[Paper Review] Spaces H^1 and BMO on ax+b-groups

Maria Vallarino|ArXiv.org|Apr 29, 2008

Advanced Harmonic Analysis Research16 references4 citations

TL;DR

This paper establishes a Hardy space $H^1$ and BMO theory on the $ax+b$-group $S = \mathbb{R}^d \ltimes \mathbb{R}^+$, a non-unimodular Lie group of exponential growth. It defines atomic Hardy spaces $H^{1,p}$ and a BMO space using Calderón–Zygmund sets, proves the John–Nirenberg inequality, identifies $BMO$ as the dual of $H^1$, and shows boundedness of singular integrals from $H^1$ to $L^1$ and from $L^\infty$ to $BMO$, with real interpolation yielding $L^p$ spaces.

ABSTRACT

Let S be the semidirect product of R^d and R^+ endowed with the Riemannian symmetric space metric and the right Haar measure: this is a Lie group of exponential growth. In this paper we define an Hardy space H^1 and a BMO space in this context. We prove that the functions in BMO satisfy the John-Nirenberg inequality and that BMO may be identified with the dual space of H^1. We then prove that singular integral operators which satisfy a suitable integral Hormander condition are bounded from H^1 to L^1 and from L^{\infty} to BMO. We also study the real interpolation between H^1, BMO and the L^p spaces.

Motivation & Objective

• To develop a $H^1$–$BMO$ theory in the context of the $ax+b$-group $S = \mathbb{R}^d \ltimes \mathbb{R}^+$, a space of exponential growth where classical doubling-based theories fail.
• To define atomic Hardy spaces $H^{1,p}$ and a BMO space using Calderón–Zygmund sets, replacing balls in the classical theory.
• To prove that functions in the defined BMO space satisfy the John–Nirenberg inequality and that $BMO$ is isometrically isomorphic to the dual of $H^1$.
• To establish boundedness of singular integral operators with kernels satisfying an integral Hörmander condition from $H^1$ to $L^1$ and from $L^\infty$ to $BMO$.
• To characterize real interpolation spaces between $H^1$, $BMO$, and $L^p$ spaces, showing $[H^1, BMO]_{\theta,q} = L^{p,q}$ with $1/p = 1 - \theta$.

Proposed method

• Define Calderón–Zygmund sets in $S$ as the geometric replacement for balls in the classical theory, based on the Riemannian metric and right Haar measure.
• Introduce atomic Hardy spaces $H^{1,p}$ via functions supported in Calderón–Zygmund sets, with vanishing integral and $L^p$-size control.
• Define the BMO space using seminorms over Calderón–Zygmund sets, analogous to the classical definition with balls.
• Prove the John–Nirenberg inequality for the BMO space by adapting techniques from the classical theory to the non-doubling, exponential growth setting.
• Establish duality between $H^1$ and $BMO$ using atomic decomposition and duality theorems, showing $BMO = (H^1)^*$.
• Apply real interpolation theory, using the duality and density results to characterize $[H^1, BMO]_{\theta,q}$ and $[L^{q_1}, BMO]_{\theta,q}$ as $L^{p,q}$ and $L^q$ respectively.

Experimental results

Research questions

• RQ1 Can a $H^1$–$BMO$ theory be developed in spaces of exponential growth, such as the $ax+b$-group, where the doubling condition fails?
• RQ2 Do functions in the defined BMO space satisfy the John–Nirenberg inequality in this non-doubling, non-compact setting?
• RQ3 Is the BMO space isometrically isomorphic to the dual of the Hardy space $H^1$ in this context?
• RQ4 Are singular integral operators with kernels satisfying an integral Hörmander condition bounded from $H^1$ to $L^1$ and from $L^\infty$ to $BMO$?
• RQ5 What are the real interpolation spaces between $H^1$, $BMO$, and $L^p$ in this setting?

Key findings

• The spaces $H^{1,p}$ for $p \in (1,\infty]$ are equivalent, generalizing the classical result to the non-doubling, exponential growth setting.
• The defined BMO space satisfies the John–Nirenberg inequality, a key property ensuring its well-behaved oscillation control.
• The BMO space is isometrically isomorphic to the dual of $H^1$, establishing a central duality in the theory.
• Singular integral operators with kernels satisfying an integral Hörmander condition are bounded from $H^1$ to $L^1$ and from $L^\infty$ to $BMO$.
• Real interpolation yields $[H^1, BMO]_{\theta,q} = L^{p,q}$ with $1/p = 1 - \theta$, and $[L^{q_1}, BMO]_{\theta,q} = L^q$ with $1/q = (1 - \theta)/q_1$, extending classical interpolation results.
• The interpolation result $[H^1, L^{\infty}]_{\theta,p} = L^p$ holds with $1/p = 1 - \theta$, confirming consistency with the classical limit case.