Page 408 -
P. 408

392    7. Pseudodifferential Operators as Integral Operators

                                               a β (x) = 0 for all |β| = m − 1 .

                           Repeating this process we finally obtain a β (x) = 0 for all |β|≤ m which
                           assures the representation (7.1.77) of our theorem.


                           7.1.4 A Summary of the Relations between Kernels and Symbols

                                                                                         m
                           As we have seen in this section, a pseudodifferential operator A ∈L (Ω)
                                                                                         c
                           can be expressed in terms of either a given Schwartz kernel or in terms
                           of its symbol. The relations between these two representations have been
                           given explicitly so far. However, for ease of reading, we now summarize these
                           relations in the following.
                           Kernel to symbol
                                                   m
                              Let the operator A ∈L (Ω) be given in the form (7.1.8) or, more pre-
                                                   c
                           cisely, in the form (7.1.60) where the Schwartz kernel of the finite part integral
                           operator in (7.1.60) is a given function in the class Ψhk κ (Ω) with κ = −m−n
                           for fixed m ∈ IR (see Definition 7.1.1). Then this kernel has an asymptotic
                           expansion in the form (7.1.2), i.e.,

                                                ∞

                                    k(x, x − y) ∼  k κ+j (x, x − y) where k κ+j ∈ Ψhf κ+j .
                                                j=0
                           In the case that m − j ∈ IN 0 , we assume that the corresponding terms in the
                           asymptotic expansion satisfy the Tricomi conditions (7.1.56), i.e.,


                                            α
                                           Θ k κ+j (x, Θ)dω(Θ) = 0 for all |α| = m − j,
                                      |Θ|=1
                           (see Theorem 7.1.7). For n = 1, again this formula is interpreted as in (3.2.14).
                           With a properly supported part A 0 of A, the symbol corresponding to A 0
                           can be computed according to the formula (6.1.23), i.e.,

                                                a(x, ξ):= e −ix·ξ A 0 (e iξ• )(x) .

                                                           m
                           The symbol a(x, ξ) admits for A ∈L (Ω) a classical asymptotic expansion
                                                           c
                                                           ∞
                                                               0
                                                  a(x, ξ) ∼   a   (x, ξ)               (7.1.80)
                                                               m−j
                                                           j=0
                           (see (6.1.42)). Each term in the expansion (7.1.80) can be calculated from the
                           corresponding pseudohomogeneous terms of the kernel expansion explicitly
                           depending on m − j.
                           For m − j< 0 we have:
   403   404   405   406   407   408   409   410   411   412   413