On the continuous dependence of the poles of the scattering matrix on the coefficients of an elliptic operator (Q794205): Difference between revisions

The system \(u_{tt}(x,t)=\nabla A_{\epsilon}(x)\nabla u(x,t)\), \(u(x,0)=f_ 1(x)\), \(u_ t(x,0)=f_ 2(x)\), is considered. With respect to the matrix \(A_{\epsilon}(x)\) it is assumed that \(<A_{\epsilon}(x)\xi,\xi>\geq c| \xi |^ 2,\quad c>0, A_{\epsilon}(x)=A_ 0(x)+\epsilon a(x)\) and \(A_{\epsilon}(x)=A_ 0(x)=I\) for \(| x| \geq \rho\). The purpose of the paper is to study poles of the scattering matrix \(S_{\epsilon}(z)\), Im z\(>0\), of this problem. These poles coincide with poles of the resolvent \((l_{\epsilon}-z^ 2)^{-1}\) of the corresponding stationary problem. Let \(z_ n\) be poles of \((l_ 0-z^ 2)^{-1}\) of orders \(q_ n\). It is proven that as \(\epsilon \to 0\) poles of \(S_{\epsilon}(z)\) lie in discs \(\{z: | z-z_ n| =O(\epsilon^{1/q_ n})\}.\)