Table of Contents
Fetching ...

Optimal subelliptic super-Poincaré and isoperimetric inequalities on stratified Lie groups

Yaozhong W. Qiu

TL;DR

The paper develops an $L^1$-oriented framework to obtain $q$-super-Poincaré and isoperimetric inequalities for exponential-power measures $d\mu = Z^{-1}e^{-N^p}d\xi$ defined via a homogeneous norm $N$ in subelliptic geometries. It relies on $U$-bounds, Hardy inequalities, and $L^1$-Caffarelli–Kohn–Nirenberg tools to derive a $q$-super-Poincaré inequality with growth $\beta_q(\varepsilon) \lesssim \exp(C\varepsilon^{-p(\alpha+1)/(q(p-\alpha-1))})$ for $q\in[1,2]$, and proves isoperimetric bounds $I_\mu \gtrsim \mathcal{U}_r$ with $r = \frac{(\alpha+1)p}{(\alpha+1) + \alpha p}$, which are optimal in several cases. The results are validated across concrete subelliptic settings, including step-two stratified Lie groups, Grushin, and Heisenberg–Greiner operators, as well as an anisotropic Heisenberg group, demonstrating that the tail behavior can be supergaussian while still yielding sharp subgaussian-type isoperimetry. The approach avoids curvature-based assumptions and provides explicit growth rates for the inequalities, with implications for spectral theory and diffusion on non-Euclidean spaces.

Abstract

We prove $q$-super-Poincaré inequalities, $q \in [1, 2]$, for a class of exponential power type probability measures defined in terms of a norm in a number of subelliptic settings, primarily on stratified Lie groups but also in the Grushin and Heisenberg-Greiner settings. Our results include generically optimal isoperimetric inequalities for such probability measures.

Optimal subelliptic super-Poincaré and isoperimetric inequalities on stratified Lie groups

TL;DR

The paper develops an -oriented framework to obtain -super-Poincaré and isoperimetric inequalities for exponential-power measures defined via a homogeneous norm in subelliptic geometries. It relies on -bounds, Hardy inequalities, and -Caffarelli–Kohn–Nirenberg tools to derive a -super-Poincaré inequality with growth for , and proves isoperimetric bounds with , which are optimal in several cases. The results are validated across concrete subelliptic settings, including step-two stratified Lie groups, Grushin, and Heisenberg–Greiner operators, as well as an anisotropic Heisenberg group, demonstrating that the tail behavior can be supergaussian while still yielding sharp subgaussian-type isoperimetry. The approach avoids curvature-based assumptions and provides explicit growth rates for the inequalities, with implications for spectral theory and diffusion on non-Euclidean spaces.

Abstract

We prove -super-Poincaré inequalities, , for a class of exponential power type probability measures defined in terms of a norm in a number of subelliptic settings, primarily on stratified Lie groups but also in the Grushin and Heisenberg-Greiner settings. Our results include generically optimal isoperimetric inequalities for such probability measures.

Paper Structure

This paper contains 11 sections, 13 theorems, 84 equations.

Key Result

Theorem 1

Assume $N$ satisfies in the distributional sense for some $\alpha > 0$ and where $\left\lvert x \right\rvert$ is the euclidean norm of a generic point $\xi \in \mathbb{R}^n$ written in the form $\xi = (x, x') \in \mathbb{R}^{n_1} \times \mathbb{R}^{n_2}$ with $n_1 \geq 2$. Assume moreover $\left\lvert x \right\rvert$ satisfies again in the distrib and If there exists a local $1$-super-Poincaré i

Theorems & Definitions (25)

  • Definition 1
  • Theorem 1
  • Lemma 1
  • proof
  • Lemma 2
  • Remark 1
  • proof
  • Proposition 3
  • Remark 2
  • proof
  • ...and 15 more