Page 116 - Vector Analysis
P. 116

112 CHAPTER 4. Vector Calculus

A third kind of coordinate chart is given as follows. Let U = (0, 2π) ˆ (0, π), and define

                                  ψ(θ, ϕ) = (sin ϕ cos θ, sin ϕ sin θ, cos ϕ).

Then ψ : U Ñ S2zt(x, 0, z) | 0 ď x ď 1, x2 + z2 = 1u is a continuous bijection with a contin-
uous inverse. We note that for any U = (θ0, θ0 + 2π) ˆ (ϕ0, ϕ0 + π), ψ is a homeomorphism
between U and an open subset of S2.

Next, we would like to define the derivative of f when f : Σ Ñ R3 is a vector-valued

function. We first talk about what the directional derivative is. Let Σ Ď R3 be a regular

surface, p P Σ, and v P TpΣ. It is intuitive to define the directional derivative of f at p in

the direction v by

                                                    dˇ          ˝  x)(t)  ,                                                           (4.6)
                                                      ˇ (f

                                                    dt ˇt=0

if the derivative exists, where x : (´δ, δ) Ñ Σ is a C 1-parametrization of a curve on Σ such

that x(0) = p and x 1(0) = v. The first question arising naturally is that if the derivative

in (4.6) depends on the choices of x. Suppose that y : (´δ, δ) Ñ Σ is a C 1-parametrization

of another curve on Σ such that y(0) = p and y 1(0) = v (note that the curve x((´δ, δ))

and y((´δ, δ)) in general are different). Let tV, ψu be a parametrization of Σ at p, and

q = ψ´1(p). Then the chain rule (Theorem 2.49) implies that

         v       =  x 1(0)  =  dˇ             ˝  ψ´1  ˝  x)(t)     =            (dˇ         (ψ´1  ˝                           )
                                 ˇ (ψ                                 (Dψ)(q) ˇ                                         x)(t)
                                                                                   dt ˇt=0
                               dt ˇt=0

and  similarly,  v  =             (    d  ˇ   (ψ´1          )      Therefore,
                        (Dψ)(q)           ˇ         ˝ y)(t) .

                                     dt ˇt=0

                           (   d  ˇ       (ψ´1   ˝        )  =            (  d  ˇ                    )
                 (Dψ)(q)          ˇ                 x)(t)       (Dψ)(q)         ˇ     (ψ´1 ˝ y)(t) .

                               dt ˇt=0                                       dt ˇt=0

The injectivity of (Dψ)(ψ´1(p)) then shows that

                                  dˇ         (ψ´1   ˝ x)(t)  =  dˇ       (ψ´1  ˝ y)(t) .
                                    ˇ                             ˇ

                               dt ˇt=0                          dt ˇt=0

Using the chain rule again,

dˇ       ˝       x)(t)  =  Dddt(ˇˇˇft=˝0(ψf)˝(ψψ´˝1(ψp)´)1(˝ddxt ˇˇˇ)t(=t0)(ψ=´D1 ˝(fy˝)(ψt)))(ψ=´1dd(tpˇˇˇ)t)=(0(ddft  ˇ          ˝        )
  ˇ (f                  =                                                                                               ˇ (ψ´1        x)(t)
                                                                                                                        ˇt=0
dt ˇt=0
                                                                                                                        ˝ y)(t) .

In other words, the derivative in (4.6) is independent of the choice of x as long as x(0) = p

and x 1(0) = v. This observation implies the following
   111   112   113   114   115   116   117   118   119   120   121