Page 348 -
P. 348

332    6. Introduction to Pseudodifferential Operators

                                                             n
                           where the cut–off function χ ∈ C (IR ) satisfies
                                                        ∞
                                                                             1
                                   χ(ξ)=1 for |ξ|≥ 1and χ(ξ)=0 for |ξ|≤        .       (6.2.19)
                                                                             2
                           We define the parametrix by
                                                        −1
                                                 Qf := F   (q −2 (ξ)F y →ξ f)
                                                        ξ →x
                                     −2   n
                           with Q ∈L   (IR ); and we have
                                                   u(x)= Qf(x)+ Rf(x)

                           with the remainder
                                                                    1
                                                      −1
                                            Rf(x)= F       1 − χ(ξ)    f(ξ) .          (6.2.20)
                                                                        )
                                                      ξ →x         a(ξ)
                           Since f ∈ C (Ω)wehave f ∈ C  ∞  from the Paley–Wiener–Schwartz theorem
                                      ∞
                                     0
                                                   )
                                                                             1
                           3.1.3 (iii). Hence, with 1 − χ(ξ) =0 for |ξ|≥ 1and   having a singu-
                                                                            a(ξ)
                           larity only at ξ = 0 and having quadratic growth, due to Lemma 3.2.1,

                                        f(ξ) defines a distribution in E . Hence, Rf ∈ C
                            1 − χ(ξ)  1 )                                           ∞  and R ∈
                                     a(ξ)
                                                                                           n
                             −∞                                                  −∞
                           L    (Ω) by applying Theorem 3.1.3 (ii). (Note that R  ∈ OPS  (Ω ×IR )).
                                             −2
                              Thus, Q + R ∈L   (Ω) defining again N := Q + R, the Newton potential
                           operator, which can now still be written as
                                                             1

                                                      −1
                                            Nf(x)= F           F y →ξ f(y) (x) .       (6.2.21)
                                                      ξ →x
                                                           a(ξ)
                           As in the previous case, we still can define the fundamental solution
                                                    E(x, y)= Nδ(x − y)                 (6.2.22)
                           where δ(x − y) is the Dirac functional with singularity at y. Note that δ ∈
                                                        n
                                              n
                                n
                           E (IR )and N : E (IR ) →D (IR ) is well defined. E(x, y) will satisfy (3.6.4).



                              We now return to the more general scalar elliptic equation (5.1.1) with
                             ∞
                           C –coefficients (but p = 1). Here, the symbol a(x, ξ) in (6.2.4) depends on
                           both, x and ξ. The construction of the symbols (6.2.8) and (6.2.9) leads to
                           an asymptotic expansion and via (6.1.18), i.e., by
                                                        ∞
                                                              ξ
                                              q(x, ξ):=   Ξ      q −2−j (x, ξ)
                                                              t j
                                                       j=0
                           we define the symbol q where Ξ ∈ C  ∞  is a cut–off function satisfying Ξ(y)=0
                                   1                                  −2      n
                           for |y|≤  and Ξ(y)=1 for |y|≥ 1. Then q ∈ S S S  (Ω × IR )and
                                   2
   343   344   345   346   347   348   349   350   351   352   353