Page 48 - Vector Analysis
P. 48

44 CHAPTER 2. Differentiation of Functions of Several Variables

Example 2.34. If f : R Ñ R is differentiable at x0, must f 1 be continuous at x0? In other

words, is it always true that lim f 1(x) = f 1(x0)?
                                                        xÑx0

Answer: No! For example, take

                                             $     x2 sin  1  if x ‰ 0,
                                             &

                                  f (x) =                  x

                                             % 0 if x = 0.

     1˝ Show f (x) is differentiable at x = 0:

                 f 1(0) =    lim  f (0 + h) ´ f (0)      =    lim  h2  sin  1  =  lim h sin   1  = 0.
                                                                            h

                             hÑ0 h                            hÑ0 h               hÑ0 h

     2˝ We compute the derivative of f and find that

                                             $  2x sin   1    ´ cos  1     if x ‰ 0,
                                             &           x           x

                                  f 1(x)  =

                                             %0                            if x = 0.

     However, lim f 1(x) does not exist.
                        xÑ0

Definition 2.35. Let U Ď Rn be open, and f : U Ñ R be a function. If the partial

derivative B f exists in U and has partial derivatives (at every point in U) with respect to
              B xj                                        B ( Bf )
xi,  then                    partial  derivatives                      is  denoted       by    B2f .
           the second-order                              B xi B xj
                                                                                              B xiB xj

     In general, if the k-th order partial derivatives B kf exists in U and has

                                                                          B xik B xik´1 ¨ ¨ ¨ B xi1

partial  derivatives  (at  every  point in   U)    with    respect     to   xik+1 ,  then the    (k     + 1)-th order
partial  derivatives         (                       )                                                   ; that is,
                        B              Bkf                                             B k+1f
                                                         is denoted by
                      B xik+1 B xik B xik´1 ¨ ¨ ¨ B xi1                        B xik+1 B xik ¨ ¨ ¨ B xi1

                             B k+1f             = B(                    Bkf              )
                                                                                           .
                      B xik+1 B xik ¨ ¨ ¨ B xi1 B xik+1 B xik B xik´1 ¨ ¨ ¨ B xi1

Theorem 2.36. Let U Ď Rn be open, a P U, and f : U Ñ R be a real-valued function.

Suppose    that  for some 1 ď i, j ď n,      Bf ,  Bf ,      B 2f      and       B 2f    exist in a       neighborhood
                                                           B xjB xi            B xiB xj
                                             B xi  B xj

of a and are continuous at a. Then

                                          B 2f (a) = B 2f (a) .
                                      B xiB xj           B xjB xi
   43   44   45   46   47   48   49   50   51   52   53