Page 409 -
P. 409

7.1 Pseudohomogeneous Kernels  393




                                a 0  (x, ξ) = lim   k κ+j x, z ψ  z  e −iξ·z dz for x ∈ Ω  (7.1.81)
                                 m−j                           t
                                            t→+∞
                                                IR n
                                                                   1
                           where the cut–off function ψ(z)=1 for |z|≤  and ψ(z)=0 for |z| > 1(see
                                                                   2
                           Lemma 7.1.4, equations (7.1.28) and (7.1.29)).
                           For m  ∈ IN 0 and m − j> 0 we have

                                           a 0 m−j (x, ξ)= p.f.  k κ+j (x, z)e −iξ·z dz  (7.1.82)
                                                           IR n
                           (see Theorem 7.1.6, formula (7.1.53)).

                           For m ∈ IN 0 and m − j ≥ 0 and if the Tricomi conditions (7.1.56) are
                           satisfied, then we have


                                                         α

                               a 0  (x, ξ)=       c α (x)(iξ) +p.f.  k κ+j (x, z)e −iξ·z dz  (7.1.83)
                                m−j
                                           |α|=m−j               IR n
                           (see Theorem 7.1.7, formula (7.1.57)).
                           Symbol to kernel
                                                    m
                              For the operator A ∈L (Ω), let the classical symbol a(x, ξ) be given
                                                    c
                           by its homogeneous classical asymptotic symbol expansion (7.1.80). Then
                           the corresponding pseudohomogeneous kernel expansion can be calculated
                           explicitly as follows.
                           For m − j< 0 we have

                                                         a
                             k κ+j (x, z)=(2π) −n  p.f.  e iz·ξ 0  (x, ξ)dξ for x ∈ Ω and z ∈ IR n
                                                          m−j
                                                   IR n
                                                                                       (7.1.84)
                           (see in the proof of Theorem 7.1.1, formulae (7.1.48), (7.1.51)).
                           For m − j ≥ 0 we have

                                                              a
                                      k κ+j (x, z)=(2π) −n  e iξ·z 0  (x, ξ)ψ(ξ)dξ
                                                               m−j
                                                       IR n
                                                                                       (7.1.85)

                                     −2     −n    iξ·z          0
                                 +|z|   (2π)     e   (−∆ ξ ) a m−j (x, ξ) 1 − ψ(ξ)  dξ
                                              IR n
                           where   ∈ IN satisfying 2 >m + n − j (see in the proof of Theorem 7.1.8 i),
                           formula (7.1.64)). This formula is valid for arbitrary m ∈ IR.
                           If m ∈ IN 0 then the Schwartz kernel is given by (7.1.85) for j =0,...,m
                           whereas the operator a 0 m−j (x, −iD) contains a differential operator of the
   404   405   406   407   408   409   410   411   412   413   414