Page 472 - DMTH504_DIFFERENTIAL_AND_INTEGRAL_EQUATION
P. 472

Unit 31: Fredholm Equations with Poincere Goursat Kernels




          In fact, conditions (26) are necessary because if equation (1) for   =   admits a certain solution  Notes
                                                                 0
            (x), then from the equation itself, it follows that
                           x
                                x
                                            x
                   x
             x
            f  ( )  0h ( )dx  ( )  0h ( )dx  0  0h ( )dx K ( , ) ( )dy
                                                  x
                                                    y
                                                        y
                           x
                                                         dx
                                                 x
                                                        x
                          ( )  ( )dx      ( )dy K ( , )  ( ) .
                                          y
                                x
                                                   y
                              0h      0               0h
          But, since   and   (x) are eigenvalue and corresponding eigenfunction of the associated Kernel,
                   0     0h
          we have
                     K ( , )  ( )dx   ( );
                       x
                                      y
                              x
                         y
                   0        0h      0h
          hence
                          x
                   f  ( )  0h ( )dx  0
                     x
          Furthermore, conditions (26) are also sufficient, since from them it can be easily deduced that the
          non-homogeneous system (7), which we shall write briefly as
                     b     b  ...,  b  ,
                   1  1  2  2   n   n
          reduces to only n   r independent equations. Consequently we can now solve it readily (carrying
          r unknowns on the right hand side), since the characteristic of matrix of the coefficients is exactly
          p = n   r.
          We can reduce the system for the following reason: Let us multiply the previous equations by
           *
               *
                    *
          B , B , ...B , respectively and add. Bearing in mind equations (25), we have
            h1  h2   hr
                   n
                      *
                    B hk  k  [(1  a 11 )B h *  1  a B *  2  ...  a B *  ]  1
                                        21 h
                                                   n
                                                   1 hn
                  k  1
                          [  a B * 1  (1  a 22 )B h * 2  ...  a B *  ]  2
                             12 h
                                                    2 hn
                                                   n
                          ..................................................................
                         [  a B *   a B *  ... (1  a  )B *  ]  0,
                             1n h 1  2n h 2       nn  hn  n
          while on the other side, by virtue of (26), we also have
                   n         n
                      *
                                *
                                   x
                                        x
                                                  x
                                                     x
                                                   f
                    B b        B Y  ( ) f ( )dx  ( ) ( )dx  0.
                     hk k       hk k            oh
                  k  1      k  1
          Among other things, form (27) of the solution demonstrates the following  obvious fact:  the
          general solution of equation (1) when D( ) = 0 can be considered as the sum of any particular
          solution  (x) and of the general solution (18) of the homogeneous equation.
          Thus we have proved for PG Kernels the following basic Fredholm theorem, which will be
          extended to general Kernels in the next section:
          Fredholm’s integral equation of the second kind
                             y
                                y
                           x
                   x
                   ( )   K ( , ) ( )dy  f  ( )
                                       x
          has, in general, one and only one solution of the class  L  given by the formula
                                                       2
                   x
                   ( )  f ( )  H ( , ; ) ( )dy ,
                                        y
                         x
                                x
                                  y
                                      f
                                           LOVELY PROFESSIONAL UNIVERSITY                                   465
   467   468   469   470   471   472   473   474   475   476   477