Restriction and decoupling estimates for the hyperbolic paraboloid in ℝ 3 \mathbb{R}^{3}

Ciprian Demeter, Shukun Wu · 2025

Read Restriction and decoupling estimates for the hyperbolic paraboloid in ℝ 3 \mathbb{R}^{3} by Ciprian Demeter, 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-2505-09037.txt.

Introduction

Ciprian Demeter and Shukun Wu present a 2025 math paper on waves tied to a saddle-shaped surface. When waves are forced to live on a curved sheet, they spread in ways that are hard to bound. The saddle shape, called a hyperbolic paraboloid, has been less well understood than the bowl-shaped case.

The pair proves a two-part bound, known as a decoupling estimate, for a cut-off piece of the saddle. They also prove a sharper form of the same bound. As a payoff, they settle a linked bound, the restriction estimate, for a range of powers above twenty-two sevenths. That range matches what was known for the bowl shape, closing a gap between the two cases.

Catalog record

CREATORCiprian Demeter, Shukun Wu
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE88 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2505-09037

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

Restriction and decoupling estimates for the hyperbolic paraboloid in ℝ 3 \mathbb{R}^{3} Thanks: The first author is partially supported by the NSF grants DMS-2055156 and DMS-2349828. Ciprian Demeter, Shukun Wu Abstract . We prove bilinear ℓ 2 \ell^{2} -decoupling and refined bilinear decoupling inequalities for the truncated hyperbolic paraboloid in ℝ 3 \mathbb{R}^{3} . As an application, we prove the associated restriction estimate in the range p > 22 / 7 p>22/7 , matching an earlier result for the elliptic paraboloid. 1. Introduction 1.1. Overview Let S\subset\mathbb{R}^{n} be a smooth compact hypersurface and let d\sigma_{S} be its surface measure. We consider the associated extension operator \widehat{fd\sigma_{S}}(x)=\int e^{2\pi ix\cdot\xi}f(\xi)d\sigma_{S}(\xi). Elias Stein conjectured the following. Conjecture 1.1 . When S has non-vanishing Gaussian curvature, (1.1) \|\widehat{fd\sigma_{S}}\|_{p}\lesssim\|f\|_{L^{p}(d\sigma_{S})} holds for all p>\frac{2n}{n-1} and all smooth functions f on S . Since Bourgain’s work [ Bou91 ] , Conjecture 1.1 was studied intensively. Most recently, [ WW24 ] posted an incidence geometry conjecture that, along with decoupling theorems, would fully solve Conjecture 1.1 when S is of elliptic type. One notable property of elliptic surfaces is that they do not contain any linear subspaces, which are typical sources of constructive interference. For example, the \ell^{2} -decoupling theorem in [ BD15 ] is known to fail when S is not of elliptic type. See [ BD17 ] . However, the existence of linear subspaces does not invalidate the L^{p} -estimate ( 1.1 ), since a surface with non-vanishing Gaussian curvature cannot contain linear subspaces of large dimension. Moreover, it is conceivable that linear subspaces are the only obstruction to orthogonality results such as the decoupling theorem. In other words, if constructive interference from linear subspaces is neutralized, then an appropriate form of \ell^{2} -decoupling may still hold. In this paper we prove decoupling inequalities that support the aforementioned philosophy. Specifically, in the setting governed by transversality (Definition 1.2 ), we establish both a bilinear \ell^{2} -decoupling and a bilinear refined decoupling inequality for functions whose Fourier transforms are supported near the hyperbolic paraboloid (1.2) \mathbb{H}=\{(\xi,\eta,\xi\eta):\;(\xi,\eta)\in\mathbb{R}^{2}\}. As an application, we prove Conjecture ( 1.1 ) when S=\mathbb{H}\cap[-1,1]^{3} , conditional to the two-ends Furstenberg conjecture introduced in [ WW24 ] . We also give an unconditional proof of ( 1.1 ) for p>22/7 , matching the best-known result in [ WW24 ] for elliptic surfaces in \mathbb{R}^{3} . Let us now briefly describe our results. 1.2. Bilinear decoupling inequalities We first introduce some notation and a critical definition. For a rectangle \tau\subset[-1,1]^{2} , define \mathbb{H}_{\tau}=\{(\xi,\eta,\xi\eta):(\xi,\eta)\in\tau\} . For a function of three variables f:\mathbb{R}^{3}\to\mathbb{C} , we write f_{\tau} for the Fourier restriction of f to \tau\times\mathbb{R} . We write \mathcal{P}_{\Delta}(\tau) for the collection of \Delta -squares in some partition of \tau . Definition 1.2 . We call a pair of squares (of arbitrary size) \tau_{1},\tau_{2}\subset[-1,1]^{2} transverse if {\rm dist}(\xi_{1},\xi_{2})\sim 1 and {\rm dist}(\eta_{1},\eta_{2})\sim 1 for each (\xi_{j},\eta_{j})\in\tau_{j} . Transversality is in fact equivalent to asking that {\rm dist}(\tau_{1},\tau_{2})\sim 1 and also that each line joining some (\xi_{1},\eta_{1})\in\tau_{1} and (\xi_{2},\eta_{2})\in\tau_{2} has slope with absolute value satisfying \left|\frac{\eta_{2}-\eta_{1}}{\xi_{2}-\xi_{1}}\right|\sim 1. In particular, the line \ell(\tau_{1},\tau_{2}) joining the centers of such squares is (quantitatively) transverse to both coordinate axes. All lines contained in \mathbb{H} are parallel to either the plane \xi=0 or the plane \eta=0 . Transversality guarantees that none of these lines intersects both \mathbb{H}_{\tau_{1}} and \mathbb{H}_{\tau_{2}} . 1.2.1. \ell^{2} -decoupling Definition 1.3 (Bilinear decoupling constant for \ell^{2} -decoupling) . Given 0<\delta<1 and R\geq 1 , we let C(\delta,R) be the smallest constant such that (1.3) \int_{\mathbb{R}^{3}}|f_{1}f_{2}|^{2}\leq C(\delta,R)\prod_{j=1}^{2}\Big(\sum_{\theta_{j}\in\mathcal{P}_{R^{-1/2}}(\tau_{j})}\big\|f_{\theta_{j}}\big\|_{L^{4}(\mathbb{R}^{3})}^{2}\Big) for each transverse \delta -squares \tau_{1},\tau_{2} and each f_{j} Fourier supported on the 1/R -neighborhood N_{1/R}(\mathbb{H}_{\tau_{j}}) of \mathbb{H}_{\tau_{j}} . Remark 1.4 . It is clear that C(\delta,R) is nondecreasing in \delta , and, at least heuristically, it is also nondecreasing in R . Remark 1.5 . Due to the Fourier support of f_{1} and f_{2} , ( 1.3 ) implies (in fact, it is equivalent to) a localized version of itself, with \mathbb{R}^{3} replaced on both sides by (some smooth approximations of) 1_{B_{R}} . Here is our first result. Theorem 1.6 (Bilinear \ell^{2} decoupling) . For all \varepsilon>0 , we have C(1,R)\lesssim_{\varepsilon}R^{\varepsilon} . The proof of Theorem 1.6 is inspired by the alternative proof of the elliptic \ell^{2} decoupling theorem given in [ FSWW18 ] . The key new observation in the non-elliptic setting here is the following. Let \tau_{1} and \tau_{2} be two transverse \delta -squares in [-1,1]^{2} , and for j=1,2 , let S_{\tau_{j}}=N_{\delta^{2}}(H_{\tau_{j}}) be an approximate \delta\times\delta\times\delta^{2} -box. Then, interpreting S_{\tau_{1}} and S_{\tau_{2}} as \delta^{2} -neighborhoods of two planes \pi_{1},\pi_{2} , the intersection \pi_{1}\cap\pi_{2} is a line whose projection onto the horizontal (\xi,\eta) -plane is transverse to both coordinate directions. We refer the reader to the proof of Proposition 2.2 for details. At the foundation of all our orthogonality arguments lies the following classical inequality. Proposition 1.7 . (Bilinear restriction) Let \gamma_{1},\gamma_{2} be two smooth curves in \mathbb{R}^{2} , such that any two of their normal vectors n_{1},n_{2} are (quantitatively) transverse. Assume f_{j} is Fourier supported in the \Delta -neighborhood N_{\Delta}(\gamma_{j}) of \gamma_{j} . Partition N_{\Delta}(\gamma_{j}) into \Delta -squares s_{j} . Then (1.4) \int_{\mathbb{R}^{2}}|f_{1}f_{2}|^{2}\lesssim\sum_{s_{1},s_{2}}\int_{\mathbb{R}^{2}}|P_{s_{1}}f_{1}P_{s_{2}}f_{2}|^{2}, where P_{s}f is the Fourier restriction of f to s . This inequality may be easily proved using simple geometric arguments that rely critically on the fact that 4=2\times 2 . This type of argument is sometimes referred to as bi-orthogonality . However, the paper [ BCT06 ] revealed that inequality ( 1.4 ) is the two-dimensional manifestation of the more general multilinear restriction phenomenon in \mathbb{R}^{n} , that registers at the critical exponent \frac{2n}{n-1} . With this perspective came a different proof of ( 1.4 ), that presents a severe departure from bi-orthogonality. Our proof of Theorem 1.6 embraces this philosophy, leading to a bi-orthogonality free argument for the bilinear decoupling inequality for the two-dimensional paraboloid (both elliptic and hyperbolic). In the elliptic case, the standard bilinear-to-linear reduction immediately recovers the linear \ell^{2} decoupling proved in [ BD15 ] , without the use of the trilinear restriction theorem from [ BCT06 ] . It remains an interesting open problem to extend our results to higher dimensions. Our bi-orthogonality free argument opens up the possibility for a similar argument in \mathbb{R}^{n} , when n is odd. By this we mean, a proof of d -linear \ell^{2} decoupling in \mathbb{R}^{n} using the d -linear restriction theorem in \mathbb{R}^{d} . This speculation is entertained by the coincidence between the multilinear restriction exponent \frac{2d}{d-1} in \mathbb{R}^{d} and the critical exponent \frac{2(n+1)}{n-1} for \ell^{2} decoupling in \mathbb{R}^{n} , when n=2d-1 . However, while this numerology is consistent in critical places of the argument, there are new difficulties in higher dimensions. These are associated with the more complex broad-narrow reduction, when trying to establish the analog of inequality ( 2.5 ). We mention that the coincidence between the two exponents was recently exploited in [ Oh25 ] , in order to produce a proof of \ell^{p} (rather than \ell^{2} ) decoupling, albeit conditional to the Restriction Conjecture. Remark 1.8 . By a standard broad-narrow argument, Theorem 1.6 recovers the decoupling inequality for \mathbb{H} (Theorem 1.8) from the recent paper [ GMO24 ] . See Section ( 2.1 ). 1.2.2. Refined decoupling The

Read the complete text