Nonlinear wave equations with slowly decaying initial data

Technical Summary: Nonlinear Wave Equations with Slowly Decaying Initial Data

Problem Setting
This paper investigates the local and global well-posedness of the cubic nonlinear wave equation (NLWE) in two dimensions, specifically addressing the case of "slowly decaying" initial data. The authors consider the equation:
$$\partial_t^2 u - \Delta_x u = \pm |u|^2 u, \quad (t, x) \in \mathbb{R} \times \mathbb{R}^2$$
with initial data $(u(0), \partial_t u(0)) = (f, g)$ belonging to $L^p$-based spaces where $p > 2$. Standard well-posedness theory for such equations typically relies on $L^2$-based Sobolev spaces ($H^s$). However, for $p > 2$, the standard Sobolev spaces $W^{s,p}$ are not invariant under the half-wave group (the linear propagator $e^{it\sqrt{-\Delta}}$), leading to a loss of regularity that complicates the analysis of nonlinear terms. The central challenge is to establish well-posedness for initial data that decays more slowly than those in $L^2$-based spaces, without incurring prohibitive derivative losses.

Methodology
The authors employ a novel framework combining $\ell^2$-decoupling inequalities with adapted function spaces.

1. Adapted Besov Spaces: The authors introduce a new scale of Besov spaces, denoted $B^s_{p,q,r}(\mathbb{R}^d)$, which are specifically adapted to the half-wave group. These spaces are defined via a decomposition involving Fourier multipliers localized to cones in frequency space (related to the geometry of the light cone). Unlike standard Besov spaces $B^s_{p,q,r}$, these adapted spaces satisfy invariance properties under the half-wave propagator, making them natural candidates for the fixed-time regularity of wave equations.
2. $\ell^2$-Decoupling: The core analytical tool is the sharp $\ell^2$-decoupling inequality for the cone, established by Bourgain and Demeter. The authors utilize this to derive local smoothing estimates in the adapted spaces. Specifically, they prove that the half-wave group maps initial data in $B^s_{p,2,2}(\mathbb{R}^d)$ to space-time Lebesgue spaces $L^p_t([0,1], L^p(\mathbb{R}^d))$ with a specific regularity gain.
3. Contraction Mapping: The well-posedness results are obtained via a standard contraction mapping argument (Picard iteration) in a space-time function space $S$. This requires establishing:
   • Linear estimates for the free evolution operator in the adapted spaces.
   • Nonlinear estimates (trilinear for the cubic case) using Strichartz estimates and the local smoothing properties derived from decoupling.
4. Global Well-Posedness: For the defocusing case, global existence is proven by adapting arguments from Dodson, Soifer, and Spencer (originally for NLS). This involves decomposing the solution into a linear part and a nonlinear remainder, then using energy and mass conservation laws (applied to the remainder) combined with a blow-up alternative to prevent finite-time blow-up.

Key Contributions and Results

• New Local Smoothing Estimates: The paper proves Theorem 1.1, establishing that for $d \ge 2$ and $p \in (2, \infty)$, the half-wave group satisfies:
  $$\|e^{it\sqrt{-\Delta}}f\|_{L^p_t([0,1], L^p(\mathbb{R}^d))} \lesssim \|f\|_{B^{s(p)+\varepsilon}_{p,2,2}(\mathbb{R}^d)}$$
  where $s(p)$ is the critical exponent related to the local smoothing conjecture. The authors note that the space $B^{s(p)+\varepsilon}_{p,2,2}$ is strictly larger (in terms of norm size) than the corresponding Hardy space for Fourier integral operators ($H^s_{FIO}$) or standard Besov spaces $B^s_{p,p,p}$, providing a strict improvement over previous bounds.
• Local Well-Posedness for Slowly Decaying Data: Theorem 1.2 establishes local well-posedness for the cubic NLWE in 2D with initial data in spaces such as:
  $$X \times Y = (B^\varepsilon_{4,2,2}(\mathbb{R}^2) + \dot{H}^{3/8}(\mathbb{R}^2)) \times (B^{\varepsilon-1}_{4,2,2}(\mathbb{R}^2) + \dot{H}^{-5/8}(\mathbb{R}^2))$$
  This result improves upon previous results in $W^{s,p}$ spaces, allowing for initial data with lower regularity and slower decay. Remark 1.3 clarifies that while the $L^6$-based result supersedes standard Sobolev results, the $L^4$-based result involves a derivative mismatch of $1/8$, reflecting the fact that $\ell^2$-decoupling does not fully resolve the local smoothing conjecture for all $p$.
• Generalizations: The methods are shown to be robust, extending to:
  • Higher dimensions ($d \ge 3$).
  • Variable coefficients (nonlinear wave equations on compact Riemannian manifolds).
  • Higher-order nonlinearities (e.g., quintic NLWE).
  • Slower decaying initial data in $L^{4n+2}$-based spaces (Theorem 5.7).
• Global Well-Posedness: Theorem 5.8 proves global well-posedness for the defocusing cubic NLWE in 2D with initial data in $B^s_{6,2,2}(\mathbb{R}^2) \times B^{s-1}_{6,2,2}(\mathbb{R}^2)$ for $s > 1/2$. This is achieved without relying on conserved quantities in the $L^p$ scale directly, but rather by controlling the difference between the solution and the linear evolution.

Significance and Claims
The authors claim that this work provides a novel approach to nonlinear wave equations with slowly decaying initial data. By introducing Besov spaces adapted to the half-wave group and leveraging $\ell^2$-decoupling, they achieve:

1. Improved Regularity: They obtain well-posedness results for initial data in spaces that are strictly larger than those accessible via standard $L^p$-based Sobolev spaces or Hardy spaces for Fourier integral operators.
2. Sharpness: The local smoothing estimates derived are shown to be sharp for the class of spaces considered, quantifying the polynomial growth of the norm under the half-wave group.
3. Robustness: The framework is flexible enough to handle variable coefficients and higher dimensions, suggesting a pathway for future analysis of wave equations on manifolds where standard $L^2$-based methods may be insufficient for slowly decaying data.

The paper explicitly notes that while their results complement the local smoothing conjecture, they do not imply it for all $p$, and the specific derivative loss observed in the $L^4$ case highlights the limitations of current decoupling techniques compared to the full conjecture. The work does not propose new experimental applications but rather advances the theoretical understanding of well-posedness in non-standard function spaces.