Spectral multipliers on Métivier groups

Lars Niedorf · 2024

Read Spectral multipliers on Métivier groups by Lars Niedorf (2024) 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-2412-07920.txt.

Introduction

Lars Niedorf proves a sharp bound rule for wave filters on a class of curved group spaces. Such spaces, called Metivier groups, carry a lap-style operator whose bands the filters reweight. A clean rule for when those filters stay tame on each power space was open. His rule needs just past a set edge tied to the space size and the power in play.

The proof leans on a curb-style bound that looks too weak at first sight. A shape quirk saves it: the first layer of these spaces runs far wider than the next. That width, kin to a known count set, feeds the weak bound with room to spare. Filter rules for kin spaces may gain from the same trick.

Catalog record

CREATORLars Niedorf
YEAR2024
COLLECTIONPAPERS
FORMATTXT
FILE SIZE123 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2412-07920

Browse the full library · About the Tower of Babel

Full text

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

Spectral multipliers on Métivier groups Thanks: The author gratefully acknowledges the support by the Deutsche Forschungsgemeinschaft (DFG) through grant MU 761/12-1. Lars Niedorf Abstract . We prove an L p L^{p} -spectral multiplier theorem under the sharp regularity condition s > d ​ | 1 / p − 1 / 2 | s>d\left|1/p-1/2\right| for sub-Laplacians on Métivier groups. The proof is based on a restriction type estimate which, at first sight, seems to be suboptimal for proving sharp spectral multiplier results, but turns out to be surprisingly effective. This is achieved by exploiting the structural property that for any Métivier group the first layer of any stratification of its Lie algebra is typically much larger than the second layer, a phenomenon closely related to Radon–Hurwitz numbers. 2020 Mathematics Subject Classification 42B15, 22E25, 22E30, 43A85 1. Introduction 1.1. Spectral multipliers The celebrated Mikhlin–Hörmander theorem [ 18 ] states that the Fourier multiplier operator T given by \widehat{Tf}(\xi)=F(\xi)\hat{f}(\xi),\quad f\in\mathcal{S}(\mathbb{R}^{n}),\xi\in\mathbb{R}^{n} is of weak type (1,1) and extends to a bounded operator on L^{p}(\mathbb{R}^{n}) for all p\in(1,\infty) whenever the Fourier multiplier F:\mathbb{R}^{n}\to\mathbb{C} satisfies \|F\|_{L^{2}_{s,\mathrm{sloc}}(\mathbb{R}^{n})}:=\sup_{t>0}{\lVert F(t\,\cdot\,)\eta\rVert_{L^{2}_{s}(\mathbb{R}^{n})}}<\infty\quad\text{for some }s>n/2. Here \eta:\mathbb{R}^{n}\to\mathbb{C} denotes some nonzero smooth function which is compactly supported in an annulus away from the origin, and L^{2}_{s}(\mathbb{R}^{n})\subseteq L^{2}(\mathbb{R}^{n}) is the Sobolev space of fractional order s\geq 0 . Note that different choices of \eta lead to equivalent norms. The regularity condition s>n/2 is sharp and cannot be decreased, see for example [ 42 ] . The Mikhlin–Hörmander theorem implies the following spectral multiplier result for the Laplacian -\Delta=-(\partial_{x_{1}}^{2}+\dots+\partial_{x_{n}}^{2}) on \mathbb{R}^{n} . Given a Borel measurable function F:\mathbb{R}\to\mathbb{C} , which we call a spectral multiplier in the following, the operator F(-\Delta) defined by functional calculus corresponds to a radial Fourier multiplier operator since (F(-\Delta)f)^{\wedge}(\xi)=F(|\xi|^{2})\hat{f}(\xi). Thus, the Mikhlin–Hörmander theorem implies that F(-\Delta) is bounded as an operator on L^{p}(\mathbb{R}^{n}) for all p\in(1,\infty) and is of weak type (1,1) whenever \|F\|_{L^{2}_{s,\mathrm{sloc}}}:=\sup_{t>0}{\lVert F(t\,\cdot\,)\eta\rVert_{L^{2}_{s}(\mathbb{R})}}<\infty\quad\text{for some }s>n/2. 1.2. Sub-Laplacians Let G be a stratified Lie group and \mathfrak{g}=\mathfrak{g}_{1}\oplus\dots\oplus\mathfrak{g}_{r} be the stratification of the Lie algebra of G , that is, [\mathfrak{g}_{i},\mathfrak{g}_{j}]\subseteq\mathfrak{g}_{i+j} for all i,j\in\mathbb{N}\setminus\{0\} (with the understanding that \mathfrak{g}_{m}=\{0\} for m>r ), and \mathfrak{g}_{1} generates \mathfrak{g} as a Lie algebra. Given a basis X_{1},\dots,X_{d_{1}} of the first layer \mathfrak{g}_{1} , we can identify it with left-invariant vector fields on G via the Lie derivative, and consider the associated left-invariant sub-Laplacian L , which is given by L=-(X_{1}^{2}+\dots+X_{d_{1}}^{2}). Let G be equipped with a left-invariant Haar measure and L^{p}(G) be the associated Lebesgue space. The sub-Laplacian L is positive (that is, (Lf,f)_{L^{2}(G)}\geq 0 for all functions f in the domain of L , which is given by \{f\in L^{2}(G):Lf\in L^{2}(G)\} ) and self-adjoint on L^{2}(G) . Hence, one can define via functional calculus for every spectral multiplier F:\mathbb{R}\to\mathbb{C} the operator F(L)=\int_{0}^{\infty}F(\lambda)\,dE(\lambda)\quad\text{on }L^{2}(G), where E denotes the spectral measure of L . By the spectral theorem, the operator F(L) is bounded on L^{2} if and only if the multiplier F is E -essentially bounded. Christ [ 7 ] , and Mauceri and Meda [ 30 ] showed that F(L) is of weak type (1,1) and bounded on all L^{p} -spaces for p\in(1,\infty) if \|F\|_{L^{2}_{s,\mathrm{sloc}}}<\infty\quad\text{for some }s>Q/2, where Q=\dim\mathfrak{g}_{1}+2\dim\mathfrak{g}_{2}+\dots+r\dim\mathfrak{g}_{r} is the homogeneous dimension of G . For sub-Laplacians on Heisenberg (type) groups, Müller and Stein [ 33 ] , and Hebisch [ 17 ] showed that the threshold s>Q/2 can even be pushed down to s>d/2 , with d being the topological dimension of the underlying group. Since then, this result has been extended to other specific classes of two-step stratified Lie groups [ 28 , 25 , 29 ] , and also to related settings [ 9 , 24 , 8 , 2 , 10 , 11 , 12 ] . Conversely, Martini, Müller, and Nicolussi Golo [ 23 ] were able to show for a very general class of smooth second-order real differential operators associated with a bracket-generating sub-Riemannian structure on smooth d -dimensional manifolds that at least regularity of order s\geq d\left|1/p-1/2\right| is necessary for having the spectral multiplier estimate \lVert F(L)\rVert_{p\to p}\leq C_{p,s}\|F\|_{L^{\infty}_{s,\mathrm{sloc}}} for every bounded Borel function F , where \|F\|_{L^{\infty}_{s,\mathrm{sloc}}}:=\sup_{t>0}{\lVert F(t\,\cdot\,)\eta\rVert_{L^{\infty}_{s}(\mathbb{R}^{n})}}. In particular, their result implies that regularity of order s\geq d/2 is necessary to have boundedness on all L^{p} -spaces. On the other hand, suppose we have a Mikhlin–Hörmander type result of the form \|F(L)\|_{p\to p}\leq C_{p,s}\|F\|_{L^{2}_{s,\mathrm{sloc}}}\quad\text{for all }p\in(1,\infty)\text{ and }s>d/2. Then, replacing \|\cdot\|_{L^{2}_{s,\mathrm{sloc}}} by the stronger norm \|\cdot\|_{L^{\infty}_{s,\mathrm{sloc}}} , we obtain \|F(L)\|_{p\to p}\leq C_{p,s}\|F\|_{L^{\infty}_{s,\mathrm{sloc}}}\quad\text{for all }s>d\left|1/p-1/2\right|, and all p\in(1,\infty) , which follows by interpolation with the trivial L^{2} - L^{2} estimate from the spectral theorem. In the range 1\leq p\leq 2(d+1)/(d+3) such p -specific spectral multiplier estimates are known to hold for radial Fourier multipliers on \mathbb{R}^{n} even with the weaker \smash{L^{2}_{s,\mathrm{sloc}}} -norm, see [ 6 , 41 , 21 ] . With the \smash{L^{2}_{s,\mathrm{sloc}}} -norm in place of the \smash{L^{\infty}_{s,\mathrm{sloc}}} -norm on the right-hand side, these p -specific spectral multiplier estimates yield in particular that the Bochner-Riesz means given by F(L)=(1-tL)^{\delta}_{+},\quad t>0 are uniformly bounded on L^{p} under the regularity condition \delta>d\left|1/p-1/2\right|-1/2 , which is the same order of regularity as in the Bochner-Riesz conjecture, see [ 44 ] . So far it is an open question if regularity of order s>\max\{d\left|1/p-1/2\right|,1/2\} with respect to the L^{2}_{s,\mathrm{sloc}} -norm is generally sufficient to provide spectral multiplier estimates. In sub-elliptic settings, the only results in the literature seem to be those of [ 5 , 37 ] for Grushin operators and those of [ 38 ] for sub-Laplacians on Heisenberg type groups. 1.3. Métivier groups versus Heisenberg type groups The purpose of this paper is to extend the spectral multiplier result for sub-Laplacians on Heisenberg type groups of [ 38 ] to the larger class of Métivier groups. Before stating the main result, we need to introduce some notation. Let G be a two-step stratified Lie group, that is, a connected, simply connected nilpotent Lie group whose Lie algebra \mathfrak{g} (which is the tangent space T_{e}G at the identity e of G ) admits a decomposition \mathfrak{g}=\mathfrak{g}_{1}\oplus\mathfrak{g}_{2} into two non-trivial subspaces \mathfrak{g}_{1},\mathfrak{g}_{2}\subseteq\mathfrak{g} , where [\mathfrak{g}_{1},\mathfrak{g}_{1}]=\mathfrak{g}_{2} and \mathfrak{g}_{2}\subseteq\mathfrak{g} is contained in the center of \mathfrak{g} . In the following, we refer to \mathfrak{g}_{1} and \mathfrak{g}_{2} as being the first and second layer of \mathfrak{g} , respectively. Let (1.1) d_{1}=\dim\mathfrak{g}_{1}\geq 1,\quad d_{2}=\dim\mathfrak{g}_{2}\geq 1\quad\text{and}\quad d=\dim\mathfrak{g}. Given a basis X_{1},\dots,X_{d_{1}} of the first layer \mathfrak{g}_{1} , we identify each element of the basis by a left-invariant vector field on G via the Lie derivative, and consider the associated sub-Laplacian L , which is the second-order differential operator (1.2) L=-(X_{1}^{2}+\dots+X_{d_{1}}^{2}). Let \mathfrak{g}_{2}^{*} denote the dual of \mathfrak{g}_{2} . Any element \mu\in\mathfrak{g}_{2}^{*} gives rise to a skew-symmetric bilinear form \omega_{\mu}:\mathfrak{g}_{1}\times\mathfrak{g}_{1}\to\mathbb{R} via \omega_{\mu}(x,x^{\prime})=\mu([x,x^{\prime}]),\quad x,x^{\prime}\in\mathfrak{g}_{1}. The group G is called a Métivier group if \omega_{\mu} is non-degenerate for all \mu\in\mathfrak{g}_{2}^{*}\setminus\{0\} , which means that \mathfrak{r}_{\mu}=\{0\} for the radical \mathfrak{r}_{\mu}=\{x\in\mathfrak{g}_{1}:\omega_{\mu}(x,x^{\prime})=0\text{ for all }x^{\prime}\in\mathfrak{g}_{1}\}. We choose a basis U_{1},\dots,U_{d_{2}} of the second layer \mathfrak{g}_{2} . Let \langle\cdot,\cdot\rangle be the inner product rendering X_{1},\dots,X_{d_{1}},U_{1},\dots,U_{d_{2}} an orthonormal basis of \mathfrak{g} . The inner product \langle\cdot,\cdot\rangle induces a norm on the dual \mathfrak{g}_{2}^{*} which we denote by |\cdot| . Let J_{\mu} be the skew-symmetric endomorphism such that (1.3) \omega_{\mu}(x,x^{\prime})=\langle J_{\mu}x,x^{\prime}\rangle,\quad x,x^{\prime}\in\mathfrak{g}_{1}. Then G is a Métivier group if and

Read the complete text