Sharp trace inequalities for conformally invariant fractional powers of the sublaplacian on the Heisenberg group and the CR sphere

Qiaohua Yang, Leyuan Yu · 2026

Read Sharp trace inequalities for conformally invariant fractional powers of the sublaplacian on the Heisenberg group and the CR sphere by Qiaohua Yang, Leyuan Yu (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-2604-16771.txt.

Introduction

Yang and Yu prove sharp Sobolev trace inequalities for conformally invariant fractional powers of the sublap on the Heis group and the CR sphere. The results extend flat work by Einav-Loss, Beckner and Bez-Machihara-Sugimoto to non-flat settings. In the limiting case, sharp trace Beckner-Onofri inequalities are also proved on the CR sphere. Proofs combine a Bez-Machihara-Sugimoto duality argument with the Frank-Lieb sharp form of Hardy-Littlewood-Sobolev inequalities. The same approach yields trace Beckner-Onofri inequalities on the standard sphere.

Catalog record

CREATORQiaohua Yang, Leyuan Yu
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE63149
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2604.16771

Browse the full library · About the Tower of Babel

Full text

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

Sharp trace inequalities for conformally invariant fractional powers of the sublaplacian on the Heisenberg group and the CR sphere Thanks: The first author was partially supported by the National Natural Science Foundation of China(No.12471056). Qiaohua Yang, Leyuan Yu Abstract . We establish sharp Sobolev trace inequalities for conformally invariant fractional powers of the sublaplacian on the Heisenberg group and the CR sphere, extending the corresponding Euclidean results of Einav-Loss, Beckner, and Bez-Machihara-Sugimoto to these non-Euclidean settings. In the limiting case, sharp trace Beckner-Onofri inequalities are also established on the CR sphere. The proofs are based on a duality argument due to Bez-Machihara-Sugimoto, together with the Frank-Lieb sharp form of the Hardy-Littlewood-Sobolev inequalities on the Heisenberg group and the CR sphere. The same approach also yields trace Beckner-Onofri inequalities on the standard sphere. 2010 Mathematics Subject Classification Primary: 46E35; 35R03; 22E25. 1. Introduction Let n\geq 3 and denote by \mathbb{R}^{n}_{+}=\{(x,t)\in\mathbb{R}^{n}:x\in\mathbb{R}^{n-1},t>0\} the upper half-space in \mathbb{R}^{n} . The Sobolev trace inequality on \mathbb{R}^{n}_{+} , first established by Escobar [ 17 ] (see also Carlen–Loss [ 7 ] ), states that for any real-valued function f on \mathbb{R}^{n}_{+} , sufficiently smooth up to the boundary and decaying rapidly enough at infinity, the following holds: (1.1) \displaystyle\left(\int_{\mathbb{R}^{n-1}}|(\tau f)(x)|^{\frac{2(n-1)}{n-2}}dx\right)^{\frac{n-2}{n-1}}\leq C_{n}\int_{\mathbb{R}^{n}_{+}}|\nabla f(x,t)|^{2}dxdt, where C_{n}=\frac{1}{\sqrt{\pi}(n-2)}\Bigl(\frac{\Gamma(n-1)}{\Gamma\bigl(\frac{n-1}{2}\bigr)}\Bigr)^{\frac{1}{n-1}}, and (\tau f)(x) denotes the trace of f on \partial\mathbb{R}^{n}_{+} . The equality in ( 1.1 ) holds if and only if f is a multiple of \Bigl(\frac{1}{(\varepsilon+t)^{2}+|x-x_{0}|^{2}}\Bigr)^{(n-2)/2},\qquad\varepsilon>0,\;x_{0}\in\mathbb{R}^{n-1}. We remark that, via the seminal extension problem of Caffarelli and Silvestre [ 6 ] , sharp Sobolev trace inequalities on \mathbb{R}^{n+1}_{+} can be established for \gamma\in(0,1) (see Carlen–Loss [ 7 ] for \gamma=\frac{1}{2} ). In [ 10 ] , Case extended the framework of Caffarelli and Silvestre by introducing a family of higher-order boundary operators, yielding a corresponding family of sharp Sobolev trace inequalities on \mathbb{R}^{n+1}_{+} . Building on the work of Case et al. (who generalized these results to a conformally invariant class of manifolds and introduced the relevant conformally covariant boundary operators for the Paneitz and sixth-order GJMS cases [ 8 , 9 , 12 ] ), the extension to Poincaré-Einstein manifolds was later established by Flynn, Lu, and the second author [ 19 ] . Analogous results in the CR setting have been developed by Frank et al. [ 22 ] and by Flynn, Lu, and the second author [ 18 ] . For more information on Sobolev trace inequalities, we refer to [ 1 , 3 , 11 , 14 , 25 , 33 ] and the references therein. In their work [ 16 ] , Einav and Loss investigated sharp trace inequalities for fractional Sobolev spaces. We briefly summarize the results as follows. Define the space D_{s}(\mathbb{R}^{n}) to be the space of all functions f\in L^{2}(\mathbb{R}^{n}) whose Fourier transform \widehat{f} satisfies \int_{\mathbb{R}^{n}}|\widehat{f}(y)|^{2}|y|^{s}dy<\infty. For f\in D_{s}(\mathbb{R}^{n}) , let \mathbb{R}^{n-m}=\{(x_{1},\dots,x_{n-m},0,0,\dots,0):x_{j}\in\mathbb{R},1\leq j\leq n-m\} and let \tau_{m} denote the restriction operator that maps functions on \mathbb{R}^{n} to \mathbb{R}^{n-m} . Einav and Loss established the following sharp Sobolev trace inequality for fractional Sobolev spaces. Theorem 1.1 . ( [ 16 , Theorem 1.1] ) Let 0\leq m<n and \frac{m}{2}<s<\frac{n}{2} . For any f\in D_{2s}(\mathbb{R}^{n}) we have (1.2) \|\tau_{m}f\|_{L^{\frac{2(n-m)}{n-2s}}}^{2}\leq C_{m,s,n}\int_{\mathbb{R}^{n}}|(-\Delta)^{s/2}f|^{2}, where C_{m,s,n}=2^{-2s}\pi^{-s}\cdot\frac{\Gamma(n/2-s)\Gamma(s-m/2)}{\Gamma(s)\Gamma(n/2+s-m)}\left\{\frac{\Gamma(n-m)}{\Gamma((n-m)/2)}\right\}^{(2s-m)/(n-m)}. There is equality only if f(x) is proportional to \int_{\mathbb{R}^{m}}\frac{1}{\left(|x^{\prime}|^{2}+|x^{\prime\prime}-y^{\prime\prime}|^{2}\right)^{(n-2s)/2}}\frac{1}{\left(\gamma^{2}+|y^{\prime\prime}-a|^{2}\right)^{(n+2s-2m)/2}}dy^{\prime\prime} for some a\in\mathbb{R}^{n-m} and \gamma\neq 0 . In particular, when s=2 and m=1 , the inequality ( 1.2 ) recovers the earlier result of Escobar. Moreover, Beckner [ 3 ] and Bez-Machihara-Sugimoto [ 4 ] established another type of Sobolev trace inequality for fractional Sobolev spaces. Unlike Beckner’s approach, Bez–Machihara-Sugimoto employed a dual argument. Specifically, they proved the following theorem: Theorem 1.2 . ( [ 3 , Theorem 5] , [ 4 , Theorem 4.1] ) Let n\geq 2 and s\in\left(\frac{1}{2},\frac{n}{2}\right) . For f\in D_{2s}(\mathbb{R}^{n}) , (1.3) \|\mathcal{R}f\|_{L^{\frac{2(n-1)}{n-2s}}(\mathbb{S}^{n-1})}^{2}\leq 2^{1-2s}\frac{\Gamma(2s-1)\Gamma\left(\frac{n}{2}-s\right)}{\Gamma(s)^{2}\Gamma\left(\frac{n}{2}-1+s\right)}\left(\frac{\Gamma\left(\frac{n}{2}\right)}{2\pi^{\frac{n}{2}}}\right)^{\frac{2s-1}{n-1}}\int_{\mathbb{R}^{n}}|(-\Delta)^{s/2}f|^{2}. Here \mathcal{R} denotes the operator which restricts f on \mathbb{R}^{n} to the unit sphere \mathbb{S}^{n-1}\subset\mathbb{R}^{n} . By stereographic projection, inequality ( 1.3 ) is in fact equivalent to the case m=1 of inequality ( 1.2 ). This equivalence can be observed from our subsequent proof in the case of Heisenberg group. The purpose of this paper is to extend Theorems 1.1 and 1.2 to the settings of the Heisenberg group and the CR sphere. To state our main results, we first introduce some necessary notations. Recall that the Heisenberg group \mathbb{H}^{n} is \mathbb{C}^{n}\times\mathbb{R} with elements u=(z,t) and group law uu^{\prime}=(z,t)(w,\mu)=\bigl(z+w,t+\mu+2\operatorname{Im}(z\cdot\overline{w})\bigr), where z\cdot\overline{w}=\sum_{j=1}^{n}z_{j}\overline{w}_{j}. Here \operatorname{Im}z and \operatorname{Re}z denote, respectively, the imaginary and real parts of z . The Haar measure on \mathbb{H}^{n} is the usual Lebesgue measure du=dzdt . We denote the convolution on the Heisenberg group by f*g(u)=\int_{\mathbb{H}^{n}}f(v)g(v^{-1}u)dv. The homogeneous norm on \mathbb{H}^{n} is given by |u|=|(z,t)|=\bigl(|z|^{4}+t^{2}\bigr)^{1/4}, and dilations of a point u=(z,t) are defined as \delta_{r}(u)=(rz,r^{2}t),\qquad r>0. Denote by Q_{n} the homogeneous dimension of \mathbb{H}^{n} . Then Q_{n}=2n+2. The sublaplacian on \mathbb{H}^{n} is the second-order differential operator \mathcal{L}_{0}=-\frac{1}{4}\sum_{j=1}^{n}\bigl(X_{j}^{2}+Y_{j}^{2}\bigr), where X_{j}=\frac{\partial}{\partial x_{j}}+2y_{j}\frac{\partial}{\partial t},\qquad Y_{j}=\frac{\partial}{\partial y_{j}}-2x_{j}\frac{\partial}{\partial t},\quad j=1,\cdots,n. The conformally invariant fractional powers of the sublaplacian on \mathbb{H}^{n} are defined by (see [ 15 , 29 ] ) \mathcal{L}_{s,\mathbb{H}^{n}}\;:=\;|2T|^{s/2}\frac{\Gamma\left(\mathcal{L}_{0}|2T|^{-1}+\frac{2+s}{4}\right)}{\Gamma\left(\mathcal{L}_{0}|2T|^{-1}+\frac{2-s}{4}\right)}, where T=\frac{\partial}{\partial t} . The operator \mathcal{L}_{s,\mathbb{H}^{n}} is left-invariant and satisfies the homogeneity property (1.4) \displaystyle\mathcal{L}_{s,\mathbb{H}^{n}}\bigl(f\circ\delta_{r}\bigr)=r^{s}\bigl(\mathcal{L}_{s,\mathbb{H}^{n}}f\bigr)\circ\delta_{r},\qquad f\in C_{0}^{\infty}(\mathbb{H}^{n}). The Sobolev space W^{s,2}(\mathbb{H}^{n}) is defined as the completion of C_{0}^{\infty}(\mathbb{H}^{n}) with respect to the norm \|f\|_{W^{s,2}(\mathbb{H}^{n})}\;:=\;\Bigl(\int_{\mathbb{H}^{n}}\bigl|\mathcal{L}_{s,\mathbb{H}^{n}}^{1/2}f\bigr|^{2}\Bigr)^{1/2}. Let \mathbb{S}^{2n+1}=\Bigl\{z=(\zeta_{1},\dots,\zeta_{n+1})\in\mathbb{C}^{n+1}\;:\;\sum_{j=1}^{n+1}|\zeta_{j}|^{2}=1\Bigr\} be the CR sphere. The sublaplacian on \mathbb{S}^{2n+1} is defined as \mathcal{L}=-\frac{1}{2}\sum_{j=1}^{n+1}\bigl(T_{j}\overline{T}_{j}+\overline{T}_{j}T_{j}\bigr), where T_{j}=\frac{\partial}{\partial\zeta_{j}}-\overline{\zeta}_{j}\sum_{k=1}^{n+1}\zeta_{k}\frac{\partial}{\partial\zeta_{k}},\quad j=1,\cdots,n+1. The transversal direction is given by the real vector field \mathcal{T}=\frac{i}{2}\sum_{j=1}^{n+1}\Bigl(\zeta_{j}\frac{\partial}{\partial\zeta_{j}}-\overline{\zeta}_{j}\frac{\partial}{\partial\overline{\zeta}_{j}}\Bigr). The conformal sublaplacian on \mathbb{S}^{2n+1} is defined as \mathcal{D}=\mathcal{L}+\frac{n^{2}}{4}. The Sobolev space W^{s,2}(\mathbb{S}^{2n+1}) is defined as the completion of C^{\infty}(\mathbb{S}^{2n+1}) with respect to the norm \|F\|_{W^{s,2}(\mathbb{S}^{2n+1})}=\|\mathcal{D}^{s/2}F\|_{L^{2}(\mathbb{S}^{2n+1})}. It is well known that L^{2}(\mathbb{S}^{2n+1}) admits a decomposition into U(n+1) -irreducible components L^{2}(\mathbb{S}^{2n+1})=\bigoplus_{j,k\geq 0}\mathcal{H}_{j,k}, where \mathcal{H}_{j,k} consists of the restrictions to \mathbb{S}^{2n+1} of harmonic polynomials p(z,\overline{z}) on \mathbb{C}^{n+1} that are homogeneous of degree j in z and degree k in \overline{z} (see [ 21 , 31 ] and references therein). We define the operator \mathcal{A}_{s,\mathbb{S}^{2n+1}} on W^{s,2}(\mathbb{S}^{2n+1}) for -Q_{n}<s\leq Q_{n} by its action on the spherical harmonics Y_{jk}\in\mathcal{H}_{j,k} : (1.5) \displaystyle\mathcal{A}_{s,\mathbb{S}^{2n+1}}Y_{jk}=\lambda_{j}(s)\lambda_{k}(s)Y_{jk}, where \lambda_{j}(s)=\frac{\Gamma\bigl(\frac{Q_{n}+s}{4}+j\bigr)}{\Gamma\bigl(\frac{Q_{n}-s}{4}+j\bigr)}. In terms of the conformal sublaplacian \mathcal{D} and the transversal vector field \mathcal{T} , the operator \mathcal{A}_{s,\mathbb{S}^{2n+1}} can be expressed as (see Section 2.2; for the case where s is an even integer, see [ 5 ] ) (1.6) \displaystyle\mathcal{A}_{s,\mathbb{S}^{2n+1}}=\frac{\Gamma\Bigl(\sqrt{\mathcal{D}-\mathcal{T}^{2}}-i\mathcal{T}+\frac{2+s}{4}\Bigr)\Gamma\Bigl(\sqrt{\mathcal{D}-\mathcal{T}^{2}}+i\mathcal{T}+\frac{2+s}{4}\Bigr)}{\Gamma\Bigl(\sqrt{\mathcal{D}-\mathcal{T}^{2}}-i\mathcal{T}+\frac{2-s}{4}\Bigr)\Gamma\Bigl(\sqrt{\mathcal{D}-\mathcal{T}^{2}}+i\mathcal{T}+\frac{2-s}{4}\Bigr)}. Set \mathbb{H}^{n-m}=\{(z,t)\in\mathbb{H}^{n}:z=(z_{1},\dots,z_{n-m},0,\dots,0)\}, considered as a subgroup of \mathbb{H}^{n} via the natural embedding. For convenience, we write a point z\in\mathbb{C}^{n} as z=(z^{\prime},z^{\prime\prime}) with z^{\prime}=(z_{1},\dots,z_{n-m})\in\mathbb{C}^{n-m},\qquad z^{\prime\prime}=(z_{n-m+1},\dots,z_{n})\in\mathbb{C}^{m}. Let \mathcal{R}_{m} denote the restriction operator that maps functions on \mathbb{H}^{n} to \mathbb{H}^{n-m} . As shown in Lemma 3.2 , \mathcal{R}_{m} extends to a bounded operator from W^{s,2}(\mathbb{H}^{n}) to W^{s-2m,2}(\mathbb{H}^{n-m}) . Correspondingly, we set (1.7) \displaystyle\mathbb{S}^{2(n-m)+1}=\left\{\zeta\in\mathbb{S}^{2n+1}:\zeta_{k}=0,\;n-m+2\leq k\leq n+1\right\}. Let \widetilde{\mathcal{R}}_{m} be the operator that restricts functions on \mathbb{S}^{2n+1} to \mathbb{S}^{2(n-m)+1} . For brevity, and when no confusion is likely, we also identify (1.8) \displaystyle\mathbb{S}^{2(n-m)+1}= \displaystyle\Bigl\{(\zeta,t):\sum_{j=1}^{n-m+1}|\zeta_{j}|^{2}=1,\;\zeta_{k}=0,\;n-m+2\leq k\leq n,\;t=0\Bigr\} \displaystyle\subset \displaystyle\mathbb{C}^{n-m+1}\times\{0\}\times\cdots\times\{0\}\subset\mathbb{H}^{n}. With this identification, the notation \widetilde{\mathcal{R}}_{m} will also be used for the restriction of functions on \mathbb{H}^{n} to \mathbb{S}^{2(n-m)+1} . Employing the method of Bez–Machihara–Sugimoto together with the sharp form of the Hardy–Littlewood–Sobolev inequality on the Heisenberg group due to Frank and Lieb, we obtain the following three types of Sobolev trace inequalities on \mathbb{H}^{n} and \mathbb{S}^{2n+1} . Theorem 1.3 . Let 1\leq m<n , 2m<s<Q_{n} , and p=\frac{2(Q_{n}-2m)}{Q_{n}-s}

Read the complete text