Spectral and scattering theory for wave propagation in perturbed stratified media (Q5917637)

The paper deals with the propagation of acoustic and electromagnetic waves in \((m+n)\)-dimensional medium which is a perturbation of a stratified medium. The propagation of such waves is described by the wave equation \[ (\partial^2_t + \widetilde H) u = 0, \] where \(H : = - f^2(z) \nabla_z g(z) \nabla_z\), \(z = (x,y) \in \mathbb{R}^{m + n}\) and \(\nabla_z \equiv (\partial/ \partial z_1, \ldots, \partial/ \partial z_{m + n})\). The authors consider media which are perturbations of stratified media where the functions \(f\) and \(g\) are functions on \(y \in \mathbb{R}^n\) only, and for this case they label these functions as \(f_0\) and \(g_0\). So, they consider also the case \((\partial^2_t + \widetilde H_0) u = 0\), where \(\widetilde H_0 : = - f^2_0(y) \nabla_z g_0(y) \nabla_z\). The aim of the authors is to prove complete results about the spectral and scattering theory of \(\widetilde H\) and \(\widetilde H_0\) under very general and natural conditions on the functions \(f,g\) and \(f_0, g_0\). These conditions are the following: (i) The real-valued functions \(f_0, g_0, f,g\) are bounded from below by strictly positive constants; (ii) The functions \(f_0\), \(g_0\) are supposed to be represented as \(g_0 = g_0' + g_0''\), \(f_0 = f_0' f_0''\) where (a) the real-valued functions \(g_0'\), \(f_0'\) are bounded from below by strictly positive constants; \(g_0'\) is Lipschitz and \(f_0'\) has Lipschitz derivatives; (b) the following decay at infinity conditions are satisfied: \[ \int^\infty_1 {\underset |y |\geq {r} {\text{ess sup}}} |y |\left[ \bigl |\nabla g_0'(y) \bigr |+ \bigl |\nabla f_0'(y) \bigr |+ \biggl |\nabla \bigl( \nabla f_0'(y) \bigr) \biggr |\right] r^{-1} dr < \infty; \] \[ \int^\infty_1 {\underset |y |\geq {r} {\text{ess sup}}} \biggl[ \bigl |g_0''(y) \bigr |+ \bigl |f_0''(y) - 1 \bigr |\biggr] dr < \infty. \] Instead of studying the operators \(\widetilde H\) and \(\widetilde H_0\), the authors consider the unitarily equivalent operators \[ \widetilde H = U_1 HU_1^{-1}, \quad \widetilde H_0 = U_0 HU_0^{-1} \tag{1} \] where \({\mathcal H}_1 \ni \psi \mapsto V_1 \psi : = f^{-1} \psi \in {\mathcal H}\), \({\mathcal H}_0 \ni \psi \mapsto V_0 \psi : = (f_0)^{-1} \psi \in {\mathcal H}\). Here \({\mathcal H}_1, {\mathcal H}_0, {\mathcal H}\) are the next Hilbert spaces: \({\mathcal H}_1 : = L^2 (\mathbb{R}^{m + n}; f^{-2} d^{m + n} z)\), \({\mathcal H}_0 : = L^2 (\mathbb{R}^{m + n}; (f_0)^{-2} d^{m + n} z)\), \({\mathcal H} \equiv L^2 (\mathbb{R}^{m + n}; d^{m + n} z)\). The main result of the paper is the following: Suppose the conditions (i)--(ii) hold, and let \(\widetilde H_0\) be the operator defined by (1). Then \(\widetilde H_0\) has a purely absolutely continuous spectrum, equal to \([0, \infty)\) and the function \[ z \mapsto f_0 (\widetilde H_0 - z)^{-1} f_0 \in B ({\mathcal H}^{-1}_{1/2,1}; {\mathcal H}^1_{-1/2, \infty}), \] which is a priori defined and holomorphic for \(\text{Im} z \neq 0\), extends to a weak \(*\)-continuous function on each of the regions \[ \mathbb{C}^\pm (\widetilde H^r) : = \bigl\{ z \in \mathbb{C} \mid z \notin \sigma_{pp} (\widetilde H^r) \cup \{0\}; \pm \text{Im} z \geq 0 \bigr\}. \]