Page 86 -
P. 86

70     2. Boundary Integral Equations

                                    1
                           operator ( I − K) has only one eigensolution ϕ , as well. It remains to show
                                    2                                0
                           that n is also the only linear independent eigensolution to V and satisfies
                           (2.3.32).
                              As we will show later on, for V , the Fredholm theorems are also valid and
                                 α
                           V : C (Γ) → C  α+1 (Γ) has the Fredholm index zero. Let τ 0 be any solution
                           of V τ 0 = 0. Then the single layer potential

                                                 u 0 = V τ 0 with p 0 = Φτ 0
                                                                          c
                           is a solution of the Stokes system in Ω as well as in Ω with u 0 | Γ = 0. Then
                                                                         c
                                      3
                           u 0 ≡ 0 in IR and the associated pressure is zero in Ω and p 0 = β = constant
                           in Ω. As a consequence we have from the jump relations
                                                                        !
                                      T x (u 0− ,p 0 ) − T x (u 0+ , 0) = σ(u 0 ,p 0 )n | Γ = −βn ,
                           therefore τ 0 = −βn.
                              On the other hand, V n| Γ = 0 follows from the fact that u := V n and
                           p := Πn is the solution of the exterior homogeneous Neumann problem of
                           the Stokes system since
                                                            1

                                               T(V u)| Γ =(− I + K )n = 0 .
                                                            2
                           and therefore vanishes identically (see [179, Theorem 1 p.60]).
                              In order to show (2.3.32), we consider the solution of the exterior Dirichlet
                                               +
                           Stokes problem with u | Γ = ϕ  = 0 which admits the representation
                                                       0
                                                    u(x)= Wϕ − V τ .
                                                              0
                           Then it follows that the corresponding simple layer term has vanishing bound-
                           ary values,
                                                                   1
                                                      1
                                       −V τ| Γ = ϕ − ( I + K)ϕ =( I − K)ϕ = 0 .
                                                  0   2        0   2        0
                           Hence, τ = βn with some constant β ∈ IR. Application of T x | Γ gives
                                                            1

                                          τ = βn = −Dϕ +( I − K )βn = −Dϕ .
                                                                              0
                                                        0
                                                            2
                              The case β = 0 would imply τ = 0 and then u(x) solved the homogeneous
                           Neumann problem which has only the trivial solution [179, p. 60] implying
                           ϕ = 0 which is excluded. So, β  = 0 and scaling of ϕ implies (2.3.32).
                             0
                                                                          0
                                              1
                           ii) For the operator ( I+K) having the eigenspace (a+b×x)| Γ of dimension
                                              2
                                                                            1
                           6 we refer to [179, p. 62]. Hence, the adjoint operator ( I + K ) also has a

                                                                            2
                           6–dimensional eigenspace due to the classical Fredholm theory since K is a
                           compact operator. For the operator D let us consider the potential u(x)=
                                      c
                           Wv(x)in Ω with v = a+b×x| Γ . Then u is a solution of the Stokes problem
                           and on the boundary we find
                                                          1
                                                   +
                                                  u | Γ =( I + K)v = 0 .
                                                          2
   81   82   83   84   85   86   87   88   89   90   91