High-order long-time asymptotics for small solutions to the one-dimensional nonlinear Schrödinger equation

1. Problem Statement

The paper investigates the global well-posedness and the long-time asymptotic behavior of small, localized solutions to the one-dimensional Nonlinear Schrödinger Equation (NLS) with a general gauge-invariant polynomial nonlinearity:
$$i\partial_t u + \frac{1}{2}\Delta u = \sum_{n=1}^d \lambda_n |u|^{2n}u$$
The authors consider initial data $u_1$ at $t=1$ belonging to weighted Sobolev spaces $H^{1,0}_x$ (finite energy) and $H^{0, 2N+1}_x$ (high-order spatial localization), where the data is sufficiently small.

Key Challenges:

• Long-range effects: In 1D, the cubic nonlinearity ($n=1$) is critical. It does not decay fast enough to be treated as a short-range perturbation, leading to "modified scattering" where the solution acquires a logarithmic phase shift.
• Higher-order expansions: While first-order asymptotics are well-understood, deriving rigorous high-order asymptotic expansions (arbitrary order $N$) for general polynomial nonlinearities without relying on complete integrability is a significant open problem.
• Regularity: Previous results often required Schwartz class data or very high regularity. This work aims to work in low-regularity weighted Sobolev spaces.

2. Methodology

The analysis relies heavily on the Space-Time Resonance Method, originally introduced by Germain, Masmoudi, and Shatah, and adapted to 1D cubic NLS in previous works (e.g., [22]). The authors extend this framework to handle arbitrary-order polynomial nonlinearities and high-order expansions.

Core Techniques:

1. Profile Formulation: The solution is transformed into a profile $f(t) = e^{-it\Delta/2}u(t)$, which removes the linear dispersion. The evolution of the profile is governed by an integral equation involving the nonlinearity.
2. Modified Profile: To handle the non-integrable cubic term (long-range effect), a "modified profile" $\hat{w}(t, \xi)$ is introduced:
   $$\hat{w}(t, \xi) = \Theta(t, \xi) \hat{f}(t, \xi), \quad \text{where } \Theta(t, \xi) = \exp\left(i \int_1^t \frac{\lambda_1 |\hat{f}(s, \xi)|^2}{s} ds\right)$$
   This transformation absorbs the leading-order logarithmic phase shift, allowing $\hat{w}$ to converge to a limit as $t \to \infty$.
3. Stationary Phase Expansion: The nonlinear terms in the Fourier space are analyzed using stationary phase approximations. The authors decompose the nonlinearity into a truncated sum (contributing to the asymptotic expansion) and a remainder term (error).
4. Bootstrap Argument: A global existence proof is established via a bootstrap argument. The authors define a norm $\|u\|_{X_T}$ involving weighted Sobolev norms of the profile and the modified profile. They show that for small initial data, the solution remains bounded in this norm for all time, allowing the extension of local existence to global existence.
5. Inductive Construction of Asymptotics: The high-order expansion is constructed inductively. By assuming the expansion holds up to order $n-1$, the authors substitute this into the evolution equation for $\hat{w}$, integrate in time, and derive the coefficients for order $n$. This process carefully tracks the interplay between powers of $t^{-1}$ and $\ln(t)$.

3. Key Contributions

• Arbitrary Order Asymptotics: The paper provides the first rigorous derivation of asymptotic expansions of arbitrary order $N$ for 1D NLS with general polynomial nonlinearities, without assuming complete integrability (unlike inverse scattering methods).
• General Nonlinearities: The results apply to any polynomial nonlinearity $\sum \lambda_n |u|^{2n}u$, not just the cubic case. The analysis explicitly handles the interaction between the cubic term (dominant long-range) and higher-order terms (quintic, septic, etc.).
• Low Regularity: The results are established for initial data in weighted Sobolev spaces $H^{1,0}_x \times H^{0, 2N+1}_x$, relaxing the requirement for Schwartz class data found in earlier high-order studies.
• Explicit Structure of Corrections: The paper details the structure of the asymptotic expansion, showing that terms involve powers of $t^{-p}$ and $\ln(t)^k$ where $k \le 2p$. It explicitly identifies how higher-order nonlinearities (like the quintic term) contribute to the expansion at specific orders.

4. Main Results

Theorem 1.1 (Global Existence and Asymptotics):
For sufficiently small initial data satisfying the weighted Sobolev bounds, there exists a unique global solution $u \in C([1, \infty), H^1_x)$ with sharp decay:
$$\|u(t)\|_{L^\infty_x} \lesssim \frac{C(\epsilon_0)}{(1+t)^{1/2}}$$
Furthermore, the solution admits an asymptotic expansion of order $N$:
$$u(t, x) = \frac{e^{i\frac{x^2}{2t} - i\lambda_1 |u_{0,0}(\frac{x}{t})|^2 \ln(t) - i\phi(\frac{x}{t})}}{(it)^{1/2}} \sum_{p=0}^N \sum_{k=0}^{2p} \frac{\ln(t)^k}{t^p} u_{p,k}\left(\frac{x}{t}\right) + u_{err}(t, x)$$
where:

• $u_{p,k} \in W^{2(N-p), \infty}_x$.
• The error term satisfies $\|u_{err}(t)\|_{L^\infty_x} = O(t^{-N-1/2-\delta})$ for some $\delta > 0$.
• The phase correction includes a logarithmic term driven by the cubic coefficient $\lambda_1$ and a bounded phase shift $\phi$.

Specific Findings on Nonlinearities:

• Cubic Term ($n=1$): Induces the primary long-range effect, manifesting as the $\ln(t)$ phase shift and modifying the scattering profile.
• Quintic Term ($n=2$): Contributes to the expansion at order $t^{-2}$ (and higher), introducing terms proportional to $|u_{0,0}|^4 u_{0,0}$. The authors explicitly compute the second-order expansion in the Appendix, showing how the quintic term interacts with the cubic phase shift.

5. Significance

• Extension of Integrability-Free Methods: This work demonstrates that the space-time resonance method is robust enough to handle high-order asymptotics for general nonlinearities, bridging the gap between integrable systems (solvable via Inverse Scattering) and non-integrable systems.
• Precision in Scattering Theory: By providing expansions to arbitrary order, the paper allows for a much more precise description of the "final state" of the solution. This is crucial for understanding the long-term dynamics of dispersive waves in 1D.
• Foundation for Future Work: The rigorous framework for handling higher-order terms and weighted norms sets a precedent for analyzing more complex systems, such as NLS with potentials or coupled systems, where complete integrability is absent.
• Validation of Physical Intuition: The explicit calculation of the second-order terms (Appendix A) confirms and refines previous heuristic derivations (e.g., from [29]), particularly regarding the vanishing nature of the quintic contribution compared to the logarithmic cubic contribution.

In summary, Jendrej and Salvi have successfully generalized the theory of modified scattering in 1D NLS, providing a rigorous, high-order asymptotic framework for small solutions with polynomial nonlinearities using modern harmonic analysis techniques.