Page 71 - 4811
P. 71

Complex inversion formula


               For s on the line segments BC and DE, again we take them to be the respective

               upper and lower edges of the cut along the negative axis. If s lies on BC, then s =
                      √       √
                  iπ
               xe ,     s = i x, and when s goes from −R to −r, x goes from R to r. Therefore,
                                            √                    √                     √
                                 ∫      ts−a s       ∫  −r   ts−a s        ∫  r  −tx−ai x
                                      e                     e                  e
                                               ds =                 ds =                   dx.           (4.22)
                                          s                    s                    x
                                  BC                   −R                   R
                                                     √         √
               Along DE, similarly s = xe       −iπ ,  s = −i x implying

                                                      √                      √
                                           ∫      ts−a s        ∫  R  −tx+ai x
                                                 e                   e
                                                         ds =                   dx.                      (4.23)
                                                    s                     x
                                             DE                  r
               Combining (4.22) and (4.23) after multiplying each by 1/2πi yields

                                                 √          √
                                   ∫  R                                   ∫  R           √
                               1        e −tx (e ai x  − e −ai x )     1       e −tx  sin a x
                                                                dx =                          dx.        (4.24)
                              2πi   r              x                   π   r         x
               Letting r → 0 and R → ∞ in (4.24), we obtain the integral

                                                                     √
                                                     ∫  ∞
                                                  1       e −tx  sin a x
                                                                         dx.                             (4.25)
                                                  π   0          x

               Above we introduced the error function

                                                                 ∫  t
                                                              2          2
                                                 erf(t) = √          e −x  dx.
                                                               π   0

               It can be shown that the integral in (4.25) can be written in terms of the error
               function, that is,
                                           ∫                √               (      )
                                         1    ∞  e −tx  sin a x                 a
                                                                dx = erf       √      ;
                                         π   0          x                     2 t
               we shall use use latter expression.
                                        iθ
                   Finally, for s = re on γ , we can take the integration from π to −π,
                                               r
                                                √                           √
                                      ∫                       ∫  −π     iθ      iθ/2
                                                                                       iθ
                                  1       e ts−a s         1        e tre −a re    ire dθ
                                                  ds =                                      =
                                 2πi         s           2πi                  re iθ
                                       γ r                      π
                                              ∫  π                         ∫  π
                                           1           iθ−a re          1
                                                           √
                                     = −           e tre      dθ → −            dθ= −1
                                           2π   −π                      2π   −π
               for r → 0. We have used here the uniform continuity of the integrand to pass the
               limit inside the integral.
                   Whence, letting r → 0, R → ∞ in (4.21) gives

                                                              √
                                              ∫  x 0 +i∞  ts−a s            (      )
                                           1            e                       a
                                    0 =                          ds + erf       √     − 1;               (4.26)
                                          2πi               s                 2 t
                                                x 0 −i∞
                                                              70
   66   67   68   69   70   71   72   73   74   75   76