Modified scattering for the cubic nonlinear Schrödinger equation with long-range potentials in one space dimension
Masaki Kawamoto, Haruya Mizutani · 2024
Read Modified scattering for the cubic nonlinear Schrödinger equation with long-range potentials in one space dimension by Masaki Kawamoto, Haruya Mizutani (2024) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2412-16872.txt.
Introduction
Masaki Kawamoto and Haruya Mizutani proved in 2024 a result on how waves governed by a cubic nonlinear equation scatter when a long-range force is present, in one space dimension. Over long times the wave should settle toward a free wave with a fixed extra phase shift. That shift depends on both the long-range force and the nonlinear term. The proof uses a simple energy method and avoids heavy global estimates.
Catalog record
| CREATOR | Masaki Kawamoto, Haruya Mizutani |
|---|---|
| YEAR | 2024 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 45 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2412-16872
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Modified scattering for the cubic nonlinear Schrödinger equation with long-range potentials in one space dimension Masaki Kawamoto, Haruya Mizutani Abstract . We consider the cubic nonlinear Schrödinger equation with long-range linear potentials in one space dimension, and prove the modified scattering in the energy space for the associated final state problem with a prescribed small asymptotic profile. Compared with the leading term of the free solution, the asymptotic profile has an additional phase correction depending both on the long-range part of the potential and on the nonlinear term. The proof is based on a simple energy method and does not rely on global-in-time Strichartz estimates for Schrödinger equations with linear potentials. In particular, the class of potentials to which our theorem applies is large enough to accommodate slowly decaying negative potentials so that the associated Schrödinger operators may have negative eigenvalues. 2020 Mathematics Subject Classification Primary: 35R03; Secondary: 35B40, 35P25 1. Introduction The present paper is concerned with the scattering problem for the following cubic nonlinear Schrödinger equation (NLS) with a linear potential in one space dimension: \displaystyle i\partial_{t}u-H_{0}u-Vu=F(u),\quad x\in{\mathbb{R}},\quad t\in{\mathbb{R}}, ( 1.1) where u=u(t,x) is a {\mathbb{C}} -valued unknown function, F(u)=\lambda|u|^{2}u with \lambda\in{\mathbb{R}} , and H_{0}+V is the Schrödinger operator with a real-valued potential V : H_{0}:=-\frac{1}{2}\frac{d^{2}}{dx^{2}},\quad V=V(x):{\mathbb{R}}\to{\mathbb{R}}. We are interested in the (small data) modified scattering for the associated finial state problem. More precisely, given a prescribed sufficiently small asymptotic profile function u_{\mathrm{p}}(t,x) , we construct a unique global solution u\in C({\mathbb{R}};H^{1}({\mathbb{R}})) to ( 1.1 ) which scatters to u_{\mathrm{p}} as t\to\infty : \lim_{t\to\infty}\|u(t)-u_{\mathrm{p}}(t)\|_{H^{1}({\mathbb{R}})}=0, where H^{s}({\mathbb{R}}) denotes the L^{2} -Sobolev space of order s . Here the term “modified” means that the asymptotic profile u_{\mathrm{p}} has an additional correction term compared with the leading term of the free solution e^{-itH_{0}}u_{+} , depending both on F(u) and on the long-range part of V (see ( 1.5 ) for the definition of u_{\mathrm{p}} ). This gives the first positive result on the modified scattering for ( 1.1 ) , and is a continuation of our resent work [ 18 ] for the two and three space dimensional cases. To state the result, we first introduce the class of potentials. Assumption A . V is decomposed into the short-range, long-range and compactly supported singular parts as V=V^{\mathrm{S}}+V^{\mathrm{L}}+V^{\mathrm{C}} with real-valued functions V^{\mathrm{S}},V^{\mathrm{L}},V^{\mathrm{C}} satisfying following: • Short-range part: V^{\mathrm{S}}\in C^{1}({\mathbb{R}}) and there exists \rho_{\mathrm{S}}>3/2 such that, for any k=0,1 , \displaystyle\left|\frac{d^{k}}{dx^{k}}V^{\mathrm{S}}(x)\right|\lesssim{\langle}x{\rangle}^{-\rho_{\mathrm{S}}} • Long-range part: V^{\mathrm{L}}\in{C^{3}}({\mathbb{R}}) and there exists \rho_{\mathrm{L}}>1/2 such that, for any k=0,1,2,3 , \displaystyle\left|\frac{d^{k}}{dx^{k}}V^{\mathrm{L}}(x)\right|\lesssim{\langle}x{\rangle}^{-\rho_{\mathrm{L}}-k}. • Compactly supported singular part: V^{\mathrm{C}}\in L^{2}({\mathbb{R}}) and V^{\mathrm{C}} is compactly supported. Under Assumption A , V is infinitesimally H_{0} -bounded, that is, for any {\varepsilon}>0 there exists C_{\varepsilon}>0 such that \|Vf\|_{L^{2}}\leq{\varepsilon}\|H_{0}f\|_{L^{2}}+C_{\varepsilon}\|f\|_{L^{2}},\quad f\in H^{2}({\mathbb{R}}). In particular, H:=H_{0}+V with domain D(H)=H^{2}({\mathbb{R}}) is self-adjoint on L^{2}({\mathbb{R}}) by the Kato–Rellich theorem, and generates the associated unitary group e^{-itH} on L^{2}({\mathbb{R}}) . A typical example of V we have in mind is of the form V(x)=Z{\langle}x{\rangle}^{-\rho}+V^{\mathrm{C}}(x) with some \rho>1/2 and Z\neq 0 . When Z<0 , it is well known that H has negative eigenvalues. We next introduce the asymptotic profile u_{\mathrm{p}} . Let c_{0}>0 and \chi\in C_{0}^{\infty}({\mathbb{R}}) be such that 0\leq\chi\leq 1 , \chi(x)=1 for |x|\leq c_{0}/4 and \chi(x)=0 for |x|\geq{c_{0}/3} . Define an effective potential \displaystyle V_{T_{1}}(t,x)=V^{\mathrm{L}}(x)\left\{1-\chi\left(\frac{2x}{t+T_{1}}\right)\right\}, ( 1.2) where T_{1}\geq 1 is a large constant specified in Proposition 1.1 below. It is easily seen that • V_{T_{1}}\equiv V^{\mathrm{L}} if t\geq 0 and |x|\geq c_{0}(t+T_{1})/4 ; • V_{T_{1}}\in C^{\infty}([0,\infty);C^{3}({\mathbb{R}})) and |\partial_{x}^{k}V_{T_{1}}(t,x)|\leq C_{\alpha}{\langle}t{\rangle}^{-\rho_{\mathrm{L}}-k} on [0,\infty)\times{\mathbb{R}} . Proposition 1.1 . Let c_{0}>0 . Then, for sufficiently large T_{1}>0 , one can construct a global solution \Psi\in C^{1}([1,\infty)\times{\mathbb{R}})\cap C([1,\infty);C^{3}({\mathbb{R}})) to the Hamilton–Jacobi equation \displaystyle-\partial_{t}\Psi(t,x)=\frac{1}{2}|\partial_{x}\Psi(t,x)|^{2}+V_{T_{1}}(t,x) ( 1.3) such that, for k=1,2,3 and t\geq 1 , \displaystyle\left\|\partial_{x}^{k}\left(\Psi(t,x)-\frac{x^{2}}{2t}\right)\right\|_{L^{\infty}({\mathbb{R}})}\lesssim\begin{cases}{\langle}t{\rangle}^{1-\rho_{\mathrm{L}}-k}&\mbox{if}\ \rho_{\mathrm{L}}\neq 1,\\ \left\langle t\right\rangle^{-k}\log(tT_{1}^{-1}+1)&\mbox{if}\ \rho_{\mathrm{L}}=1.\end{cases} ( 1.4) In particular, for any \rho_{\mathrm{L}}^{\prime}<\rho_{\mathrm{L}} , \left\|\partial_{x}^{k}\left(\Psi(t,x)-\frac{x^{2}}{2t}\right)\right\|_{L^{\infty}({\mathbb{R}})}\lesssim{\langle}t{\rangle}^{1-\rho_{\mathrm{L}}^{\prime}-k},\quad t\geq 1,\quad k=1,2,3. Proof. The proposition is basically well-known in the context of long-range linear scattering theory (see e.g. [ 5 , Sections 1.5, 1.8 and A.3] ). Moreover, the proof is essentially same as that of [ 18 , Proposition 3.1] . We thus give an outline of the proof in Appendix A below, and refer to [ 18 , Proposition 3.1] for more details. ∎ Using the function \Psi(t,x) , we define u_{\mathrm{p}} by \displaystyle u_{\mathrm{p}}(t,x)=[\mathcal{M}_{\Psi}(t)\mathcal{D}(t)w_{\mathrm{p}}](t,x)=(it)^{-1/2}e^{i\Psi(t,x)}e^{-i\lambda|\widehat{u_{+}}(x/t)|^{2}\log|t|}\widehat{u_{+}}(x/t), ( 1.5) where u_{+}(x) is a given function (often called the scattering datum) and \displaystyle w_{\mathrm{p}}(t,x) \displaystyle=e^{-i\lambda|\widehat{u_{+}}(x)|^{2}\log|t|}\widehat{u_{+}}(x), \displaystyle\mathcal{D}(t)f(x) \displaystyle=(it)^{-1/2}f(x/t), \displaystyle\mathcal{M}_{\Psi}(t)f(x) \displaystyle=e^{i\Psi(t,x)}f(x), \displaystyle\widehat{f}(x)={\mathcal{F}}f(x) \displaystyle=\frac{1}{\sqrt{2\pi}}\int_{\mathbb{R}}e^{-ix\cdot\xi}f(\xi)d\xi. We here recall that the leading term of the free solution e^{-itH_{0}}u_{+} satisfies e^{-itH_{0}}u_{+}=\mathcal{M}(t)\mathcal{D}(t)\widehat{u_{+}}+o(1) in L^{2} as t\to\infty , where \mathcal{M}(t)f(x)=e^{i|x|^{2}/(2t)}f(x) and the phase function \Psi_{0}=|x|^{2}/(2t) solves the free Hamilton–Jacobi equation -\partial_{t}\Psi_{0}=\frac{1}{2}|\partial_{x}\Psi_{0}|^{2} . Thus, compared with this free profile \mathcal{M}(t)\mathcal{D}(t)\widehat{u_{+}} , u_{\mathrm{p}} has the additional phase correction terms e^{-i\lambda|\widehat{u_{+}}(x/t)|^{2}\log|t|} and e^{i(\Psi-\Psi_{0})} depending on F(u) and V^{\mathrm{L}} , respectively. Let H^{s,r}({\mathbb{R}}) be the weighted L^{2} -Sobolev space defined by \displaystyle H^{s,r}({\mathbb{R}}) \displaystyle=\{f\in\mathcal{S}^{\prime}({\mathbb{R}})\ |\ \|f\|_{H^{s,r}}<\infty\}, \displaystyle\|f\|_{H^{s,r}} \displaystyle=\|{\langle}\partial_{x}{\rangle}^{s}{\langle}x{\rangle}^{r}f\|_{L^{2}({\mathbb{R}})}=\|{\langle}\xi{\rangle}^{s}{\mathcal{F}}[{\langle}x{\rangle}^{r}f]\|_{L^{2}({\mathbb{R}})}, where {\langle}x{\rangle}=\sqrt{1+|x|^{2}} . Note that H^{s}({\mathbb{R}})=H^{s,0}({\mathbb{R}}) . We now state the result. Theorem 1.2 . Suppose that V satisfies Assumption A . Let c_{0}>0 and T_{1} in Proposition 1.1 be fixed. Let b>2 and \delta\leq 1 be such that \frac{1}{2}<\delta<\min\{\rho_{\mathrm{L}},\rho_{\mathrm{S}}-1\}. Then, for any u_{+}\in H^{1,2\delta}({\mathbb{R}}) with sufficiently small \|\widehat{u_{+}}\|_{L^{\infty}} and \mathop{\mathrm{supp}}\nolimits\widehat{u_{+}}\subset\{|\xi|\geq c_{0}\} , there exists a unique solution u\in C({\mathbb{R}};H^{1}({\mathbb{R}})) to ( 1.1 ) satisfying the prescribed asymptotic condition: \displaystyle\|u(t)-u_{\mathrm{p}}(t)\|_{H^{1}({\mathbb{R}})}\lesssim t^{-\delta}(\log t)^{b},\quad t\to\infty. ( 1.6) Remark 1.3 . Since \|u_{\mathrm{p}}(t)\|_{L^{\infty}}\leq|t|^{-1/2}\|\widehat{u_{+}}\|_{L^{\infty}} , the solution u also enjoys the same L^{\infty} -decay estimate as for the free solution e^{-itH_{0}}u_{+} by the Sobolev embedding: \|u(t)\|_{L^{\infty}({\mathbb{R}})}\lesssim t^{-1/2},\quad t\to\infty. The modified scattering for the standard NLS \displaystyle i\partial_{t}u-H_{0}u=\lambda|u|^{2/d}u,\quad x\in{\mathbb{R}}^{d},\quad t\in{\mathbb{R}},\quad\lambda\in{\mathbb{R}},\quad d=1,2,3, ( 1.7) has been extensively studied both for the final state and Cauchy problems. It is impossible to list all of the known literature, instead we refer to some of important results [ 28 , 8 , 10 , 1 , 13 , 6 ] for the final state problem and [ 12 , 4 , 19 , 17 , 15 ] for the Cauchy problem. We also refer to [ 21 , 22 , 16 ] for more recent development. For the cubic NLS ( 1.1 ) satisfying the very short-range condition {\langle}x{\rangle}V\in L^{1}({\mathbb{R}}) , there also many works on the modified scattering (see e.g. [ 27 , 26 , 7 , 2 , 3 , 29 , 20 ] ), where the correction term of the asymptotic profile compared with the free solution is independent of the potential. However, if both of the linear potential V and nonlinear term F(u) are of long-range type, then the literature is much more sparse. In fact, although it has been shown in [ 25 ] that no non-trivial solutions scatter to the free solution, there was no previous positive result before [ 18 ] in such a mixed situation. In the previous paper [ 18 ] , we established the modified scattering for the following NLS: \displaystyle i\partial_{t}u-H_{0}u-Vu=\lambda|u|^{2/d}u,\quad x\in{\mathbb{R}}^{d},\quad t\in{\mathbb{R}},\quad d=2,3. ( 1.8) where V is of long-range type and assumed to satisfy not only a similar condition as Assumption A , but also the repulsive condition, which roughly means that V is positive and monotonously decreasing in the radial direction. The proof essentially relied on the global-in-time Strichartz estimates for e^{-itH} proved by [ 24 , 30 ] . Very recently,