Page 144 - Vector Analysis
P. 144

140 CHAPTER 4. Vector Calculus

Theorem 4.83. Let Σ Ď R3 be an oriented C 1-surface with C 1-boundary B Σ, N : Σ Ñ S2
be a continuous unit normal vector field on Σ, and T : B Σ Ñ S2 be tangent vector on B Σ
such that T is compatible with N (which means T ˆ N points away from Σ). Then

     żż                                                @ u P TΣ X C 1(Σ; R3) X C (Σ; R3) ,

          u ¨ (T ˆ N) ds = divΣu dS

       BΣ Σ

where divΣ is the surface divergence operator.

Proof. Let tVm, ψmuKm=1 denote a collection of local parametrization of Σ such that ψm(Vm)X
B Σ = H for 1 ď m ď J, and ψm(Vm) X B Σ is non-empty and connected for J + 1 ď m ď K.
W.L.O.G., we can assume that Vm = Bm ” B(0, rm) for some rm ą 0. Write Um = ψm(Vm),
and let tgmuKm=1 be the associated metric tensor, as well as the associated first fundamental
form gm = det(gm). Let tζmuKm=1 be a partition-of-unity of Σ subordinate to t UmuKm=1.
Then

     ż Kż

     divΣu dS = ÿ                    ζmdivΣu dS

     Σ m=1 UmXΣ

                   =    J    2ż            ˝  ψm   )   B    [?gm  gmαβ  (     ˝  ψm)  ¨  B  ψm  )]
                            ÿ                         B yα               (u                       dy
                      ÿ                                                                  B yβ
                                      (ζm
                      m=1
                           α,β=1 Bm

                      +      K      2   ż          ˝  ψm)    B    [?gmgmαβ    (    ˝  ψm)   ¨   B  ψm    )]    .
                                             (ζm            B yα               (u                          dy
                           ÿ      ÿ                                                             B yβ
                                          Bm+
                         m=J +1  α,β=1

Let n denote the outward-pointing unit normal on either BBm for 1 ď m ď J or BBm+ for
J + 1 ď m ď K. Since ζm ˝ ϑm = 0 on BB(0, rm) for 1 ď m ď J, and ζm ˝ ϑm = 0 on
ty2 ą 0u X BB(0, rm) for J + 1 ď m ď K, the divergence theorem (on R2) implies that

ż    divΣu  dS  =      K 2ż                   [?gmgmαβ      (     ˝  ψm)   ¨  B ψm  )]    B    (ζm    ˝  ψm) dy
                                                             (u               B yβ       B yα
  Σ                   ÿÿ

                   ´

                       m=1 α,β=1 ψm´1(UmXΣ)

                   +      K      2   ż                 ˝    ψm  )nα[?gmgmαβ       (      ˝  ψm)    ¨  B ψm )]  dy1
                                                                                   (u
                        ÿ      ÿ                  (ζm                                                 B yβ

                      m=J +1  α,β=1    BmXty2=0u

                       Kż               (u ¨ ∇Σζm) ˝ ψm?gm dy

                      ÿ
                =´

                      m=1 ψm´1(UmXΣ)

                          K   ż                                         [     2   nα?gmgmαβ           B ψm  ]
                                                                                                      B yβ
                   +    ÿ                  (ζm  ˝  ψm)(u    ˝   ψm)  ¨      ÿ                                  dy1  .

                      m=J +1    BmXty2=0u                                  α,β=1
   139   140   141   142   143   144   145   146   147   148   149