Mixed norm estimates for dilated averages over planar curves
Junfeng Li, Zengjian Lou, Haixia Yu
TL;DR
The paper analyzes mixed-norm estimates for dilated planar curve averages, establishing sharp space-time and time-space bounds for the operator along a general plane curve with curvature conditions. By reducing T to a Fourier integral operator with cinematic curvature and decomposing into dyadic pieces, the authors derive precise ranges for (1/p,1/q,1/r) through stationary-phase analysis and interpolation. They prove nontrivial, nearly sharp bounds for a broad class of curves, including the parabola, and identify borderlines where sharpness fails, supported by endpoint counterexamples and decay estimates. The results extend the understanding of L^p-improving phenomena and Strichartz-type estimates for curved averaging operators in two dimensions, with implications for related Fourier integral operators and maximal functions.
Abstract
In this paper, we investigate the mixed norm estimates for the operator $ T $associated with a dilated plane curve $(ut, uγ(t))$, defined by \[ Tf(x, u) := \int_{0}^{1} f(x_1 - ut, x_2 - uγ(t)) \, dt, \] where $ x := (x_1, x_2) $ and $γ$ is a general plane curve satisfying appropriate smoothness and curvature conditions. More precisely, we establish the $ L_x^p(\mathbb{R}^2) \rightarrow L_x^q L_u^r(\mathbb{R}^2 \times [1, 2]) $ (space-time) estimates for $ T $, whenever $(\frac{1}{p},\frac{1}{q})$ satisfy \[ \max\left\{0, \frac{1}{2p} - \frac{1}{2r}, \frac{3}{p} - \frac{r+2}{r}\right\} < \frac{1}{q} \leq \frac{1}{p} < \frac{r+1}{2r} \] and $$1 + (1 + ω)\left(\frac{1}{q} - \frac{1}{p}\right) > 0,$$ where $ r \in [1, \infty] $ and $ ω:= \limsup_{t \rightarrow 0^+} \frac{\ln|γ(t)|}{\ln t} $. These results are sharp, except for certain borderline cases. Additionally, we examine the $ L_x^p(\mathbb{R}^2) \rightarrow L_u^r L_x^q(\mathbb{R}^2 \times [1, 2]) $ (time-space) estimates for $T $, which are especially almost sharp when $p=2$ or $p\in [1, \frac{3}{2}]\cup [4, \infty]$.
