Page 195 - Schaum's Outline of Theory and Problems of Advanced Calculus
P. 195

186                   APPLICATIONS OF PARTIAL DERIVATIVES                  [CHAP. 8



                        The result can be extended to determine the envelope of a one-parameter family of surfaces
                      ðx; y; z; Þ. This envelope can be found from
                                                ðx; y; z; Þ¼ 0;      ðx; y; z; Þ¼ 0                 ð12Þ
                     Extensions to two- (or more) parameter families can be made.



                     DIRECTIONAL DERIVATIVES
                        Suppose Fðx; y; zÞ is defined at a point ðx; y; zÞ on a given space curve C.  Let
                     Fðx þ  x; y þ  y; z þ  zÞ be the value of the function at a neighboring point on C and let  s denote
                     the length of arc of the curve between those points. Then
                                             F        Fðx þ  x; y þ  y; z þ  zÞ  Fðx; y; zÞ
                                         lim    ¼ lim
                                         s!0  s   s!0                s                              ð13Þ
                     if it exists, is called the directional derivative of F at the point ðx; y; zÞ along the curve C and is given by
                                                   dF  @F dx  @F dy  @F dz
                                                   ds  ¼  @x ds  þ  @y ds  þ  @z ds                 ð14Þ
                     In vector form this can be written

                                   dF    @F   @F    @F     dx   dy   dz         dr
                                                                                  ¼rF   T           ð15Þ
                                                             i þ
                                                      k
                                      ¼
                                    ds   @x    @y   @z     ds   ds   ds         ds
                                                 j þ
                                                                       k ¼rF
                                                                  j þ
                                            i þ
                     from which it follows that the directional derivative is given by the component of rF in the direction of
                     the tangent to C.
                        In the previous chapter we observed the following fact:
                        The maximum value of the directional derivative is given by jrFj.
                        These maxima occur in directions normal to the surfaces Fðx; y; zÞ¼ c (where c is any constant)
                     which are sometimes called equipotential surfaces or level surfaces.
                     DIFFERENTIATION UNDER THE INTEGRAL SIGN
                                                      ð
                                                      u 2
                          Let                    ð Þ¼   f ðx; Þ dx  a @   @ b                       ð16Þ
                                                      u 1
                     where u 1 and u 2 may depend on the parameter  . Then
                                             d    ð  u 2  @ f    du 2       du 1
                                             d   ¼  @   dx þ f ðu 2 ; Þ  d     f ðu 1 ; Þ  d        ð17Þ
                                                   u 1
                     for a @   @ b,if f ðx; Þ and @ f =@  are continuous in both x and   in some region of the x  plane
                     including u 1 @ x @ u 2 , a @   @ b and if u 1 and u 2 are continuous and have continuous derivatives for
                     a @   @ b.
                        In case u 1 and u 2 are constants, the last two terms of (17) are zero.
                        The result (17), called Leibnitz’s rule,is often useful in evaluating definite integrals (see Problems
                     8.15, 8.29).



                     INTEGRATION UNDER THE INTEGRAL SIGN
                        If  ð Þ is defined by (16) and f ðx; Þ is continuous in x and   in a region including
                     u 1 @ x @ u 2 ; a @ x @ b, then if u 1 and u 2 are constants,
   190   191   192   193   194   195   196   197   198   199   200