perturbation of the nonlinear Schrödinger equation by a localized nonlinearity
Gong Chen, Jiaqi Liu, Yuanhong Tian · 2025
Read perturbation of the nonlinear Schrödinger equation by a localized nonlinearity by Gong Chen, Jiaqi Liu, Yuanhong Tian (2025) 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-2508-11463.txt.
Introduction
Gong Chen, Jiaqi Liu, and Yuanhong Tian revisit a classic theory for how small disturbances evolve in integrable wave equations, aiming for cleaner proofs of its core bounds. The machinery, built by Deift and Zhou, tracks solutions of systems like the nonlinear Schrödinger equation over long times. The authors supply new, simpler arguments for the key estimates that keep errors under control.
As a test case, they perturb the one-dimensional defocusing cubic Schrödinger equation with an extra localized higher-order term. Their improved estimates tame the added term's strength. The conclusion is that the disturbed equation keeps the same long-time behavior as its clean, fully integrable version. The authors frame the work as a step toward focusing problems, where solutions can spike rather than spread, and toward applying the same machinery to other integrable models beyond this one equation.
Catalog record
| CREATOR | Gong Chen, Jiaqi Liu, Yuanhong Tian |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 78 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2508-11463
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
perturbation of the nonlinear Schrödinger equation by a localized nonlinearity Thanks: The first author was partially supported by NSF grant DMS-2350301 and by Simons foundation MP-TSM-00002258. The Second author is partially supported by NSFC grant No.12201605 and CAS grant E4ER0101A2. Gong Chen, Jiaqi Liu, Yuanhong Tian Abstract . We revisit the perturbative theory of infinite dimensional integrable systems developed by P. Deift and X. Zhou [ 8 ] , aiming to provide new and simpler proofs of some key L ∞ L^{\infty} bounds and L p L^{p} a priori estimates. Our proofs emphasizes a further step towards understanding focussing problems and extends the applicability to other integrable models. As a concrete application, we examine the perturbation of the one-dimensional defocussing cubic nonlinear Schrödinger equation by a localized higher-order term. We introduce improved estimates to control the power of the perturbative term and demonstrate that the perturbed equation exhibits the same long-time behavior as the completely integrable nonlinear Schrödinger equation. 1. Introduction 1.1. Background It is well-known that the 1 -d defocussing cubic nonlinear Schrödinger equation (NLS) (1.1) iq_{t}+q_{xx}-2|q|^{2}q=0 is a canonical example of infinite dimensional integrable systems, in the sense that it is the compatibility condition for the following commutator: (1.2) \displaystyle\left[\partial_{x}-U,\partial_{t}-W\right]=0, where \displaystyle U \displaystyle=iz\sigma+\left(\begin{array}[]{cc}0&q\\ \bar{q}&0\end{array}\right), \displaystyle W \displaystyle=-iz^{2}\sigma-z\left(\begin{array}[]{cc}0&q\\ \bar{q}&0\end{array}\right)+\left(\begin{array}[]{cc}-i|q|^{2}&i\partial_{x}q\\ -i\partial_{x}\bar{q}&i|q|^{2}\end{array}\right). Moving beyond integrable structures, in the seminal paper by P. Deift and X. Zhou [ 8 ] , after exploring the L^{p} a priori estimates of the resolvent operator associated to the Riemann-Hilbert problem, the authors investigated the perturbation of the cubic NLS with higher order nonlinearity: (1.3) iq_{t}+q_{xx}-2|q|^{2}q-\varepsilon|q|^{\ell}q=0. with \ell>7/2 . The authors show that for \varepsilon\ll 1 whose size depends on the H^{1,1} Sobolev norm of the initial data, solutions to perturbed NLS equation behave as t\to\infty like solutions of the cubic NLS equation ( 1.1 ) which is completely integrable. Their work is the first major step towards building a rigorous perturbation theory for infinite dimensional integrable systems. [ 8 ] relies partly on the L^{p} a priori estimate developed in [ 7 ] . As a byproduct, Deift and Zhou calculated the long time asymptotics for Equation ( 1.1 ) whose initial data belongs to the H^{1,1} space. Later, in [ 10 ] (see also a published version of this paper [ 11 ] ), Dieng and McLaughlin re-investigated the long time asymptotics of ( 1.1 ) and developed the \overline{\partial} version of the nonlinear steepest descent method which is originally developed in [ 6 ] . We also mention that in [ 2 ] , the authors investigate the evolution of scattering data under perturbations of the Toda lattice. In this paper, we revisit the perturbative method of [ 8 ] and give new proofs of L^{p} a priori estimates and also L^{\infty} bounds aiming at a further step towards understanding perturbation of focussing problems and the applicability of [ 8 ] to other integrable models. To illustrate applications of improved estimates by our new approach, we consider the long time dynamics for the following 1-d defocussing NLS with a localized nonlinearity: (1.4) \displaystyle iq_{t}+q_{xx}-2|q|^{2}q-\varepsilon a(x)|q|^{l}q=0, \displaystyle q(x,t=0)=q_{0}(x)\in H^{1,1}(\mathbb{R}). Here 0<\varepsilon\ll 1 , l>3 and a(x) is a Schwarz function (We do not aim to pursue the optimal conditions for a(x) here). If we replace the RHS of ( 1.2 ) with \varepsilon G(x) where (1.5) G(x)=-ia(x)|q|^{l}\left(\begin{array}[]{cc}0&q\\ -\bar{q}&0\end{array}\right) then it is easy to check that (1.6) \left[\partial_{x}-U,\partial_{t}-W\right]=\varepsilon G(x) is the compatibility condition for ( 1.4 ). In the following, we will set up problems and introduce notations based on the concrete example ( 1.4 ). 1.2. The inverse scattering for the NLS Recall that ( 1.1 ) generates an isospectral flow for the problem (1.7) \frac{d}{dx}\Psi=iz\sigma\Psi+Q(x)\Psi where \sigma=\left(\begin{array}[]{cc}{\frac{1}{2}}&0\\ 0&{-\frac{1}{2}}\end{array}\right),\,\,\,Q(x)=\left(\begin{array}[]{cc}0&{q(x)}\\ {\overline{q(x)}}&0\end{array}\right). This is a standard AKNS system. If u\in L^{1}(\mathbb{R}) , Equation ( 1.7 ) admits bounded solutions for z\in\mathbb{R} . There exist unique solutions \Psi^{\pm} of ( 1.7 ) obeying the the following space asymptotic conditions \lim_{x\to\pm\infty}\Psi^{\pm}(x,z)e^{-ixz\sigma_{3}}=\left(\begin{array}[]{cc}{1}&0\\ 0&{1}\end{array}\right), and there is a matrix S(z) , the scattering matrix, with \Psi^{+}(x,z)=\Psi^{-}(x,z)S(z) . The matrix T(z) takes the form (1.8) S(z)=\left(\begin{array}[]{cc}{a(z)}&{\breve{b}(z)}\\[10.0pt] {b(z)}&{\breve{a}(z)}\end{array}\right) and the determinant relation gives a(z)\breve{a}(z)-b(z)\breve{b}(z)=1 Combining this with the symmetry relations (1.9) \displaystyle\breve{a}(z)=\overline{a(\overline{z})},\quad\breve{b}(z)=\overline{b(\overline{z})}. we arrive at |a(z)|^{2}-|b(z)|^{2}=1 and conclude that a(z) is zero-free. By the standard inverse scattering theory, we formulate the reflection coefficient: (1.10) r(z)=\breve{b}(z)/a(z),\quad z\in\mathbb{R}. In [ 14 ] , it is shown that for k,j integers with k\geq 0 , j\geq 1 , the direct scattering map \mathcal{R} maps H^{k,j}(\mathbb{R}) onto H^{j,k}_{1}=H^{j,k}(\mathbb{R})\cap\{r:\left\|r\right\|_{L^{\infty}}<1\} where H^{j,k} norm is given by: (1.11) \left\|q\right\|_{H^{i,j}(\mathbb{R})}=\left(\left\|(1+|x|^{j})q\right\|_{2}^{2}+\left\|q^{(i)}\right\|_{2}^{2}\right)^{1/2}. and the map \mathcal{R}:q\mapsto r is Lipschitz continuous. Since we are dealing with the defocussing NLS, only the reflection coefficient r is needed for the reconstruction of the solution. The long-time behavior of the solution to the NLS equation is obtained through a sequence of transformations of the following RHP: Problem 1.1 . Given r(z)\in H^{1,1}(\mathbb{R}) , find a 2\times 2 matrix-valued function m(z;x,t) on \mathbb{C}\setminus\mathbb{R} with the following properties: (1) m(z;x,t)\to I as |z|\to\infty , (2) m(z;,x,t) is analytic for z\in\mathbb{C}\setminus\mathbb{R} with continuous boundary values m_{\pm}(z;x,t)=\lim_{\varepsilon\to 0}m(z\pm i\varepsilon;x,t), (3) The jump relation m_{+}(z;x,t)=m_{-}(z;x,t)e^{i\theta\ad\sigma}v(z) holds, where (1.12) e^{i\theta\ad\sigma}v(z)=\left(\begin{array}[]{cc}{1-|r(z)|^{2}}&{r(z)e^{i\theta}}\\[10.0pt] {-\overline{r(z)}e^{-i\theta}}&{1}\end{array}\right) and the real phase function \theta is given by (1.13) \theta(z;x,t)=xz-tz^{2} with stationary points (1.14) z_{0}={\dfrac{x}{2t}}. From the solution of Problem 1.1 , we recover (1.15) \displaystyle q(x,t) \displaystyle=\lim_{z\to\infty}2izM_{12}(x,t,z) where the limit is taken in \mathbb{C}\setminus\mathbb{R} along any direction not tangent to \mathbb{R} . 1.2.1. The Beals-Coifman solution For z\in\mathbb{R} , introduce (1.16) w_{\theta}=\left(w^{-}_{\theta},w^{+}_{\theta}\right)=\left(\left(\begin{array}[]{cc}{0}&{r(z)e^{i\theta}}\\[10.0pt] {0}&{0}\end{array}\right),\left(\begin{array}[]{cc}{0}&{0}\\[10.0pt] {-\overline{r(z)}e^{-i\theta}}&{0}\end{array}\right)\right) corresponding to the factorization of the jump matrix: (1.17) v_{\theta}=\left(v^{-}_{\theta}\right)^{-1}v^{+}_{\theta}=\left(\begin{array}[]{cc}{1}&{-r(z)e^{i\theta}}\\[10.0pt] {0}&{1}\end{array}\right)^{-1}\left(\begin{array}[]{cc}{1}&{0}\\[10.0pt] {-\overline{r(z)}e^{-i\theta}}&{1}\end{array}\right). Denote the associated singular integral operator, (1.18) C_{v_{\theta}}h=C^{+}\left(hw^{-}_{\theta}\right)+C^{-}\left(hw^{+}_{\theta}\right) and let \mu\in I+L^{2}(\mathbb{R}) solves the equation (1.19) \left(1-C_{v_{\theta}}\right)^{-1}\mu=I. Then a simple calculation shows that (1.20) m_{\pm}=I+C^{\pm}\left(\mu\left(w^{+}_{\theta}+w^{-}_{\theta}\right)\right)=\mu v^{\pm}_{\theta} is the unique solution to Problem 1.1 . Set (1.21) \mathbf{Q}(x,t)=\int_{\mathbb{R}}\mu(x,s)\left(w_{\theta}^{+}(s)+w_{\theta}^{-}(s)\right)ds. Then direct calculations give: (1.22) \left(\begin{array}[]{cc}{0}&{q(x,t)}\\[10.0pt] {\overline{q(x,t)}}&{0}\end{array}\right)=\frac{1}{2\pi}\mathrm{ad}\sigma(\mathbf{Q}). We now state the important result from [ 9 ] : Theorem 1.2 . Given q_{0} the initial data of Equation ( 1.1 ), then we have the direct scattering map (1.23) \mathcal{R}:H^{1,1}(\mathbb{R})\ni q_{0}\mapsto r(z)\in H^{1,1}_{1}(\mathbb{R}) and inverse scattering map (1.24) \mathcal{R}^{-1}:H^{1,1}_{1}(\mathbb{R})\ni e^{i\theta}r(z)\mapsto q(x,t)\in H^{1,1}(\mathbb{R}). Both \mathcal{R} and \mathcal{R}^{-1} are Lipschitz continuous. Moreover, \mathcal{R}(q(t)) evolves linearly (1.25) \mathcal{R}(q(t))=\mathcal{R}(q(0))e^{-itz^{2}}=r(z)e^{-itz^{2}}. 1.3. The inverse scattering for the perturbed NLS Let q(x,t) be the solution to Equation ( 1.4 ). We then let (1.26) Q(x,t)=\left(\begin{array}[]{cc}{0}&{q(x,t)}\\[10.0pt] {\overline{q(x,t)}}&{0}\end{array}\right) be the potential of ( 1.7 ). Following the standard direct and inverse scattering formalism, we can as well formulate the following Riemann-Hilbert problem: Problem 1.3 . Find a 2\times 2 matrix-valued function m(z;x,t) on \mathbb{C}\setminus\mathbb{R} with the following properties: (1) m(z;x,t)\to I as |z|\to\infty , (2) m(z;,x,t) is analytic for z\in\mathbb{C}\setminus\mathbb{R} with continuous boundary values m_{\pm}(z;x,t)=\lim_{\varepsilon\to 0}m(z\pm i\varepsilon;x,t), (3) The jump relation m_{+}(z;x,t)=m_{-}(z;x,t)e^{ix\ad\sigma}v(z;q(t)) holds, where (1.27) e^{ix\ad\sigma}v(z;q(t))=\left(\begin{array}[]{cc}{1-|r(z;q(t))|^{2}}&{r(z;q(t))e^{ix}}\\[10.0pt] {-\overline{r(z;q(t))}e^{-ix}}&{1}\end{array}\right). Define (1.28) r(t)(z)=e^{itz^{2}}r(z;q(t)) and the following integral equation is originally obtained in [ 12 ] and [ 13 ] : (1.29) r(t)(z)=r_{0}(z)+\varepsilon\int_{0}^{t}dse^{iz^{2}s}\int_{-\infty}^{\infty}dye^{-iyz}\left(m_{-}^{-1}(z;y,q(s))G(q(y,s))m_{-}(z;y,q(s))\right)_{12} where (1.30) r_{0}(z)=\mathcal{R}\left(q_{0}\right)(z),\quad q_{0}=q(t=0). 1.4. Main results Now we introduce our main results for the asymptotics of ( 1.4 ). Due to their technicality, we postpone the presentation of our improved estimates in Section 2 and Section 3 . Theorem 1.4 . Assume q_{0}=q(x,0)\in H^{1,1} with \left\|r_{0}\right\|_{H^{1,1}}=\eta and \left\|r_{0}\right\|_{L^{\infty}}=\rho<1 . There exists \varepsilon(\rho,\eta)\ll 1 such that q(x,t)\in C\left([0,\infty),H^{1,1}\right) solves ( 1.4 ) with \varepsilon=\varepsilon(\rho,\eta) and