An annulus multiplier and applications to the limiting absorption principle for Helmholtz equations with a step potential (Q2225633): Difference between revisions

The authors are interested in the equation \[ -\Delta u + Vu - \lambda u = f \tag{1}\] in \(\mathbb{R}^n\), \(n\geq 2\), as well as a nonlinear pendant, where \(V\) is assumed throughout to be a model step potential taking the constant value \(V_1\) in the upper half-space and \(V_2 < V_1\) in the lower half space. Denote by \(\mathcal{R}(\mu) := (-\Delta + V - \mu)^{-1}\) the resolvent operator associated with (1), \(\mu \in \mathbb{C} \setminus \sigma (-\Delta + V)\). The main result (Theorem 1) is the \(L^p\)--\(L^q\) resolvent estimate \[ \sup_{0 < |\varepsilon| \leq 1} \|\mathcal{R}(\lambda + i\varepsilon)\|_{L^p(\mathbb{R}^n) \to L^q(\mathbb{R}^n)} < \infty \] for all \(\lambda > V_1\) and all pairs \((p,q)\) in a somewhat involved region related to, and almost as large as, the corresponding region in the corresponding result for constant potential \(V \in \mathbb{R}\) (the precise region here is also related to \textit{T. Tao}'s restriction problem in Harmonic Analysis, cf. [in: Fourier analysis and convexity. Boston, MA: Birkhäuser. 217--243 (2004; Zbl 1083.42008)]). In this case, the authors also obtain convergence, as \(\varepsilon \to 0\), of \(\mathcal{R} (\lambda + i\varepsilon)\) to \(\mathcal{R} (\lambda \pm i0)\) in the weak topology of bounded linear operators from \(L^p (\mathbb{R}^n)\) into \(L^q (\mathbb{R}^n)\). Two other results are obtained as corollaries: firstly, for the problem (1), with \(f \in L^p(\mathbb{R}^n)\), for the ``outgoing'' and ``incoming'' solutions \(u_+ := \mathbb{R} (\lambda + i0)f\) and \(u_- := \mathbb{R} (\lambda - i0)f\), respectively, under the same conditions as the main theorem we have \[ u_\pm \in L^q (\mathbb{R}^n) \cap W^{2,p}_{\textrm{loc}}(\mathbb{R}^n) \] together with the estimate \[ \|u_+\|_{L^q(\mathbb{R}^n)} + \|u_-\|_{L^q(\mathbb{R}^n)} \leq C(V_1,V_2,n,p,q,\lambda) \|f\|_{L^p (\mathbb{R})}. \] The second application is to prove the existence of a nontrivial solution \(u \in L^q (\mathbb{R}^n) \cap W^{2,r}_{\textrm{loc}}(\mathbb{R}^n)\) for all \(r<\infty\), of the nonlinear equation \[ -\Delta u + Vu - \lambda u = \Gamma |u|^{q-2}u \] in \(\mathbb{R}^n\), \(n\geq 2\), where \(\Gamma\) is assumed to be a strictly positive \(L^\infty\) function decaying at \(\infty\), and \(\frac{2(n+1)}{n-1} \leq q < \frac{2n}{n-2}\). The proof of the main theorem is based on Fourier restriction theory as well as the analysis of, including obtaining \(L^p\)--\(L^q\) estimates for, a certain singular Fourier multiplier supported in an annular region (Theorems 2 and 3).