthe vector field » = (3x2, yz, + 22). ‘ Solution 2 (y, 62-1, 0) ® a exe + 2¢e422) = Vom 26 Serr VV 1) =(0,0,62-1) Viv = (VG), Vi), VG +22)) = (6x, 0,0) Ths WV) — Vir = (65,0, 6 = VX) ‘Similar verifications for other identities are suggested in Exercises 3.3.6. Example 3.16 Maxwell's equations in ive space may be written, in Gaussian units, as (@) divit=0, (by divE=0 Lae, (©) cul = Vx (@) ewlE= Ve: where c is the velocity of light (assumed constant), Show that these equations are satisfied by a nal2mndgxk B= cor" i 122+ 2 gto bere 6 sates np bdo vee k ar and Kis a unit vector along the = axis, 12 gad px Solution () H=12mndgxk £2 urcuntigadd)— (ead g)-cur kh fom @.194) By G 21), curl(grad 6) = 0, and since & isa constant vector, curl k= 0, 80 that av3.3 DERIVATIVES OF A VECTOR POINT FUNCTION 213 » ‘dy grad @ al by (3.20) have fo cute 1Zewientoxe = 12 1a Vgrate he (divgndo)~ (gnde-Ve+ gmdg(F 4) from (3.198) 19/2 era 9 since his a constant vector 13 rad RV" ‘ 6 owe ) + Zant ane 9 |. since cu grad §= Dby (3.21) IS ale =214 vecToR caLcuLus a7 50 3 3.3.6 Exercises Show tha ig isa fantion ofr = Ux 2) hen = tds, wade Fr Deeduce tha fw is veto ed hen @ Ve andy) carlowte Show tha position veeber= (x vector and isthe 2) hen sv (grad ("1M = 100) Verify te identity Vee, by calculating each ern separa, the eres iv ux p= e-ear a weer Fis the usual position vector r= (1,9, show dat o conten) el lk end!) 52 53 54 55 Aina constant vector and este position vector (a )42) show that, (22) tt ( 22) = 24 Biaxerxe 3n, frist poston vector ¢= Cry, 2)and wand are constant vectors, io dat f@) Vxr=0 ©) @ Vena fo) Vel 96 (bal (@) Ve(a-nb— tbr) ) By evaluating ¥- Ifthe eurve CAD is represented by = hy(y) and the curve CBD by x = h(, we can write the integral as [frre is) -| if Jost UF the dbl integral eviss the hee two fess ae egal andi going frm one the cher we have changed te order of negation. Note ta hein of negratin are also changed in the process. Often, when evaluating an integral analytically itis ‘easier to perform the evaluation one way rather than the other. Zrome ay Evaluate fJq (0° + y")d over the triangle with vertices at (0, 0}, (2.0) and (1, D. ° ‘The domain of integration is shown in Figure 3.28(a). The triangle is bounded by the lines y= 0, y=.xand y=2 =x,222 VECTOR CALCULUS (0) Integrating with respect tox fist, 28 indicates in Figure 3.23(b), gives forum “| Uv asrelgyay fedveayt fy ldyet indicated in Figure 3.23(6), gives reff ee [Note that because the upper boundary of the region ft has different equations for italong cifferent parts, the integral has to be split up into convenient subintegras, Evaluating the integrals we have [[eeruael rewire Ne fo) Integrating with respect toy fist, fies ff + asbefore ‘Cleary, in this example iti easier to integrate with respect tox first Example 3.21 Evahnte Jfe(x+ 29) "4 over the region x— 2y< Landy > y+ |. Figure 3.24 Domin3.4 TOPICS IN INTEGRATION 223 Solution Example 3.22 Solution than y= Gr~ 1)!, Thus we obeain [Jussrene cea ecw I (2(0+2y)"IE dy A(4y+ 1)? ort Diy AAs indicated carie, the evaluation of integrals over a domain R is not restricted to the use of seviangular cariesian coordinates (x, 3). Example 322 shows how polar ‘coordinates can be used in some eases to simplify the avalytical process. Evaluate ffpx7yd4, where R is the region x? + y? <1 ‘The fact tha the domain of integration is ac natura choice forthe integration process. Then, from (3.26), 44 = rd6dr, andthe integral becomes Jem ff settee Je suggests that polar coordinates are & cos 0,9 ='sin 6 and F cos’ rind rd0dr Note that in tis exampte de integration is such that we can separate ihe variables and and write [ml [aosnsee Furthermore, since the limits of integration with respect to @ do not involve 7, we can fete eae224 VECTOR CALCULUS and the double integral inthis case reduces 102 product of integrals, Thus we obiain Jer a= (Ub cos 6) = Reflecting on the natu ofthe integrand and the domain of integration, this isthe result ‘one would anticipate (@)_Thereare several ways ofevaluaing double integrals using MATLAB, The simplest uses the command dbiquad (£, x» ¥y» ¥)+ For example, consider i I G4 y)drdy Here we define the integrand es an Stine function f= inline (-x.-2 +2", °x7, '¥") (Note that cis taken at a vector argument) I = dbiquad Pe os! returns the answor r= 16 For non-rectangular domains, the same command is used but the integrand is ‘modified as shown below. Consider I I (ety de dy from Example 3:20 (b). Here we define the integrand as the in f = Intine (7(x.°2 + ¥°2)." =x ne fimotion Sn Cay Ay there the logical expression {y.~ x - O,y> Oandety< 1. ‘ketch the demain of ategraton and evaluate © ff cone evaluate ffsin nox +) dr dover the triangle Winse vericesare (0,0), 2,1), (1,2). ‘Sketch the demains of integration ofthe double imegrals w feof 2 Change the order of integration, and ence evaluate the integrals. Evalue ‘ketch the domain of integration ofthe double imegral I 73 74 76 ‘Express the integral in polarevordinates, and heres stow that is yal is | ‘Sketch the domain of integration ofthe double integral J de [ ate ay ody We and eval the inegr “lia ‘overtheportion ofthe fst qudran ving inside the sicker ry =e Dy using polar coordimes, evaluate the double integral ‘vertherepcnin the fist quant bounded hy the are fof the panboiay* = (1 ~ 1) andthe eoowdinte axes ‘By rasfonning to polar coordinates, show that dhe double negra taken over the area common othe two circles Shy anand. +)" shy sah 3.4.5 Green's theorem in a plane ‘This theorem shows the relationship between line integrals and double integrals, and will also provide a justification for the general change of variables in a double imegral Consider a simple closed curve. C, enclosing the rezion 4 as shown in Figure 325. If P(x.) and Q(x») ate continuous functions with continuous partial derivatives then frssom (e@-B« en226 VECTOR caLcuLuS Example 3.23 Solution “where Cis traversed in the positive sense (that is, so that the bounded area is always on the lo). This result is called Green's theorem in « plane. “The proof of this result is straightforward. Consider the first term on the right-hand sido, Then, with reference to Figure 3.25, {jee0-{[fl e+} frees )- Oats). dy {anno [oun oes, rrar=f 9 Simitary, Co ‘An elementary application is shown in Example 3.23. Bvahute f [x(x +) de +(e! +974 9°) dy] around the square with vertices a (0,0), (1,0) (1) and Q, |) illustrated in Figure 326, Here Px, y) = 2ute + y) and Ot and 90/8 aP/ay =y. Thus the line ay + y4 90 that P/Q) = 2, AQIAr = 26+ y egeal tansforms into an easy double integral3.4 TOPICS IN INTEGRATION 227 oH Tot Figure 3.26 Path of inezration for Example 323. ifilows inact fam Green's therom (327 ht hr are olen by he decane eget) foveeran [fer forf Suppose that under transformation of coordinates + becomes “enclosing an area 4’ Then f(ForrZe) Blew (04 v) andy = (9), the curve ‘de ‘This implies that de element of area du-do is equivalent o the element (5-35) Here the module sign is imtrodaced to preserve the orientation of the curve under the mapping. Similarly, we may prove that ar dy 28)228 VECTOR CALCULUS Solution Figure 327 Domain of integration for Example 324: yin tho x9) plane: () inthe (a) plane. “This enables us to make a general change of coaninates in « double integral: (=a 5 ‘where 4s the region inthe (u, 0) plane corresponding to 4 inthe (x,y) plane Note that the above discussion confirms the result as shown in Section 3.1.3, Using (3.29), the result 3.26) when using polar coordinates is readily confirmed, Evaluate {xy dv dy over the region in.x > 0, > 0 bounded by y=? +4, ye 6x and y= 12-2 ‘The domain of integration is shown in Figure 3.27(a). The bounding curves can be rewriten as y—2'= 4, y+? = 6 and y-+5°= 12, so that a natural ebange ‘of coordinates is to set Under this transformation, the region of integration becomes the rectangle 6 = «= 12, (0. v4, as shown in Figure 5.27(b), Thus since ron 3d ea] ok the integral simplifies to I scape I) wade3.4 TOPICS IN INTEGRATION 229 Figure 228 ‘We remark in passing that Green’s theorem in a plane may be generalized to three sions. Note thatthe result G.27) may be written as fer 2.9 r= |f [email protected] kavdy For a general surface $ with bounding curve C as shown in Figure 3.28 this identity becomes frioare J | eo Fin) dS where dS = dS is the vector element of surface area and 4 isa unit vector along the normal. This generalization is called Stokes" theorem, and will be diseussed in Section 3.4.12 after we have formally introduced the concept ofa surface integral, Tteesimensional generaliation of Green's theorem, 3.4.6 77 evaluate the i f taken sures Exercises ine ‘ety yar anver ng Geter nine 78. Use Gres thonm na plane torte Isinyde (e-em 981 . fier narrerrian the anctocovise sense, where Cis the e perimeter ofthe rangle fore by the nes 8a double intr, where Cis the tangle with verties a (0, 0), (2, 0) nd 2, 2) ads reversed x0 in the anticlockwise diction230 VECTOR caLcuLus 79. Evaluate the line integral 81 : fos acted) ete Cis the clsed curve consising ofy fom s=Oton= Landy = x fom x= 1 sng the tansfomaton oF esordinats a = +», 6. Conti your anewerby applying Green's theorem 82) Using the taneormation inthe plane and esluatis {80 Use Greens theorem ins planeto evaluate the line imesrl where Ci the cree x! + y to evaluate te double integral.) nor fie arervce seein fel "asa doable mera. show tha (int se pote 347 Serta) Qin) Girne Figure 229 Params canes on a surfce Surface integrals “The exiensions ofthe idea ofan integral to line and doubse integrals are not the only ‘generalizations that can be mae, We can also extend the idea to integration over 8 ‘general surface S. Two types of such integrals occu: o [foc 0 {f ro-4ssff reas In case (a) we have a scalar field f(r) and in ease (b) a vector field F(x). Note that AS = sid is the vector element of arca, where fis the unit outward-drawn normal ‘vestor tothe element dS. In general, the surface S can be deseribed in terms of two parameters wand » say, sotthat on $ uso) (5 UW), a0) “The surface 5 can be specified by a scalar point function Cle) = c, where ¢ is a ‘constant, Curves may be drawn on that surface, and in pariular if we fix the value of fone of the 1wo parameters wand v then we obtain two families of curves. On one Gra, v9) the value of w varies while v is fixed, and on the other, Crt 0), the value of v varies while w is fixed, as shown in Figure 3.29. Then as indicated in Figure 3.29, the vector element of area dS is given by3.4 TOPICS IN INTEGRATION 231 Bo) (MB 2) ang = nits j+saduan Se Se Se) Ge FE) ites estan whore eae _ ay de Ou ov dv du B30) Hence [eo-as- [fees von esanae [roses ffraoruresissiraeae where F(0)= (P, Q, R)and 4 isthe region of the (x 1) plane corresponding to S. Here ‘of course, the terms in the integrands have to be expressed in terms of 1 and. In particular wand v canbe chosen as any two of, yand =. For example i deserbes a setace asin Figure 3.30 then 26,9) with x and yas independent variables. This oy) and [feos eats [fro [f rosin sff-Qy-BJ]os eam Figure 3.30 Asurice desorbed by Peat232 VECTOR CALCULUS Example 3.25 Evaluate the surface integral few ‘where J isthe portion of the sphere x? + «har ies inthe frst quadran. () quadrant of ‘eck inte 9) Plane. Winey) 6.0.0 © o Solution Using @.17) then gives fferrenasefcerse- 1 ava aera ‘where dis the quacrant ofa circle in the (x) plane ilustrated in Figure 3.31(0).3.4 TOPICS IN INTEGRATION 233 An altemative approach to evaluating the surface integral in Example 3.25 i o evaluate it drvetly over the sursce ofthe sphere using spherical polar coordinate, As illustrated in Figure 3.32, on the eurface ofa sphore of radius we have x =asin @ cos, sin sing zsacos, dS=a"sin9dodd Figire222 Surface : clement in spherical 00 poltr coordinates aaa In the sphere of Example 3.25 the radius a= 1, 0 that formas (sin @ cos9+ sin@ sing + cos 6)sin@ deed [}rrcasg+}nsing+ c= as determined in Example 3.25.