Weighted L 2 L^{2} estimates with applications to L p L^{p} problems
Shukun Wu · 2025
Read Weighted L 2 L^{2} estimates with applications to L p L^{p} problems by Shukun Wu (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-2506-02650.txt.
Introduction
Shukun Wu, writing alone, proves weighted estimates for a central operator of harmonic analysis and draws out what they buy. This 2025 paper concerns the Fourier extension operator, which builds waves from frequency data on a surface. The estimates are of the weighted kind, bounding the operator's output in a square-integrable sense against a weight, for the plane.
Three applications follow. The first concerns the maximal Schrödinger operator, which tracks the worst-case size of a quantum evolution over time. The second is a maximal extension operator of the same family. The third reaches fractal territory: how a fractal measure's transform decays around circles. It also feeds a conjecture in partial differential equations, an analogue of the Mizohata-Takeuchi conjecture, about solutions concentrating along lines. Wu presents the weighted bounds as the single engine under all of these.
Catalog record
| CREATOR | Shukun Wu |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 53 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2506-02650
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Weighted L 2 L^{2} estimates with applications to L p L^{p} problems Shukun Wu Abstract . We establish some weighted L 2 L^{2} estimates for the Fourier extension operator in ℝ 2 \mathbb{R}^{2} , and discuss several applications to L p L^{p} problems. These include estimates for the maximal Schrödinger operator and the maximal extension operator, decay of circular L p L^{p} -means of Fourier transform of fractal measures, and an L p L^{p} analogue of the Mizohata–Takeuchi conjecture. 1. Introduction The purpose of this paper is to prove certain weighted L^{2} estimates for the Fourier extension operator and to present a couple of applications to problems related to L^{p} estimates. Previously, these problems shared similar obstructions arising from limitations of the bilinear restriction theorem. 1.1. Overview We begin with a gentle introduction to our weighted L^{2} estimate. Let S be a compact curve in \mathbb{R}^{2} with non-zero curvature. Typical examples are the unit circle and the truncated parabola \{(\xi,\xi^{2}):|\xi|\leq 1\} . For this curve S , let \sigma_{S} be the surface measure on S , and we define the extension operator E_{S}:S\to\mathbb{R}^{2} as (1.1) E_{S}f(x)=\int e^{ix\cdot\xi}f(\xi)d\sigma_{S}(\xi). Let X\subset\mathbb{R}^{2} be a set. We are interested in controlling \|E_{S}f\|_{L^{2}(X)} in terms of \|f\|_{2} and the information encoded in the weight X . For instance, when considering L^{p} inequality, X often represents an upper level set, and an estimate of the form (1.2) \|E_{S}f\|_{L^{2}(X)}\lesssim|X|^{\alpha}\cdot\|f\|_{2} for some \alpha>0 would be desirable. This is because, by Hölder’s inequality, such an estimate is essentially equivalent to an (L^{2},L^{p}) estimate (1.3) \|E_{S}f\|_{p}\lesssim\|f\|_{2}, where p is chosen so that 1/2-1/p=\alpha . In other words, questions about L^{p} estimates can often be reduced to the study of weighted L^{2} estimates. The classical Stein-Tomas estimate shows that ( 1.2 ) can only be true if \alpha\geq 1/3 . Via the bilinear restriction theorem, this lower bound can be improved to \alpha\geq 1/4 if E_{S} is replaced by its bilinear analog \text{Bil}E_{S} . That is, we have (1.4) \|{\rm Bil}E_{S}f\|_{L^{2}(X)}\lesssim|X|^{1/4}\cdot\|f\|_{2}, which we also refer to as the bilinear restriction estimate. The bilinear restriction theorem has proven useful in various problems. However, when it comes to problems related to weighted L^{p} estimates (e.g., Proposition 3.2 ), it is often inefficient. This is because ( 1.4 ) is sharp only when X is a union of r -balls for some r much larger than 1 , exhibiting a full-dimensional structure (since each r -ball has full dimensions), whereas in weighted L^{p} problems, the weight is usually a union of unit balls obeying certain fractal conditions. In this paper, we prove a weighted L^{2} estimate that improves ( 1.4 ) when X is a 1-dimensional fractal set, and when the bilinear operator \text{Bil}E_{S} is replaced by a slightly smaller operator, the “broad” operator {\rm{Br}}_{A}E_{S} . Weighted L^{2} estimates involving {\rm{Br}}_{A}E_{S} were investigated in [ LW24 ] as a tool for the study of the convergence of the Bochner-Riesz means. The broad operator {\rm{Br}}_{A}E_{S} was introduced by Guth in [ Gut16 ] . Below we recall its definition and a notion for fractal sets. Definition 1.1 . Let K\geq 1 and A\in\mathbb{N} . Let \Sigma=\{\sigma\} be a set of finite-overlapping K^{-1} -arcs that covers S . Let f_{\sigma}=f{\bf 1}_{\sigma} . For any f:S\to\mathbb{C} and x\in\mathbb{R}^{2} , define (1.5) {\rm{Br}}_{A}E_{S}f(x):=\max_{\begin{subarray}{c}\Sigma^{\prime}\subset\Sigma,\\ \#\Sigma^{\prime}=A\end{subarray}}\min_{\sigma\in\Sigma^{\prime}}|E_{S}f_{\sigma}(x)|. Definition 1.2 (Katz-Tao (\delta,s,C) -set) . Let \delta\in(0,1) be a small number. For s\in(0,n] , a finite set E\subset\mathbb{R}^{n} is called a Katz-Tao (\delta,s,C) -set (or simply a Katz-Tao (\delta,s) -set if C is not important in the context) if (1.6) \#(E\cap B(x,r))\leq C(r/\delta)^{s},\hskip 8.5359pt\forall x\in\mathbb{R}^{n},\,r\in[\delta,1]. A similar definition applies when E is a union of \delta -balls. Now we can state our main theorem. Theorem 1.3 . Let R\geq 1 , \varepsilon>0 , and K\leq R^{\varepsilon^{4}} . Let X\subset B_{R} be a union of unit balls such that the R^{-1} -dilate of X is a Katz-Tao (R^{-1},1) -set. Let \varepsilon^{\prime}\in(0,\log K/\log R) . Then when A\geq R^{\varepsilon^{\prime}} , we have (1.7) \|{\rm{Br}}_{A}E_{S}f\|_{L^{2}(X)}\leq C_{\varepsilon,\varepsilon^{\prime}}R^{\varepsilon}|X|^{2/9}\cdot\|f\|_{2}. The proof of Theorem 1.3 uses the two-ends Furstenberg estimate established in [ WW24 ] . It also uses induction on scales and the refined decoupling theorem. It is appealing to determine the smallest \alpha>0 such that (1.8) \|{\rm{Br}}_{A}E_{S}f\|_{L^{2}(X)}\lessapprox|X|^{\alpha}\cdot\|f\|_{2} holds for all weights X whose R^{-1} -dilate is a Katz-Tao (R^{-1},1) -set. The following example demonstrates that \alpha cannot be smaller than 1/6 . Let S=\{(\xi,\xi^{2}):\xi\in[-1,1]\} be the truncated parabola. Assume that R^{1/2} is an integer. Let f be the characteristic function of the (100R)^{-1} -neighborhood of R^{-1/2}\mathbb{Z}\cap[-1,1] . By the uncertainty principle, when (x_{1},x_{2})\in B_{R} , we have (1.9) |E_{S}f(x_{1},x_{2})|\sim R^{-1}\Big|\sum_{k=1}^{R^{1/2}}e\big(x_{1}\frac{k}{R^{1/2}}+x_{2}\frac{k^{2}}{R}\big)\Big|, Let X be the 10^{-2} -neighborhood of the set (1.10) \{(x_{1},x_{2})\in B_{R}:x_{1}=R^{1/2}(\frac{a}{q}+\mathbb{Z}),\,x_{2}=R\frac{b}{q},\,\text{ $a,b,q\sim R^{1/6}$ are integers}\}, so |X|\sim R . Also, by evaluating the Gauss sum in ( 1.9 ), we have {\rm{Br}}_{A}E_{S}f(x)\sim|E_{S}f(x)|\gtrsim R^{-7/12} for all x\in X . Thus, \|{\rm{Br}}_{A}E_{S}f\|_{L^{2}(X)}\sim R^{-7/12}R^{1/2}\sim|X|^{1/6}\|f\|_{2} , which shows \alpha\geq 1/6 . 1.2. Applications We give three applications to our weighted L^{2} estimate ( 1.7 ), with the help of the standard broad-narrow analysis. 1.2.1. L^{p} circular decay of Fourier transform of fractal measures Let \mu be a probability Frostman measure supported in [0,1]^{2} such that \mu(B_{r})\lesssim r for all B_{r}\subset[0,1]^{2} . Let \sigma be the surface measure on the unit circle \mathbb{S}^{1} . Wolff [ Wol99 ] studied the decay of the circular L^{2} -means of \widehat{\mu} and prove that, for any such 1 -dimensional probability Frostman measure, the following is true for all \varepsilon>0 and R\geq 1 : (1.11) \Big(\int_{\mathbb{S}^{1}}|\widehat{\mu}(R\xi)|^{2}d\sigma(\xi)\Big)^{1/2}\leq C_{\varepsilon}R^{\varepsilon}R^{-1/4} In the same paper, he posted a similar question regarding the circular L^{p} -means (see also the discussion in [ Gut25 , Section 10.2] ), which can be formulated using the following definition. Definition 1.4 . Define \sigma_{p}(1) be the supremum of all \sigma such that for all probability Frostman measure \mu supported in [0,1]^{2} with \mu(B_{r})\lesssim r for all B_{r}\subset[0,1]^{2} , it holds (1.12) \Big(\int_{\mathbb{S}^{1}}|\widehat{\mu}(R\xi)|^{p}d\sigma(\xi)\Big)^{1/p}\leq C_{p}R^{-\sigma}. Wolff’s question is about identifying \sigma_{p}(1) . In particular, ( 1.11 ) is equivalent to \sigma_{2}(1)\geq 1/4 . As an application of Theorem 1.3 , we prove the following result, generalizing Wolff’s estimate ( 1.11 ) to circular L^{p} -means. Theorem 1.5 . Let \mu be a probability Frostman measure supported in [0,1]^{2} such that \mu(B_{r})\lesssim r for all B_{r}\subset[0,1]^{2} . Then \sigma_{p}(1)\geq 1/2p for all p\in[9/5,2] . The lower bound \sigma\geq 1/2p is sharp in ( 1.12 ) for fixed R . That is, for fixed R , there exists a measure \mu such that the L.H.S. of ( 1.12 ) is \gtrsim R^{-1/2p} . In fact, we can just take d\mu(x)=Re^{iRx_{2}}\varphi(x_{1})\varphi(Rx_{2})dx_{1}dx_{2} , where \varphi is a bump function on [-1,1] . In this case, \widehat{\mu} is essentially supported on a horizontal R\times 1 -rectangle with height R , so \int_{\mathbb{S}^{1}}|\widehat{\mu}(R\xi)|^{p}d\sigma(\xi)\sim R^{-1/2} , yielding (\int_{\mathbb{S}^{1}}|\widehat{\mu}(R\xi)|^{p}d\sigma(\xi))^{1/p}\gtrsim R^{-1/2p} . Wolff’s estimate ( 1.11 ) is connected to Falconer distance set problem. For this connection and higher dimensional generalizations of ( 1.11 ), we refer to [ DZ19 ] . 1.2.2. Estimates for the maximal Schrödinger operator and the maximal extension operator Let I be a compact interval, and let (1.13) e^{it\Delta}f(x)=\int_{\mathbb{R}^{n}}e^{i(x\xi+t\xi^{2})}\widehat{f}(\xi)d\xi be the solution of the Schrödinger equation on \mathbb{R}^{n}\times I (1.14) iu_{t}=\Delta u,\hskip 14.22636ptu(\cdot,0)=f. We are interested in the maximal Schrödinger operator