On the asymptotic behavior of nonlinear waves in the presence of a short-range potential (Q818805)

The authors consider the existence of a unique global classical solution, scattering and nonscattering for the following nonlinear wave equation with short range potential: \[ \partial^2_1u-\Delta u+V (x)\cdot u=f(u)\text{ in }\mathbb{R}^n\times\mathbb{R}\tag{1} \] where \(F(u)= |u|^p\) or \(F(u)=|u|^{p-1}u\) for some \(p> 1\). It is expected that the solution to (1) with sufficiently small and rapidly decreasing potential \(V(x)\) behaves like the solution to \[ \partial^2_tu -\Delta u=F(u)\text{ in }\mathbb{R}^n\times\mathbb{R}.\tag{2} \] Indeed, such an expectation is true on the property for global existence of a small amplitude solution under some suitable assumptions to \(p\) and the initial data for the global existence of the solution to (1). On the other hand, the authors observed that the situation is quite different when we focus the associated scattering problem. Indeed, the usual scattering property is valid if the power of the nonlinearity \(p\) satisfies some condition concerning with the decaying order of the initial data as \(|x|\to\infty\). Precisely, the properties of main theorems of this paper are described as follows: Let \(u\) be a solution of (1) in the sense of the associated integral equation: \[ u=u_0^-+{\mathcal L} F(u)-{\mathcal L}(Vu),\tag{3} \] where \(u_0^-\) denotes the solution of the free wave equation \[ \partial_t^2u_0-\Delta u_0=0,\tag{4} \] and the Duhamel operator \({\mathcal L}\) is defined by \[ [{\mathcal L}G] (x,t)=\frac {1} {4\pi}\int^t_{-\infty}\frac{1}{t-\tau}\int_{|y-x|=t-\tau}G(y,\tau)\,dS_y \,d\tau. \] Suppose that the initial data and the potential satisfy the following conditions: \[ \sum_{|\alpha|\geq 3}\bigl|\partial_x^\alpha \varphi(x) \bigr|+\sum_{|\alpha|\leq 2}\bigl|\partial_x^\beta\psi(x)\bigr| \leq\varepsilon \bigl(1+|x|\bigr)^{-k-1} \] and \[ \sum_{|\alpha|\leq 2}\bigl| \partial_x^\alpha V(x)\bigr|\leq V_0\bigl(1+|x|\bigr)^{-2-\delta} \] for some \(k>0\), \(\delta>0\) and some small \(\varepsilon>0\) and \(V_0\). Under the assumption \(p>1+\sqrt 2\) with \(k\geq 2/(p-1)\), the authors proved the followings: (I) If \(\varepsilon\) and \(V_0\) are sufficiently small, then the integral equation (3) has a unique classical solution (Existence). (II) Assume that \(\varepsilon\) and \(V_0\) are sufficiently small, \(kp\geq k+2\) for \(k>1\) for the case \(V(x)\not\equiv 0\) and for \(kp>5/2\) for the case \(V(x)=0\). Then there exists a unique solution \(u_0^+\) to (4) such that the solution \(u\) provided by (I) satisfies \(\|u-u_0^\mp\|_E\to 0\) as \(t\to\mp\infty\), where \(\|\cdot \|_E\) denote the usual energy norm. In particular, one can define the scattering operator \(S\): \(u_0^-\to u_0^+\). (III) Assume that \(\varepsilon\) and \(V_0\) are sufficiently small, \(kp\geq k+2\) for \(k\leq 1\) for the case \(0\not\equiv V(x)\leq 0\) and for \(kp\leq 5/2\) for the case \(V(x)\equiv 0\). If the initial data are chosen so that \(\varphi(x)\equiv 0\), \(\psi(x)\leq 0\) and \(|\psi(x)|\geq \varepsilon_1(1+|x|)^{-k-1}\) for some \(\varepsilon_1\), then the solution \(u\) provided by (I) does not satisfy \(\|u-u_0^-\|_E\to 0\) as \(t\to-\infty\).