Restriction problems on the three-dimensional Heisenberg nilmanifold

Hajer Bahouri and Véronique Fischer · 2026

Read Restriction problems on the three-dimensional Heisenberg nilmanifold by Hajer Bahouri and Véronique Fischer (2026) 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-2605-29781.txt.

Introduction

Hajer Bahouri and Veronique Fischer prove a restriction theorem for spectra on the three-dimensional Heisenberg nilmanifold. Since the manifold forms a circle bundle over the flat torus, the result gives a torus-based analogue of Zygmund's restriction theorem, in sub-elliptic form. The team also establishes sharpness through the discrete short-time Fourier transform, so the bound cannot be tightened in this form. The short paper thus extends a classical restriction result into a curved, sub-elliptic setting.

Catalog record

CREATORHajer Bahouri and Véronique Fischer
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE94261
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2605.29781

Browse the full library · About the Tower of Babel

Full text

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

Restriction problems on the three-dimensional Heisenberg nilmanifold Hajer Bahouri and Véronique Fischer Abstract . In this paper, we prove a spectral restriction theorem on the three-dimensional Heisenberg nilmanifold. Since this manifold is an 𝕊 1 \mathbb{S}^{1} -bundle over the flat torus 𝕋 2 \mathbb{T}^{2} , the result provides a sub-elliptic counterpart of Zygmund’s restriction theorem on 𝕋 2 \mathbb{T}^{2} [ 70 ] . We also establish its sharpness by means of the discrete short-time Fourier transform. 2020 Mathematics Subject Classification 43A80, 58J50, 35H20, 42B10, 43A30 1. Introduction 1.1. Restriction problems on the torus We start with the classical inequality on the two-dimensional torus \mathbb{T}^{2} : (1.1) \Big\|\sum_{|\omega|=r}c_{\omega}\chi_{\omega}\Big\|_{L^{4}(\mathbb{T}^{2})}\leq 5^{\frac{1}{4}}\|(c_{\omega})_{\omega\in\mathbb{Z}^{2}}\|_{\ell^{2}(\mathbb{Z}^{2})}, valid for any r>0 and (c_{\omega})_{\omega\in\mathbb{Z}^{2}}\in\ell^{2}(\mathbb{Z}^{2}) , and where we have denoted for any \omega\in\mathbb{Z}^{2} , |\omega|:=\sqrt{\omega_{1}^{2}+\omega_{2}^{2}}\qquad\mbox{and}\qquad\chi_{\omega}(x,y):=e^{2\pi i(\omega_{1}x+\omega_{2}y)},\ (x,y)\in\mathbb{T}^{2}. This estimate, or rather its dual, was proved by Zygmund [ 70 ] in the 1970’s. Zygmund’s result may be summarised as (1.2) \|\Pi_{r,\sqrt{\Delta}}\|_{\mathscr{L}(L^{2}(\mathbb{T}^{2}),L^{4}(\mathbb{T}^{2}))}=\|\Pi_{r,\sqrt{\Delta}}\|_{\mathscr{L}(L^{4/3}(\mathbb{T}^{2}),L^{2}(\mathbb{T}^{2}))}\leq 5^{1/4}. Above, the operator \Pi_{r,\sqrt{\Delta}} is given by (1.3) \Pi_{r,\sqrt{\Delta}}f=\sum_{|\omega|=r}(f,\chi_{\omega})_{L^{2}(\mathbb{T}^{2})}\chi_{\omega},\qquad f\in L^{2}(\mathbb{T}^{2}). This is the spectral projector onto the r -eigenspace for \sqrt{\Delta} where \Delta=-\partial_{1}^{2}-\partial_{2}^{2} denotes the standard Laplacian on the two-dimensional torus \mathbb{T}^{2} . This result is specific to the two-dimensional case: as we will see in Section 1.3.2 , its proof relies mainly on an underlying simple arithmetic structure. This is no longer the case in higher dimensions: for \Pi_{r,\sqrt{\Delta}} , with \Delta=-\partial_{1}^{2}-\cdots-\partial_{n}^{2} on the n -dimensional torus \mathbb{T}^{n} , it has been established by Bourgain [ 11 ] and Bourgain-Demeter [ 12 ] that, for all n\geq 4 , 1\leq p\leq\frac{2n}{n+3} and \epsilon>0 , there holds (1.4) \|\Pi_{r,\sqrt{\Delta}}\|_{\mathscr{L}(L^{p}(\mathbb{T}^{n}),L^{2}(\mathbb{T}^{n}))}\lesssim_{\epsilon}r^{n(\frac{1}{p}-\frac{1}{2})-1+\epsilon}\,, while for n=3 , 1\leq p\leq\frac{3}{2} and \epsilon>0 , Bourgain proved in [ 13 ] the following estimate (1.5) \|\Pi_{r,\sqrt{\Delta}}\|_{\mathscr{L}(L^{p}(\mathbb{T}^{3}),L^{2}(\mathbb{T}^{3}))}\lesssim_{\epsilon}r^{\epsilon}\,. Such problems are known as discrete restriction problems. Given the wide range of their applications, they have been extensively studied in recent decades and we discuss a brief historical overview in Section 1.3 below. For the latest developments on this topic, we refer the interested reader to the surveys by Guth [ 39 ] and Demeter [ 19 , 20 ] . 1.2. Setting of the problem and statement of the main results The aim of this paper is to show a discrete restriction problem for the three-dimensional Heisenberg nilmanifold. Here, the 3-dimensional Heisenberg group \mathbb{H}_{1} is realised as the set of 3\times 3 -matrices with real entries, upper triangular and unipotent (that is, with entries equal to 1 on the diagonal): (1.6) \left(\begin{array}[]{ccc}1&a&c\\ 0&1&b\\ 0&0&1\end{array}\right),\qquad a,b,c\in\mathbb{R}. We consider its subgroup \Gamma consisting of the elements above corresponding to integer entries a,b,c\in\mathbb{Z} . The standard three-dimensional Heisenberg nilmanifold is the resulting right quotient: M:=\Gamma\backslash\mathbb{H}_{1}. It is a compact smooth manifold of dimension 3 that may be described as an \mathbb{S}^{1} -bundle over the flat torus \mathbb{T}^{2} . A fundamental fact is that M is equipped with a measure, a sub-Riemannian structure, and a canonical sub-Laplacian \mathcal{L}_{M} ; this will be described in Section 2 . The spectral decomposition of \mathcal{L}_{M} is well-known from a spectral viewpoint [ 21 ] and in harmonic analysis [ 47 , 65 ] ; it will be recalled in Section 3.1 . In this paper, we will prove the following result: Theorem 1.1 . Considering the canonical sub-Laplacian \mathcal{L}_{M} on M , we denote by \Pi_{\mu,\sqrt{\mathcal{L}_{M}}} the spectral projector of \sqrt{\mathcal{L}_{M}} . We have for all \mu>0 and \epsilon>0 (1.7) \|\Pi_{\mu,\sqrt{\mathcal{L}_{M}}}\|_{\mathscr{L}(L^{2}(M),L^{4}(M))}=\|\Pi_{\mu,\sqrt{\mathcal{L}_{M}}}\|_{\mathscr{L}(L^{4/3}(M),L^{2}(M))}\leq C_{\epsilon}\mu^{\frac{1}{2}+\epsilon}. To our knowledge, Theorem 1.1 is the first restriction result in the challenging context of nilmanifolds, which is both spectrally discrete and sub-elliptic. The estimate ( 1.7 ) is sharp up to the factor \mu^{\epsilon} . More precisely, we will also show the following proposition: Proposition 1.2 . The exponent 1/2 in Theorem 1.1 is optimal in the sense that there exists C>0 such that we have for all \mu>0 in the spectrum of \sqrt{\mathcal{L}_{M}} \|\Pi_{\mu,\sqrt{\mathcal{L}_{M}}}\|_{\mathscr{L}(L^{2}(M),L^{4}(M))}=\|\Pi_{\mu,\sqrt{\mathcal{L}_{M}}}\|_{\mathscr{L}(L^{4/3}(M),L^{2}(M))}\geq C\mu^{\frac{1}{2}}. Compared to Zygmund’s result ( 1.2 ), the factor \mu^{\frac{1}{2}} can be interpreted as a fractional loss of derivatives. Here, the notion of derivative is associated to the nilmanifold M and its canonical sub-Riemannian structure; it can also be defined in terms of \sqrt{\mathcal{L}_{M}} and its functional analysis. This concept goes back to the seventies with the founding papers of Hörmander [ 42 ] and Rothschild-Stein [ 57 ] . However, it is not clear whether the factor \mu^{\epsilon} in ( 1.7 ) is unavoidable: it occurs often in discrete frameworks (see e.g. ( 1.4 ) and ( 1.5 )), and arises with questions related to number theory. This is also the case in our setting. Theorem 1.1 improves significantly the trivial estimate given by the so-called Bernstein inequality in this context: Proposition 1.3 (Sub-elliptic Bernstein Inequality) . With the notation of Theorem 1.1 , there exists a constant C>0 such that the following estimate holds for any \mu>0 : (1.8) \|\Pi_{\mu,\sqrt{\mathcal{L}_{M}}}\|_{\mathscr{L}(L^{2}(M),L^{4}(M))}=\|\Pi_{\mu,\sqrt{\mathcal{L}_{M}}}\|_{\mathscr{L}(L^{4/3}(M),L^{2}(M))}\leq C\mu. The proof of the sub-elliptic Bernstein inequality will be given in greater generality in Theorem C.3 , see also Remark C.4 for the case of the Heisenberg nilmanifold. It should be emphasised that the sub-elliptic Bernstein inequality is governed not by the topological dimension of M , but by the homogeneous dimension of \mathbb{H}_{1} , which is equal to four. 1.3. A brief historical overview on restriction problems 1.3.1. The Euclidean framework Fourier restriction’s problem which was introduced by E. Stein dates back to the 60s: given a hypersurface \hat{S}\subset\hat{\mathbb{R}}^{n} endowed with a smooth measure d\sigma , the restriction problem asks for which p an inequality of the form (1.9) \|\mathcal{F}(f)|_{\hat{S}}\|_{L^{2}(\hat{S},d\sigma)}\leq C\|f\|_{L^{p}(\mathbb{R}^{n})}\,, holds for all f in the Schwartz space \mathcal{S}(\mathbb{R}^{n}) , where \mathcal{F}(f) denotes the Fourier transform of f on \mathbb{R}^{n} . When the Gaussian curvature of \hat{S} does not vanish at every point, this problem is fully understood, see [ 26 , 60 , 61 , 66 ] and the optimal range of indices p satisfying Inequality ( 1.9 ) is [1,p_{TS}] with p_{TS} the Tomas-Stein index given by p_{TS}=(2n+2)/(n+3) . It turns out that this problem is closely related to many questions in harmonic analysis, partial differential equations, Fourier analysis, spectral theory, number theory, and many others. As a matter of fact, this type of problem, which has been generalised to other hyper-surfaces in Euclidean spaces and other settings, remains a topical issue to this day. We refer the interested reader to the surveys on the subject with various points of view [ 2 , 20 , 39 , 63 ] and the references therein. In this work, we focus on the case of the nilmanifold M for which a Fourier transform is unavailable. We will then adopt the reformulation of the problem of Fourier restriction into a spectral problem as featured for instance in [ 2 , 59 ] . As we shall see in Section 3.1 , the spectral decomposition of the sub-Laplacian \mathcal{L}_{M} includes the spectral decomposition of the Laplacian on the two-dimensional flat torus given by ( 1.3 ). So, we dedicate the following paragraph to the strategy of proof of Zygmund’s theorem [ 70 ] regarding the Fourier restriction’s problem on the two dimensional torus. This emphasises the link between the restriction problem in the discrete case and number theory. 1.3.2. Sketch of the proof of Zygmund’s theorem Here we present the ideas

Read the complete text