Page 147 - Vector Analysis
P. 147

§4.7 The Stokes Theorem                                                                                                                        143

as a (constant multiple of) measurement of the speed of rotation. The limit of the quantity
above, as r Ñ 0, is then a good measurement of the rotation speed of u at the point a about
the axis in the direction N.

    We start from the case that N = e3 so that P be parallel to the x1x2-plane. With
u1, u2, u3 denoting respectively the first, the second and the third components of u, by the
change of variable ds = rdθ and the L’Hôspital rule (to obtain the second “=”) we find that

lim 1 ¿ u ¨ T ds
rÑ0 2πr Cr r

     =  lim   1      ż 2π           [(       +   (r  cos  θ,  r  sin  θ,    )  cos  θ  ´    (   +  (r  cos   θ,  r  sin  θ,    )  sin   ]  dθ
             2πr                     u2 a                                 0)              u1 a                               0)        θ
        rÑ0            0

     =  1   dˇ                 ż 2π     [(       +                       )             ´       ( )]
        2π     ˇ                         u2 a       (r cos θ, r sin θ, 0) cos θ              u1 a + (r cos θ, r sin θ, 0) sin θ dθ
                                 0
            dr ˇr=0

     =  1    ż 2π    [ B u2 (a) cos2             θ  +  B u2 (a) cos θ sin θ       ´    B u1 (a) cos θ  sin θ     ´   B u1 (a) sin2      ]
        2π                                                                                                                             θ dθ
               0      B x1                             B x2                            B x1                          B x2

        1 [ B u2 (a)                B   u1      ]      1   2          B uj (a) .                                                           (4.21)
        2 B x1                              (a)
                                                          ÿ           B xi

     =                         ´    B x2            =  2 ε3ij

                                                          i,j=1

Now suppose the general case that N ‰ e3. Let pe3 = N and choose pe1 and pe2 so that
␣pe1, pe2, pe3( is an orthonormal basis following the right-hand rule (that is, pe1 ˆ pe2 = pe3).
Then the vector field u has two representations

                                        u = u1e1 + u2e2 + u3e3 = v1pe1 + v2pe2 + v3pe3 .                                                   (4.22)

Let  O  =   [pe1...  pe2  ...  pe3  ]   and  introduce           a  new   Cartesian       coordinate       system        y   =    OTx.      Note
                                     ,

that y is the coordinate with coordinate axis parallel to the basis ␣pe1, pe2, pe3(. In this new

Cartesian coordinate system, (4.21) implies that

                                            lim   1¿        u¨T       ds  =    1 [ B v2 (b)  ´  B  v1     ]  ,
                                                 2πr Cr       r                2 B y1                  (b)
                                            rÑ0                                                 B y2

where b = OTa.

     Now we transform the result above back to the original coordinate system (so that the

limit is in terms of derivatives of uj w.r.t. xi). Note that (4.22) implies that v = OTu so
that vj = pej ¨ u. Moreover, with ejk denoting the k-th component (w.r.t. the ordered basis
te1, e2, e3u) of pej; that is, pej = ej1e1 + ej2e2 + ej3e3, the chain rule provides that

      B     =  e11    B             +   e12   B     +  e13   B            and           B    =  e21   B    +    e22   B    +  e23   B    ;
     B y1            B x1                    B x2           B x3                       B y2          B x1            B x2          B x3
   142   143   144   145   146   147   148   149   150   151   152