Scribd Logo
Carica

Benvenuto in Scribd!

  • Calculo Del Sistema Cartesiano

    Caricato da

    Manuel Alberto Peñaranda
    9 visualizzazioni38 pagine
    Calculo de los sistemas tensores.

    Copyright

    © © All Rights Reserved

    Formati disponibili

    PDF o leggi online da Scribd

    Hai trovato utile questo documento?

    Questo contenuto è inappropriato?

    Segnala questo documento
    Calculo de los sistemas tensores.

    Copyright:

    © All Rights Reserved
    Scarica in formato PDF o leggi online su Scribd
    Segnala contenuti inappropriati
    9 visualizzazioni38 pagine

    Calculo Del Sistema Cartesiano

    Caricato da

    Manuel Alberto Peñaranda
    Calculo de los sistemas tensores.

    Copyright:

    © All Rights Reserved
    Scarica in formato PDF o leggi online su Scribd
    Segnala contenuti inappropriati
    di 38
    Cartesian Vectors and Tensors: Their Calculus 3.11. Tensor functions of a time-like variable In the last chapter we considered only the algebraic manipulations and relationships of tensors. We now want to find out what we can say about the behavior of tensors when they arefunctionsof continuousvariables. Certainly this will be necessary in applications for if we think of an unsteady Aid ‘motion we realise that we shall have to consider its velocity at any point or time. Now velocity is a vector so we shall have a vector v whose components are functions of the coordinates xj, xq x and time f. Any variable which is independent of the coordinates can be called time-ike since in many applica- tions it wil be the time. We will consider fist tensors whose components are functions of one such variable, ¢ Suppose Ay, 4,40 is a tensor function; then it is clear that all the derivatives of it that exist are themselves tensors, for the / of the trans” formation are independent of ¢ and we may differentiate the relation Af = Nal ASD Gut) ‘as many times as we are able, to give AMO = habe ZA, (01 Gut) ip lala fey » BL Curves in Space 2” We may integrate the tensor components and preserve their character ‘equally well "The most important example of this has already been cited. If x, the position vector of particle, i function of time, then its first derivative is the velocity | YOK, tym daly Gu and its acceleration is a) =8(), a = dela ans ‘The differentiation of products of tensors proceeds according to the usual ules of differentiation of products. In particular, 4 a b+a 2 2, anny dt dt ¢ > y= and a a * Sarma tary ane. 6.119 Exercise 3.1.1. Show that if @ particle moves with constant speed its acceleration is normal to its velocity. Exercise 3.11.2. Show that te acceleration ofa particle moving in the surfece ‘ofa sphere of radius rasa radial component —/r, where visits speed. Exercise 3.11.4. Wf a(), Mi, and e(?) are thee mutually arthaganal wnit ‘vectors, show that ther fist derivatives ae coplanar. Exercise 3.114. Show that if the position vector of particle its velocity, and its scenleration are coplanat, then ll the higher derivatives ara inthe same plane. Exercise 2.115. Show that x, éx{dt) =O is the condition that x() should ‘emai parallel to itself Exercise 3.116. Uf ¢;is the component of a vector a with respect to a system ‘Of base vectors ba, By. by show that bg = Pale 3.12. Curves in space ‘The variable position vector x() describes the motion of a particle. For finite intervalof1, saya S Sb, wecan plotthe position as a curve in space, I the curve does not cross itself (that is, f (0) # x(t), = 1-< 1 5 B)itis ‘called simple; if x(a) = x(6) the curve is closed. ‘The variable tis now just 2 ‘Parameter along the curve which may be thought of as the time in the motion 0 care lan Vectors and Tensors: Their Celevlue (212 of the particle only if such picturesqueness* is desired. If and ¢’ are the ‘parameters of two points, the chord joining them is the vector x(¢’) — x(). ‘As 1—# this vector approaches (#’ — #)&(¢) and so inthe limit it s propor: tional to &(). However, the limit of the chord isthe tangent so that &() is in the direction of the tangent. If s? = +S we can construct a unit tangent vectors = fe. If x( and x(¢+ dt) are two very close points, Mt +d) = x00 + dt C0 + OCd* 7 and the distance between them is a = (ate + a) — x(0)} + (att + a — x(9} 2 X10) 50 dt + 04, @ ‘The arc length from any given point 4 t= ais therefore re x= [it80 see ar. 124 ‘is the natural parameter to use/on the eurve, and we observe that $-45-0/f ‘A curve for which the length can be so calculated is called rectifiable. From this pont on we wil repr s athe parameter, ening wth sanding the dot denote differentiation with respect tos. This 1) = 509) 129 isthe unit tangent vector, (Cr. Fi. 3.1) Let (9), x(¢+ and x(e ~ db thse neaby points onthe curve, The plane containing ther must also contain the vstrs EE 28)_ a6 + 088) me EM RB EHH wy 4 949 ‘Thus, in the limit when the points are coincident, the plane reaches a limiting position defined by the frst two derivatives (1) and 38). This limiting plane {s called the osculating plane and the curve appears to le in this plane in the + For en amusing use ofthis word by pare mathematician se the prefaet9 E. C. ‘Titchoars'e "Theory of Fourier Integrals.” Oxford: 1837 3.122) IL Curves in Space 4 immediate neighborhood of the point. Now X= so R= + and since wre sta0 ons so that the vector + is at right angles to the tangent. Let tee ens and ot 6.12.6 ‘Then v is a unit normal and defines the direction of the so-called principal normal to the curve. ‘To interpret p, we observe thatthe small angle d0 between the tangents at s and 5 + dis given by 1 cos = x15) -a(5 + a) or [MOE eae pets t petite. SIA Dd + since += 0 and so t-#4 += 0. Thus, 3.12 2-9 12.7) is the reciprocal of the rate of change of the angle of the tangent with arc length, that 15, the radius of curvature. Its reciprocal I/p 1s the eurvature. ‘A second normal to the curve may be taken to form a right-hand system with + and v. This is called the unit binormal, Ba=ray, G28) Since 8 = 1, 6-8 = Oand (is at ight angles to. However, = Oso sep ‘p= Oso Bis also at rghtangles to © and so must be inthe direction of v. Let b= vio G.129) ‘where l/s a scalar known as the torsion, Clearly, 1/o isthe magnitude rate ‘of change of the direction ofthe binormal, just as I/p was the rate of change of the tangent. Further, since v= 8 A, we have sapart pata —tvarttpay @.12.10, 2 Carcesian Vectors and Tentors: Their Calculus §212 ‘These three formulae, @.2at) are known as the Sertet-Frenet formulae. It may be shown that if two curves have the same dependence of curvature and torsion upon are length, then they are the same curve apart from some translation or rotation in space. These two functions p(s) and o(8) are the intrinsic equations of the curve. Exercise 3.12.1. Interpret the curve given by c0s(sc0s ala). x. xyorssin ‘and find its curvature and torsion. asin (s cos ala). _Everoe 5.12.2, Show th ifthe tangent toa curve makes constant angle With a fixed ditetion then the ratio ofits curvature and torsion is onatnt. Such curve i called «helix Everise 3.12.5. Prove that ptok «(8 ®X) = Ian hence tha he curve es in 1a plane if (A) =O. 343. UI If F(%y Xe 3), of more briefly F(x), is function of position and C is the arcofa simplecurvex = x(1),a Sf S b, wecan define the integral of Falong Cas [Fedde = FEO, 00, x40] de (3.13.1) provided this second integral exists. Ifthe curve Cis composed of a number ‘of arcs which have to be given by diferent equations, then an integral like the right-hand side of @.13.1) must be calculated for each are. If and Bare the end points of the curve (given by ¢ = a and ¢ = b, respectively) the integral above isthe integral along C from A to B. The integral in the opposite ‘direction from B to is obtained by reversing the limits and therefore has the same absolute value but the opposite sign. If x(@) = x{b), the curve C is closed and the integral is sometimes writen Feo ar (IL Line Incegrale 8 ‘An important case, and one which we shall frequently use, oocurs when the parameter ¢is the arc lengths. J, Feo és = [etxtoytsco - co} at. 13.2) In general the line integral depends on F, the two end points and the path ‘between them. If, however, the integral around any simple closed curve vanishes, the value of the integral from 4 to B is independent of the path. To see this we take any two paths between A and 2, say C, and Cz, and denote by C the closed curve formed by following C, from A to B and C back from ee rare Ltral + Ltr = [tr as), ~ [[ras] es @.B3) and this vanishes by hypothesis, so that the integrals along the two different paths are equal If a(x, X54) is any vector function of position, a + is the projection of a tangent to the curve. ‘The integral around a simple closed curve C of a+ tis called the circulation of around C, fac ds Jason x46), fa] 740) ds G34) Since dx ds = x this integral is sometimes written fa, dx, or fads we shall prefer however to write the integral with the unit tangent vector explicit. rece 3.131, Evalue the ioegn fae db whee Cit the ci of radius @ and center the origin lying in the plane m-x =0 and 4 = Ax. Fis the unit tangent to C. iple closed curve in the plane O12 then Exercise 3.13.2. Show that if Cis it encloses an area Saleh factemc tl lie aoe all the integrals being taken around C. Exercise 3.133. Cis any simple closed curve in space and tits tangent; C, 4s the projection of C on a plane O12 and C, is also a simple closed curve. Show that the area of C, is given by the same formulae as in the previous question. “ Cartesian Vectors and Tensors: Their Caleulus $3.14 3.14, Surface integrals ‘We shall need also to consider integrals over surfaces and should say something about thet construction and the surfaces for which they can be constructed. A closed surface is one which ies within a bounded region of space and hasan inside and an outside. We can pass from any inside point £0 ‘another inside point by acurve which does not cross the surface and, sary, from any outside point to any other. However, to pass from an inside point to an outside point the path must cross the surface. Fanliar examples of closed surfaces are the sphere and surfaces that could be deformed into a sphere. A continuous surface which has no inside or ouside, known asthe Klein bottle, is shown in Fig. 324. These pathological surfaces are the happy hunting ground of the topologst; they erve to preserve the engineer from becoming complacent in his assumed aormality. Ifthe nocmal to the surface varies continuously over a part of the surface that part is called smooth, Some closed surfaces (gee for example Fig. 32a) are smooth everywhere, others are made up ofa number of subzezions which aresmooth (ig. 3:26) and are called piece-wise smooth. A closed curve on a surface which can be continuously shrunk to a point is called reducible, as for ‘example the equator of sphere which can be continuously moved to any line fo late until it shuiaks (9 « pole. Ifall closed cuives on a suiface ate Feducible, the surface is called simply connected. The sphere is simply ‘connected but the torus or anchor rng isnot, fora closed curve such as is shown in Fig. 32s isnot reducible. fa surface is not closed it normally hase space curve as is boundery, as for example a hemisphere withthe equator as Boundary. It has two sides iit is impossible to go from a point on one side to a point on the other (in Dttcular te point, which fs jest on te other side) omg 4 continuum cute that does not cross the boundary. The surface is sometimes called the cap of the space curve. Again itis not nesessary for a surface to have two sides; the ‘Mobius strip (Fig. 32) isthe well-known example ofthis. Indeed a closed curve could be capped by ether aone- ora two-sided surface forthe Boundary curve of the hemisphere (Fig. 3.22) would be distorted into the boundary of the Mobius strip and the hemisphere would be a two-sided cap. These sophisticated considerations need not daunt us however, for given a simple closed curve we can imagine a soup film across it which can then be distorted into a pice-wise smooth cap. Just asthe line integral is @ natural extension ofthe common integral over an interval, 0 the suface integral san extension of the double integral. The double integral over a region R in the O12 plane of a function F(x, x) is weiten Jfreu x0 axe, a 4“ Cartesian Vectors and Tensors: Their Calculus (3.44 andi constructed as follows. ‘The region R is divided into a large numberof small areas by a ged of lines xj = ay in -+-v %1= By By oo By which over the region (ee Fig. 33). Ite cori wy (Pil rectangle, ay tn <% 5 Gp Bey ES by which is wholly of panly in Ry we may select a point (yin this and in R ay and evaluate Fat this point. Then the nm sum over all the small areas E Ray A) dSn where d5,, is the area of the part of the rectangle which is within R, is an approximation to the integral, We now let M and N increase without ig. limit but abvays insist thatthe largest 0 ae erase" subdivision of area d,, must tend to Fig.33 z2r0, then if the sum tends to a limit that ‘limit isthe integral. [Now if Sis piece-wise smooth surface with two sides in three-dimens space, we can divide it up into a large number of small regions by a grid much the same way as with a plane region R. If we are given a function F ‘defi om the surface, it oan be evaluated for s0me point ofeach cubrogion of the surface and the sum ZFS, computed. Then, as the subdivisions increase in number and become finer, the limit that this may tend to is called ‘the double integral of F over S [fras. ‘The arguments of F have been left vague here. It may be that Fis given as a ‘unetion of position in space and we therefore evaluate it at a point on the surface, or it may be defined only on the surface itself. ‘The area of a curved surface is not an easy thing to define though more mystery than 18 necessary is often accorded if (see J. Serrin, Amer. Math. Mon. 68, May, 1961, p. 435 for an elegant discussion). For many common surfaces however, we may relate the area of an element ofthe curved surface 445 to the area ofits projection on a coordinate plane (say O12) by dys = ‘nS where nis the third component ofthe normal. This is shown in Fig. 34 ‘from which itis clear what must be done to calculate the areas of a cap. The ‘cap S with boundary curve I’ has to have a projection on the O12 plane’ ‘consisting of the simiply connected region with boundary C and for any point x, %4 of R there must correspond only one point of S. If the normal to the surface mis defined everywhere on it, and ny #0, then ast 4 Surface Integrals 9 and : irs Vf Figo cease we can divide the surface up into parts that are 3clementary, and ‘evaluate the integral as the sum of the integrals over the several parts. If ‘ny = 0 then the surface is either I-elementary or 2-lementary meaning that ‘we can project the area on to the 023 or the O31 planes in a similar way. Exercise 3.14.1. S isthe hemisphere f mxz0, where m is a fixed unit vector, Cakulate ff asm a for an astitary Constant vector 8 # 6 Cartesian Vectors and Tensors: Their Calculus (2.16 [Exereue 3.14.2. A surface can be given parametially by three functions #1 ett). Show that (#8 , 28) au, du (Bengt) see ‘Deduce that area is independent of the choice of Cartesian coordinates. ads 3.15. Volume integrals We shall also have occasion to integrate ove the volume inside or outside of closed surface. Ifa volume is such that a line parallel to © 3 meets its bounding surface in two and only two points (ay 34 =f), % = F.l%y%), J, 2f.) we may cal it Selementary.* A sphere i the most ‘obvious example of this. Similar dfinéons apply to the terms Ie and elementary. Ifa volume can be divided into a number of smaller volumes ‘ach of whichis 3elementary, the volume may be called 3composite. A ‘epion wnlch fs -, 2 and Scomposte wil be called & composte volume ‘An integration throughout the volume V is written [[Jrewanser ets) and is, as before, the limit of a sum of the products ofa very smal subdivision of the volume and the function F evaluated somewhere within it. The limit is taken by letting the number of subdivisions increase without limit and the size of the largest tends to zero. If the volume Vis 3-slementary and its projection on the O12 plane is a repion R, then [[frer= {Joes where @.152) ot. = fe Rona ‘This is shown in Fig. 3.5, The integral over a 3-composite volume may be calculated as the sum of such parts. ‘The double integral may be similaely reduced. We can cal a simple closed ‘curve C in the plane elementary i it can be traversed 80 that the slope of its tangent does not decrease. A composite curve is one that can be divided into elementary curves, as in Fig. 36. If is bounded by an elementary closed + This convenient nomenclature is taken from W. L. Ferra, Integral Calulus, Oxford, 1988, 50 Carcesian Vectors and Tansors! Thole Calculus 15 curve C whose projection on the O1 axis isthe interval (2,5), it can be specified by two functions g (3) and g-(x,) as shown in Fig, 36. Then [foaa= [mayas, where * G.153) Hs) = JG) Glo 9 a Exercie 3.15.1. Evaluate f(r) a when Wis the sphere r = nd 2 = x 3.16. Change of variable with multiple integrals In Cartesian coordinates the element of volume dis simply the olume oF a rectangular parallelepiped of sides di, dry, dry and so Bm di dey dy 6.161) Suppose, however, that itis convenient to describe the position by some other coordinates, sy €, fy dy. We may ask what volume is to beassocited with the three small changes dé, df, dy ‘The change of coordinates must be given by specifying the Cartesian point x that is to correspond to given set, fy by #1 (Eb @.162) Then by partie diferentiaion the small differences corresponding to change df, are Let ax! be the vectors with components (8/86) df, for J= 1, 2, and 3 Then the volume element is w= a axe? saa) ax, nt? ~ +) des GA a a5 a6, a = Jab, db dey (3.16.3) where Arn m8) One Das dy BE bb) 36, ey OF is called the Jacobian of the transformation of variables. (BM, The Vector Operator V—Gradient of « Sea 51 Exercise 3.16.1. Show that the volume element in the frame OI33 of coordinates £, = lyr is dF, dt di. Exercise 3.162. Obtain the volume elements in cylindrical and spherical polars by the Jacobian and check with a simple geometrical picture. 3.21, Vector fields ‘When the components of @ vector or tensor depend on the coordinates wospeak of vector oF tensor field. The low of fui isa perfect alization of # vector fel for at each point inthe region of flow we have a certain velocity vector Wx, %)- I the low is unsteady then the velocity depends ‘on the time as wellas postion, = vs, yy). When itis necessary to be Specie we shall fer to this a ame dependent vector Bel. Iris somtimes Convenient to abbreviate these tow) and v(x, 2, oF to we the index notation and writ of, 1), Auf Dye. ‘Associated with any ector field a(x) are its trajectories, which is the name sven to the family of curves everywhere tangent to the local vector a. They fre the solutions ofthe simultaneous equations & sic, Ge a OF Oe Ge wheres is parameter along the trajectory. (It willbe the arc length i a i always a unit vetor.) Though we have yet to define chem, the streamlines ofa sneady flow are probably suficieniy familir to be mentioned as the realica- tion ofthese trajectories. For a time dependent vector field the trajectories will ls be time dependent since they ae solutions of fm ean sn ds oa IF Cis any closed curves the vector field and we take the trajectories through all paints of C, thay describe « wucface known a8 4 sector tube of the eld en eau 3.22. The vector operator V—gradient of a scalar ‘The symbol V (enunciated as “del” or “nabla”) is used for the symbolic vector operator whose i component is @/@xy. Thus iV operates ona scalar fanction of positon it produces a vector Vp with components dp/2x,. We shoold of course stabi that Up ie indeed vector” Tn the coordinate fame OT35 the vctor Twill have components dp/@f,, However, ap _ a9 ax, ag eae 6221) since x, = Jy, 80 that Vo i. vestor 32 Cartesian Vectors and Tensors: Theie Calculus $9.22 I Cartesian coordinates the operation of pat ifferentation with expect tothre coordinates gives the components of a tnsor ofthe nex higher order. To show this, suppose thats. ) sa second onder tensor field, then the BI quantes 6:4, /8x, ds, are the components ofa fourth order tensor, Fos like 3p a, 5,25, I ve wte gay f0 the second derivative BAg[2yB, then in the frame os @222) labia dais which shows that the Ay» are components of a fourth order tensor. “The sufix notation. forthe paral derivative with respect tox i8 a very convenient one and will be taken over for the generalization of ths operation that must be made for non-Cartesian frames of reference. The notation “grad” for V is often used and refered to asthe gradient operator. Thus red p= Vp isthe vector which i the gredioat of the scalar, and grad A ‘would bea tensor Ay, "Visalo sometimes written @/@x and can be expanded nieim Vm ey axe 6223) 1 ly ip %) = 9H) 38 scalar function of postion and ay =m, dr small displacement inthe direction m then tim Ct Bar) — ox) hr is the derivative in the direction m and is sometimes denoted by p/2n. By Taylor's theorem, Hx +d) = 92) + (ads Vp +00, 2 vpn vy. e224) Exercise 3.22.1. 16 Fp) = that VF) =f() Vo. IL-l be convenient to refer to the following formulae (2)-(4) later. They are all elementary but the exercise of establishing them is valuable, Exercise 3.22.2. Vey) = ¥ Ve +9 Vy. Exercise 3.22.3. Wp) = (Ay + AW%; ey if Ais constant. Gummation oni), * flo) da isthe indefinite integral of f(), show 2. The Divergence of a Vector Field 33 Exercise 3.224. Vf) = (f' OID where # = x+x. Exercise 3.22.5, Show that if ¢(x) = ¢ is a surface, Vp is normal to the surface. Exercise 3.22.6. 1f Ayxer, = 1 is a central quedric defined by a symmetric tensor Ay, then Ay transforms the radius vector x into @ vector normal to the surface, ‘The angle between the redius and the normal is given by tant 0 = (RXR AnAantn) — Cede? (xox) Atex) — (ee Ax 3,23, The divergence of a vector field ‘The symbolic sealar or dot product of vector and the operator Vis called the divergence of the vector field. Thus for any differentiable a(x, x5), we write = 0 5 204 3 “Fe + ae + Ba G23.) “The vergence is a lar since itis the contraction (or race of the cond eer tense Scppose tha an elementary paralelpied is st up with one comer Pat xy Xp x4a0d the diagonally opposite one Q at x, + dry, % + diy x9 + dy tS shown in Fig 37. Theoutward nit wormal to the face through Q which Popa 001i egy meres te Dutvard norma tothe paral face through P is egy On the fist of thee no we ale + dis dy 8) where diva=Vea nS Aasutdy and BSE Smt Oty ‘Whereas on the second it is a(x, 6). Fig. 37 ‘Thus ifm denotes the outward normal ‘and dS is the area dry dx ofthese faces, we have a contribution from them to the surface integral ff am dS of Taslng + das Fa fd) — a4 be BD] de dy = lat) + oe de, — aC y xD] xy dy + OC) Sh asd dey + 0, 54 Cartesian Vectors and Tensors: Thele Calculus $9.23 ‘ere (dt) denotes terms proportional to fourth powers ofthe dx. Similiar terms with day, y/9x, wil be given by the contributions of the other ‘ces s0 that forthe whole parallelepiped whose volume dV — diy did we have 1 ay 4 ay , Bay 4 = 21 5 es, 209 1g, 22) Fplle-nas fot Set +0. 0.3.2) If we le the volume shrink to z2r0 we have lim 2 fa-nds =v 6233 stouseieatin star af 6s aeW n oc at cok A vector fei whose divergence vanishes identically icalled solenoidal. Ifthe flux field ofa certain property is solenoidal there is ao generation of that property within the field, for al that ows into en infinitesimal element flows out again, TE ais the gradient of szalar function Vo, its divergence i called the Laplacian of @ Vip dv ga 9 gu 8 BE 6234) ‘A function that satisfied Laplace's equation Vip — 0 is called a potential Per as ease, th oation div Ais omens ued fre veto Ay ‘then div’ would be 4.3 We shall generally prefer the index notation for Exercise 323.1. V-(ya) = Vea + AV a) Exercise 23.2, V-(UNW) = (VA8)-d—a-(VAD). Exercive 3233. YQ =fO+Y Oh, Pax Exercise 3.23.4. (2) denotes the symmetsic tensor of order n+ 1 ep + Mey ky EAR hy te Fel Whereas axis the tensor axa -. x, of order m+ 2. Show that div (ax) = {ula} if di Exercise 3.23.5, Interpret. Vip physically by thinking oft as V-(Vp). Exercise 3.23.6. I 9(8) = 9%, %) isa potential function and r# = x+x, ‘show that oe) 42 £6(%) -2 = is ao potential funesion and tha its normal derivative o the sphere r= avanishes, (Weiss, P. Proc. Camb. Phil. Soc., 0, (1944), 249.) PAK, The Corl of s Vector Field 55 3.24. The curl of a vector field ‘The symbolic vector or cross product of V and a vector field ais called the curl of the vector field. Tes the vector Vaa=curla with three components (8a _ 209) (2s _ a) (@@ = & x) Ge ax)’ Na ay C28D It’s connected, as we shall ee, with the rotation ofthe field and is sometimes, writen rot a in older texts * In dyadic, notation cur a = (grad a). Consider an elementary rectangle fn the plane normal to 01 with one comer P at (i, Xp 4) and the diagon- ally opposite one Oat (xs x2-+ dip ++ du), as shown in Fig. 38. We ‘wish to calculate the in integral around this elementary cireit ofa ts, where ‘is the tangent. Now the line through P parulel 1003 Tuas tangent ea) and the prallel side through Q has Fens tangent ¢qy, and each is of length dx. Accordingly. they contribute to ast ds an amount LCi + di 89 ain te BD] dey = Let f+ BY dg — on te i + OP) = s xs dry +O) Similary, frm the other two sides, there isa contribution, I > xy dy + Od), ‘Thus writing 44 = didi, we have Tfetan 2) Fhatam 24) +00 (3243) pte Ros and in the limit + rot as wed by Toupin (Handbuch der Physik /, Sect. 268) to denote the ant- symmetric tensor of which cul a the vector; this usage lends itself to generalization, 56 Cartesian Vectors and Tensors: Their Calculus 824 "The sufix 1 has been put on the integral siga to show thatthe line integral isin « plane normal to O1, and Eq. (324.4) shows that the limit is the first ‘component of the curl” An entirely similar treatment would give the other ‘wo components for line integrals ‘around rectangles in planes perpendic- ular to the 02 end 03 exes, ‘Wecan treat an ininitesimaltriangle (OR of Fig. 39) ina similar way. IF thelength of POis dry = ds cos 0, and the length of PR is dr = dssin 6, the 2 areas 4A = jedi 005 0 sin 6. Figa9 ‘The unit tangents around the triangle Aare gi, (0, 05 0, sin 6), —e. Thus approximating each part ofthe line integra by the length of side multiplied by et evaluated af its midpoint, $actds = ais + Msn ds) dy + (00s Oakey y+ Btn ae + eM) $ sin O aglxp xe + bdey xy + bdxq)} ds ~ op + ha) dy = odin ta + bt ta + by) ad ty m0 + dds) ds sin = feels Xe + ddr Xy + bdr) — antes, oy + bey, %)} dsco5 0 _ (®_ 3 -@ 32) 4a cos sin + 0, Again, as ds, and so dA, tends to zero, 1 tim dA, fin fe fortdemat a, and similar forms hold for triangles in planes normal to 02 and 03. ‘Consider now a fourth point S at (xy + dxy 2p 4) 50 that QRS isa plane triangle of area dA whose unit normal isn. ‘Thus the areas of the eiangles POR, PRS, PSQ are A= nds, ddym md, dpm ngld respectively. However, the line integral around QRS can be taken to be the sum of those around POR, PRS, PSQ since the parts PQ, PR, PS are BL The Curl ofa vestor Fld 7 traversed once in ach destion and o cane Ge Fg 3.0). Ths i ais very small, Gongs tds artis + Garters fareds ‘Thea in the limit 8 6.245) By working a litle harder (Exercise 3.24.9), it can be shown that if any small curve in the plane with normal n shrinks on the point x, the limit of the {at ds divided by the area isthe projection of curl a on the normal, a. a e s s Fig. 210 ‘A vector field a for which Vaa ula =0 6246) is called irottona, for evidently the circulation around any infinitesimal curve vanishes. Exercise 324.1. V-(VA2) =0. Exercise 3.242. 0 A(V9) Exercise 3243. V Aga) = (U9) Aa-+ 9V Aa. Exercise 3.244. V ACAD) = Vb) — WT 8) + (b+ Va — (a Dh. Exercive 3245. VA(W Aa) = Wa) — Va, Brercise 3.246. V(a-¥) = (a Wh + (Yat an(Vab) + bACVAw Exercise 3247. a (WA) = HV(0 8) ~ (Ve Exercise 3.248. I (x) i the velocity ofa rigid body due to a rotation w, show that VAY = 20. se Cartesian Vectors and Tentors: Their Calculus §224 Exercise 3.249. If Cis a iafsitesimal curve about xg in a plane with normal 'n, use Taylor's theorem to approximate the line integral. feds = $atn) + (Gc — x) Vjaag +... Ttds Hence, using Ex. 3.13.2, show that this integral is VA ax) +m dS. Exercise 3.26.10. If Cis the projection on plane normal to Of of a simple ‘closed curve C and is self simple, show thatthe areas Cy, Cy Cy are components of the veetor af, (wanes. 3.31. Green's theorem and some of its variants One of the most valuable transformations of tensor analysis is that of Green* which relates a certtin volume integral to an integral over the bounding surface. We shall give two demonstrations of it.The surface sntegral Occurring iste integral of a -m, Where mis the OUWArG mortal To the surface. If we think of # a5 the flax of some physical property the integral of +m over the whole surface is thus the total flux out of a closed volume. ‘Green's theorem says tha this total x equals the integral of Va through ‘out the enclosed volume Suppose ¥ is a volume with a closed surface S and a any vector field defined in V and on S.. Then if Sis piecewise smooth with outward normal and a continuously aiferentiable, JfJv-aav= ffarnas. @31) ¥ 8 We have shown for infinitesimal volumes that Vendy acnds and the integral on the left-hand side is by definition the limit ofthe sum of a ‘numberof ffinitesimal volumes into which V can be divided. However ifit is s0 divided, the contributions of a+n dS from the touching faces of two adjacent elements of volume are equal in magnitude but opposite in sign since the outward normals point n opposite directions. Thus ina summation ‘of am dS, the only tems that survive are those onthe outer surface S, and Eq. G31.) follows. “Another form of Green's theorem isthe following. Let ¥ be a composite Ti 0 pao to enter the dia sbjct ofthe proper atbution of names to theorems Ths torem was give in various forms by Lagrange (1762) Gass (813) ae ‘Onrogradsky (HDD. Tes bt known source, however, 8 Green's “Ema onthe application ‘of mathematical analysis othe theories of electricity and magaetsm” (828). "The noo ‘denominational nae "Divergence theorem” i ako tse. BAI. Green's Theorem and Some of les Varianes 59 volume with pieee-wise smooth boundary S and F a continuous function whose derivatives Fare continuous in V. Then [[Jrcav= [Jenas. esta) Let us prove this for a volume which is elementary and with i= 3; then by joining elementary volumes ic will hold for a 3-composite volume and so by extension to a composite volume for i= 1,2, or 3. Figure 3.5 illustrates @ 3-elementary volume and by Eq. (3.15.2) we can write 2 ye fl [fF wr= Jods an where ea) Gm fe hg = Flt false )] — Flt tof y 3) 1-8 and Ris the projection of the volume ¥ on the O12 plane. However, if we use a suffix + or — to denote the upper and lower surfaces, respectively, the ‘element of area in R dry dia = mu dS. or —Ma. dS_ G31) since the elements of area are all positive but m,_ is negative. Thus, subst- tating in Eq. 3.31.3), [54 [Flow selon sD dS +f [Fos xn Sly xno dS. However, the right-hand side is now just the definition of the surface integral Fad, so the theorem is established fora 3-elementary region. A3-composite volume can be divided up into a finite number of 3-elementary volumes in cach of which Eq. (3.31.2) will hold. However, on adjacent faces of the clementary volumes the outward normals point in opposite directions so that ‘the contributions tothe total surface integral cancel. ‘Thus in summing over All the elementary volumes the only contributions to the right-hand side of Eq. (3.31.2) that survive are those from the outer surface of the composite volume. ‘The theorem can be extended 10 «function F which is continuous in ¥ but whose derivative F, is only piece-wise continuous. The volume ¥ may increase without limit or the integrand have singularities provided that the integrals converge. For further lightening of restrictions the reeder is referred to the book by Kellogg mentioned in the Bibliography atthe end of this chapter. o Carcesian Vectors and Tentors: Their Calculus 331 F may be a scalar or the component of a vector or tensor. If F = a, we hhave the form of Green's theorem originally given, If F then Fy = suds it the KY component of V Aa, whereas Fry = eyina, is the E™ ‘component of a 4.8. Thus [fJovnsav= Jian sas. G31.) ¥ 8 Tra = Vp we have [[fvrear= [fve-nas— [fBas, e116 here 2p/dn denotes the derivative in the ditetion ofthe outward normal. Ifa = yp, then [[fore+ vy var [fvieas @3t7 hence [fore ovrmar= [f(v32— 0%) as. aaa Exercise 331.1. Use Green's theorem to obiain the answer to Ex. 3.14.1 ‘without integration. xereae 231.2, Show that Sifter nd ad+ Adve +a(V-b) + 0-2) av = {ft a Dyadic notation is used here, so n-(ab+ba) is the vector lady + ba). Exercise 3.31.3. By taking a to be independent of x, and ay = 0, show that if A isan area in the O12 plane bounded by a curve C, then J fea, ‘where tis the unit tangent vector to C. Exercise 3.31.4, Deduce from the preceding question that $, (eat, + ta) ds — ff (aay — a9) dA. (ab + ba) — (a +¥)a] 45, + ayy) dA = faite — ay) ds PR. Stokes! Theorem 6 Exercise 331.5. V is compoted of two volumes ¥, and Vz divided by a surface S'. The veetor ais continuousin V = % + Veand its derivatives are continuous in Vy and Vy separately and the normal derivative is continuous across S'. Show that Green's theorem holds good for any volume ¥* within V. 3.32, Stokes’ theorem ‘The previous theorem concerned the relation of an integral over a closed volume and its relation to an integral over the bounding surface. Stokes" {heorem* relates the surface integral ovr a cap to line integral around the tounding curve. The line integral appearing is that of af, tat i the total circulation, and the theorem says that this equal to the surface integral ofthe normal component of cul a. We shall again give two demonstrations We have shown earlier in Section 3.24 and Es. 3.24 that for aninfitesimal teiangolar area the line integral f, atds=(Was)-nds. 6321) ISisthe cap of closed curve C, wecan divide its surface into a large number cof small triangles foreach of which Eq. (3.321) true. ‘Then in suraming the right-hand sides we shall have the srface integral over the whole cap, Whereas onthe left-hand sies contributions from adjacent sides of triangles will canel since they will be traversed in opposite directions. ‘The only remaining contribution from the line integrals wl hes be thove fom the bounding curve Cand fyerterm [[ovanenes 0322) The convention that the normal to the curve should be right-handed with respect to the direction of traversing the curve C is observed throughout. “More precisely stated, Stokes’ theorem says that for any two-ided piece- ‘wise smooth surface spanning the closed curve Cand any continuous vector field a whose partial derivatives are continuous, Eq. 3.32.2) holds. We shall give another proof depending on Green's theorem. Suppose first that the cap, Scan be specified by a single function xy = ftxy, x) in the region R within the closed curve C’, the projection of C on the’ plane O12. Consider the terms in a, namely, foowdem {f (mga —mge) + Te appear thatthe atbation of tit theorem to Stoke eles appropriate than that of the previous one to Gren. Its actaly due to Kelvin though fe wage of Stoke? ‘ame it too entrenched to be changed. See russel, "Kinematis of Vortisty,”" foots, pie a Carcesian Vectors and Tentors: Their Calculus $9.32 Now on S, x5 = fle %4) and 0, = als Xe ft 29) = BC 10) 2 _ da 5 265 F Oxy Ay OKs Oty However, nny = —2ffbx because the direction cosines of the normal are proportional to Aff8x, ffBrq and —1, respectively. Hence, i ($3 ml) as = Jf — FE nas a = Se = aa by Eq. G.15.1). However, by putting a, = 0, a, = gin Ex. 3.31.3 which isa ‘Wo-dimensional form of Green's theorem, ofp Baden faiide = fobs sy fon I a, ‘where 1; and ds’ are the component of the tangent and element of are C” spectively. However, 1; ds’ pral around C andthe theo through for ay and dp and for surfaces that can be decomposed into parts ‘hich are either 1, 2, oF 3-elementary. A further discussion willbe found in Kelloge's book. Other forms of Stokes’ theorem may be derived by inserting various ‘ectors; in particular the components a, may bea set of components from a tensor, s4y Aya with j& fixed, Thus it is convenient to write Sfeantrsas = $ Paty ds 6323) 43 eo Exercive 232.). Relate the results of Exercises 3.13.1 and 3.14.1 by Stokes’ theorem. Exercise 3.32.2, Show that the vanishing of the integral of (V 4 a) -m over a closed surface is a consequence both of Green's and Stokes" theorems. Exercise 3323. Show that J er ~ a4) dS = f eyo ds GAL, The Classification and Representation of Vector Fiel 3 vere 3324, Show that $, Wray-tds~ [f[Zeo-a) acer] as vere 4425, Te Bux of & though the cap Sis [J amd. Suppose (1% xy and S moves in space its velocity being given by a vector field v. ‘Consider the volume bounded by S, the position of $ at time 1 and 5; its position at time 1 + dt and a bounding surface of vectors ‘dt forall points ofthe boundary C. Then applying Green's theorem to the volume, show that AU tin [fast a-nas —ffos.-nat] = [fen] nase, “J [2+ 07-ay + vatenn) nas (ands Exercise 3.3256, 16 ¥ an x ure ws int Ex. 3.31.5 save tna this time itis he ‘tangential derivative that is continuous across S', show that Stokes" theorem still holds. 3.41. The classification and representation of vector fields We have already noted two distinct types of vector field; namely, the solenoidal, for which V+ = 0, and the lrotationa, for which V aa = 0, Such fields occur physically, and itis of interest to explore theit properties a further. We are particularly interested in relating the three components of a to certain scalar functions of position. For example, if is the gradient of a scalar function 9, then it is certainly irrotational for VA Vp = 0 identically. Thus if «problem calls for an irotational vector fel, it may be possible to turn it into a problem that requires finding only the function Certainly an easier matter than finding ll three components of a. Before this ‘can be done however it must be proved that all irrotational vector fields can be represented as the gradient of a scalar function, which is slightly more

    Potrebbero piacerti anche