Page 60 - 4811
P. 60

Fourier-Mellin formula


                   Summarizing the result claimed in (4.7):

                   Theorem 4.2È

                   Suppose that f is continuous and f piecewise continuous on [0, ∞), with

                   f of exponential order α on [0, ∞). If F(s) = L(f(t)), for Re(s) = x > α,
                   also satisfies the growth condition

                                                               M
                                                   |F(s)| ≤        , p > 0,
                                                              |s| p

                   for all |s| sufficiently large and some p (or condition 1) above), and if F(s)
                   is analytic in C except for finitely many poles at z , z , . . . , z , then
                                                                                              n
                                                                                    2
                                                                                1
                                                   ∫                         n
                                               1      x+i∞                 ∑
                                                             ts
                                     f(t) =                 e F(s)ds =          Res(z ),              (4.12)
                                                                                       k
                                              2πi
                                                     x−i∞
                                                                           k=1
                                                                             ts
                   where Res(z ) is the residue of the function e F(s) at s = z .
                                  k
                                                                                               k
               . . . . . . .
                   In view of the properties of the inverse Fourier transform, we have the next
               result.

                   Corollary 4.12
                   If f is only piecewise continuous on [0, ∞), then the value returned
                   by the complex inversion formula (4.12) is


                                                           +
                                                       f(t ) + f(t )
                                                                     −
                                                               2

                   at any jump discontinuity t > 0.
               . . . . .


                   Remark 4.2E The preceding theorem and corollary can be shown to hold under
                                                                                                 √
                   less restrictive conditions on f, so that functions such as f(t) = 1/ t are not
                   excluded by the inversion process. Essentially, the Laplace transform of f should
                   converge absolutely and f should be of “bounded variation” in a neighbourhood of

               . . . . . .  the point t > 0 in question.



                   Example 4.1,
                                                                   1
                                                     F(s) =              .
                                                               s(s − a)

                                                                                                            2
                   Then F(s) has a simple pole at s = 0 and s = a, and |F(s)| ≤ M/|s|
                                                                                        2
                   for all |s| sufficiently large, say |F(s)| ≤ 2/|s| if |s| ≥ 2|a|.
                   Moreover,
                                                                           e ts       1
                                                          ts
                                      Res(0) = lim se F(s) = lim                 = − ,
                                                   s→0               s→0 s − a        a
               . . . . .


                                                              59
   55   56   57   58   59   60   61   62   63   64   65