Sharp local smoothing estimates for curve averages

Technical Summary: Sharp Local Smoothing Estimates for Curve Averages

Problem Statement
The paper addresses the problem of establishing sharp local smoothing estimates for averaging operators along nondegenerate curves in $\mathbb{R}^n$ for all dimensions $n \ge 2$. Specifically, the authors consider the dilated averaging operator $A_t f(x) = \int_{\mathbb{R}} f(x - t\gamma_n(s))\chi(s) ds$, where $\gamma_n$ is a smooth nondegenerate curve (e.g., the moment curve $(s, s^2, \dots, s^n)$) and $\chi$ is a smooth bump function. The central goal is to determine the optimal range of $L^p$ Sobolev regularity $\sigma$ such that the local smoothing estimate
$$\left( \int_1^2 \|A_t f\|_{L^p_\sigma(\mathbb{R}^n)}^p dt \right)^{1/p} \lesssim \|f\|_{L^p(\mathbb{R}^n)}$$
holds.

A primary application of these estimates is the $L^p$ boundedness of the helical maximal operator $M_{\gamma_n} f(x) = \sup_{t>0} |A_t f(x)|$. Prior to this work, the sharp range of $p$ for which $M_{\gamma_n}$ is bounded was known only for $n=2$ (Bourgain) and $n=3$ (independently by K. Lee, Oh, and others; and by Beltran, Guo, Hickman, and Seeger). For $n=4$, the problem was open, and for $n \ge 5$, the known bounds were not sharp.

Methodology
The proof relies on a Fourier-analytic approach, decomposing the problem into geometric and analytic components. The main technical innovation is a novel wave envelope estimate adapted to moment curves, which avoids the direct analysis of stationary phase points and the Van der Corput lemma that become intractable in higher dimensions ($n \ge 4$).

The methodology proceeds through the following steps:

1. Fourier-Side Reduction: The averaging operator is expressed as a convolution with a surface measure $\mu_\Gamma$ supported on a two-dimensional cone $\Gamma$ generated by the curve. The authors analyze the Fourier decay of $\mu_\Gamma$ by decomposing the frequency space into anisotropic "planks" (rectangular boxes) aligned with the Frenet frame of the curve. This involves a dyadic decomposition of the frequency region based on the distance to the cone and the geometry of the curve's derivatives.

2. Wave Envelope Estimates: The core analytic tool is a new class of wave envelope estimates for nondegenerate curves and cones. Unlike previous work which focused on $L^4$ estimates or square function estimates, this paper establishes sharp $L^p$ wave envelope estimates for $p > 4$.

   • The authors introduce a geometric framework using "wave envelopes" (tilings of space by dual boxes) to control the oscillatory integrals.
   • A key distinction from prior results (e.g., for the parabola or cone in $\mathbb{R}^3$) is the use of an $\ell^{p/2}$ expression within an $L^{p/2}$ norm, rather than the standard $\ell^2$ square function. This modification allows for a more flexible "high/low" frequency analysis, accommodating the specific geometry of moment curve planks in higher dimensions.

3. High/Low Analysis and Induction: The proof utilizes a high/low method (originating from Bourgain and Guth) combined with a broad/narrow analysis.

   • The problem is reduced to proving estimates for polynomial curves via Taylor expansion.
   • An induction on the dimension $n$ is employed. The estimate for curves in $\mathbb{R}^n$ is deduced from estimates for cones in $\mathbb{R}^n$ and curves in $\mathbb{R}^{n-1}$.
   • The "broad" part of the function is controlled using multilinear restriction estimates, while the "narrow" part is handled by rescaling and applying the induction hypothesis.

4. Kakeya-Type Estimates: To bridge the gap between frequency-localized pieces and wave envelope quantities, the authors establish Kakeya-type estimates for intermediate-scale planks. These estimates control the overlap of the geometric objects (planks) associated with different points on the curve.

Key Contributions and Results

• Theorem 1.1 (Sharp Local Smoothing): The authors prove the sharp local smoothing estimate for curve averages in all dimensions $n \ge 2$. For $2 \le p < \infty$, the estimate holds for $\sigma < \sigma(p, n) := \min\{ \frac{1}{n}, \frac{1}{n}(\frac{1}{2} + \frac{2}{p}), \frac{2}{p} \}$. This range is shown to be optimal by testing standard counterexamples.
• Theorem 1.2 (Helical Maximal Operator): As a corollary, the paper establishes the sharp $L^p$ boundedness of the helical maximal operator $M_{\gamma_n}$ for all $n \ge 2$. The operator is bounded if and only if:
  • $p > n$ for $n = 2, 3, 4$.
  • $p > 2(n-2)$ for $n \ge 5$.
    This resolves the conjecture for $n=4$ and improves upon the best known results for $n \ge 5$ (previously obtained by K. Lee, Oh, and others).
• New Wave Envelope Estimates (Theorem 2.1): The paper introduces sharp $L^p$ wave envelope estimates for nondegenerate curves in $\mathbb{R}^n$ for $4 \le p \le n^2+n-2$. These estimates are distinct from and not implied by previously known square function estimates, offering a new tool for analyzing oscillatory integrals associated with curves.
• Bochner-Riesz Estimates (Theorem 2.2): As a byproduct, the authors derive sharp Bochner-Riesz type estimates for nondegenerate curves in all dimensions.

Significance
The paper claims to solve a long-standing open problem regarding the helical maximal operator in dimension $n=4$ and to improve the state-of-the-art for all higher dimensions. The significance lies in the development of a novel wave envelope framework that successfully generalizes techniques previously limited to low dimensions ($n=2, 3$) or specific curves (parabolas). By avoiding the direct analysis of stationary points in high dimensions and instead utilizing a geometric wave packet decomposition with $\ell^{p/2}$ structures, the authors overcome the algebraic barriers that prevented the extension of the Van der Corput lemma approach to $n \ge 4$. The results provide the definitive range of $L^p$ boundedness for these operators, confirming the conjecture that $M_{\gamma_n}$ is bounded if and only if $p > n$ for $n \le 4$ and $p > 2(n-2)$ for $n \ge 5$.