Page 351 -
P. 351

6.2 Elliptic Pseudodifferential Operators on Ω ⊂ IR n  335

                                           2κ−n
                            L 0 (x, y)= |x − y|    00 (x, |x − y|, Θ) + log |x − y|  01 (x, |x − y|, Θ) (6.2.34)
                                                            )
                                                                                    )
                                          :
                                                                                  ∞
                           with Θ =(x − y) |x − y| where the functions   00 and   01 are C (Ω × IR + ×
                                )
                                   n
                           {Θ ∈ IR ||Θ| =1}). The integral operator defined by the kernel L 0 (x, y)
                             )
                                      )
                           will be denoted by N ;
                                            ∼0
                                                                      i(x−y)·ξ
                                                            1        e
                             (N v)(x):=    L 0 (x, y)v(y)dy =               v(y)dydξ .  (6.2.35)
                                                                      0
                              ∼0                          (2π) n     a (y, ξ)
                                                                      2κ
                                                                 n
                                         Ω                     IR Ω
                           This is a pseudodifferential operator of the form (6.1.26) with the amplitude
                           function
                                                              0
                                                  a(x, y, ξ)= a (y, ξ) −1  ;
                                                              2κ
                           and L 0 (x, y) is the Schwartz kernel of N .
                                                             ∼0
                                                                     0
                              We remark that the only singularity of a (y, ξ) −1  at ξ = 0 can be
                                                                     2κ
                           handled, by introducing the cut–off function χ(ξ), in the same manner
                                                                                        n
                                                           0
                           as in Remark 3.2.2. Hence, χ(ξ)a (y, ξ) −1  ∈ S S S −2κ (Ω × Ω × IR )and
                                                           2κ
                                  −2κ
                           N ∈L      (Ω).
                            ∼0
                              To derive higher order Levi functions, we return to (6.2.30). Define the
                           pencil of pseudodifferential operators by
                                                             −1     0      −1
                                      A 0 (x, −iD; y)f (x)  = F   a (y, ξ)  F z →ξ f(z)
                                                             ξ →x  2κ

                                                        =     E 0 (x, z; y)f(z)dz .    (6.2.36)
                                                            Ω
                                              −2κ
                           Then A 0 (x, D; y) ∈L  (Ω) for every y ∈ Ω.Moreover,
                                 P 0 (y)A 0 (x, −iD; y)= I and A 0 (x, −iD; y)P 0 (y)= I.  (6.2.37)
                              In accordance with the Neumann series, we introduce the recursive se-
                           quence of operators
                                               j


                               A j (x, −iD; y):=  A 0 (x, −iD; y)T(y) A 0 (x, −iD; y)  (6.2.38)
                                               =0
                                           = A 0 (x, −iD; y)+ A j−1 (x, −iD; y)T(y)A 0 (x, −iD; y) ,
                           j =1, 2,... . We note that
                                                             j

                            PA j (x, −iD; y)  =  P 0 (y) − T(y)  A 0 (·, −iD; y)T(y) A 0 (x, −iD; y)
                                                             =0
                                                                   j+1

                                           = I − T(y)A 0 (·, −iD; y)                   (6.2.39)
                           for every y ∈ Ω and every j ∈ IN 0 .
   346   347   348   349   350   351   352   353   354   355   356