Page 113 - Vector Analysis
P. 113

§4.2 Conservative Vector Fields                                                                                                  109

Choose δ ą 0 such that B(x, δ) Ď D. Let C be a piecewise smooth curve joining
P0, and L be the line segment joining x and x + hej, where 0 ă h ă δ and ej =
(0, ¨ ¨ ¨ , 0, 1, 0, ¨ ¨ ¨ , 0) is the unit vector whose j-th component is 1. Then with the
parametrization of L: γ(t) = x + tej for t P [0, h], we have

                       ϕ(x +  hej )  ´ ϕ(x)  =  1  ż     ¨  dr    =             1  żh        +   tej )  ¨  ej  dt ;
                                h               h     F                         h      F(x

                                                     L                               0

       thus passing to the limit as h Ñ 0, we find that

                                             Bϕ    (x)  =   F(x)             ¨  ej  .
                                             B xj

As a consequence, F(x) = (∇ϕ)(x) which implies that F is conservative.                                                           ˝

    Let D Ď R2, and F = (M, N ) : D Ñ R2. If F is conservative, then M = ϕx and N = ϕy
for some scalar function ϕ : D Ñ R; thus if ϕ is of class C 2, we must have My = Nx. In
other words, if F : D Ñ R2 is a smooth vector field, then it is necessary that My = Nx. The
converse statement is not true in general, and we have the following counter-example.

Example 4.31. Let D        Ď  (Rx,2yb) e=t(hxe2a+ynnyu2 l,axr2´r+exgyi2o)n.  D     =   ␣(x,  y)  ˇ  1  ă   x2  +     y2  ă  4(,  and
consider the vector field  F                                                                     ˇ

                                                                             Then

                              B y = x2 ´ y2 = B ´x ;
                              B y x2 + y2 (x2 + y2)2 B x x2 + y2

however, if F = ∇ϕ for some differentiable scalar function ϕ : D Ñ R, we must have

                                                           y
                                         ϕx(x, y) = x2 + y2

which further implies that

                                     ϕ(x, y) = arctan x + f (y) .
                                                          y

Using  that  ϕy(x, y)  =     y,      we  conclude  that    f  is  a             constant     function;     thus

                          x2 + y2

                                         ϕ(x, y) = arctan x + C .
                                                              y

Since ϕ is not differentiable on the positive x-axis, F ‰ ∇ϕ.
   108   109   110   111   112   113   114   115   116   117   118