Page 358 -
P. 358

342    6. Introduction to Pseudodifferential Operators

                           where T(y; D) denotes the difference of the differential operators, i.e.,

                                                T(y; D)= a 0 (y, D) − a(x, D) .

                           Now, the inverse of a(x, D) can formally be written in the form of Neumann’s
                           series
                                                                         (−1)

                                       a −1 (x, D)=      a (−1) (y; D)T(y; D) a 0  (y; D) .
                                                       0
                                                   ≥0
                           This leads us to the following successively defined sequence of kernel functions
                                  N (0) (x, y)  := a (−1) (y; D)((δ(x − y)δ jk )) p×p ,  (6.2.57)
                                                 0
                                N ( +1) (x, y)  := a (−1) (y; D)T(y; D)N ( ) (x, y)for   =0, 1,... .
                                                 0
                           The Levi function of order µ to the system of differential equations (6.2.10)
                           assumes the form
                                                             µ
                                                 L (µ) (x, y):=    N ( ) (x, y) .      (6.2.58)
                                                             =0
                              In the same way as for the scalar case we may seek the solution u(x)=

                            u 1 (x),...,u p (x)  to the system (6.2.10) in the form
                                                  p
                                                        (µ)
                                         u k (x)=     L   (x, y)φ   (y)dy for x ∈ Ω    (6.2.59)
                                                        k
                                                  =1
                                                    Ω

                           where φ(x)= φ 1 (x),...,φ p (x)  denotes an unknown vector–valued den-
                           sity. By applying the differential operator A to (6.2.59) we obtain a system
                           of domain integral equations of the second kind,
                                                  p
                                                                (µ+1)
                                  f j (x)= φ j (x) −  T jk (y, D)N  (x, y)φ m (y)dy    (6.2.60)
                                                                km
                                                 k=1
                                                    Ω
                           by using the relation

                                     AL (µ) (x, y)= δ(x − y) − T(x, D)N (µ+1) (x, y) .  (6.2.61)

                           The integral operator in (6.2.60) has the kernel

                                                    T(y, D)N  (µ+1) (x, y)
                                                    λ
                           which belongs to the class C (Ω×Ω) with λ = µ+1−n−max k=1,...,p |s k |.For
                             sufficiently large and compact Ω, the integral equation (6.2.60) is a classical
                           Fredholm integral equation of the second kind with continuous kernel in Ω.
   353   354   355   356   357   358   359   360   361   362   363