[Paper Review] A new look at the John-Nirenberg and John-Stromberg theorems for BMO. Lecture Notes

Michael Ćwikel, Yoram Sagher+1 more|arXiv (Cornell University)|Nov 2, 2010

Advanced Harmonic Analysis Research18 references3 citations

TL;DR

This paper presents a new geometric approach to the John-Nirenberg and John-Strömberg theorems for BMO functions, introducing a conjectural geometric condition involving measurable subsets of cubes. An affirmative answer to this condition would yield a dimension-free version of the John-Nirenberg inequality with explicit, geometrically interpretable constants.

ABSTRACT

We develop some techniques for studying various versions of the function space BMO. Special cases of one of our results give alternative proofs of the celebrated John- Nirenberg inequality and of related inequalities due to John and to Wik. Our approach enables us to pose a simply formulated "geometric" question, for which an affirmative answer would lead to a version of the John-Nirenberg inequality with dimension free constants. A more detailed summary of the main ideas and results of this paper can be found at http://www.math.technion.ac.il/~mcwikel/bmo/CwikSaghShvaSummary.pdf

Motivation & Objective

• To re-derive the John-Nirenberg inequality using a novel geometric perspective on BMO functions.
• To investigate whether the constants in the John-Nirenberg inequality can be made independent of dimension.
• To formulate a geometric condition (Question A) that, if satisfied, would imply dimension-free BMO bounds.
• To establish a precise link between geometric properties of measurable sets and analytic norms in BMO spaces.
• To show that answering Question A in special cases (e.g., dyadic cubes) suffices for a general solution.

Proposed method

• Introduces a geometric condition (Question A) involving two disjoint measurable subsets $E_+$ and $E_-$ of a cube $Q$, requiring a positive lower bound on their relative measure and the existence of a subcube $W$ intersecting both with positive measure.
• Defines the function $\mathbf{J}(f,Q,s)$ to measure the distribution of oscillation of $f$ on $Q$, linking it to the distribution of the decreasing rearrangement $f^*$.
• Uses scaling and translation invariance to show that the geometric condition is preserved under affine transformations, enabling reduction to standard cubes.
• Applies the decreasing rearrangement $f^*$ to compare $BMO$ norms of $f$ and $f^*$, establishing equivalence between geometric and analytic conditions.
• Proves left-continuity and monotonicity of $s \mapsto \mathbf{J}(f,Q,s)$, crucial for analyzing the distribution of oscillation.
• Reduces the general problem to dyadic subcubes, showing that a solution in the dyadic case implies a solution in general.

Experimental results

Research questions

• RQ1 Can the John-Nirenberg inequality be proven using a purely geometric condition on measurable subsets of cubes?
• RQ2 Is there a universal pair of constants $\tau \in (0,1/2)$ and $s > 0$ independent of dimension $d$ satisfying the geometric condition in Question A?
• RQ3 Does the existence of such constants imply a dimension-free version of the John-Nirenberg inequality?
• RQ4 Can the equivalence between geometric and analytic $BMO$-norm conditions be fully characterized?
• RQ5 Is it sufficient to verify Question A for dyadic subcubes to establish it in full generality?

Key findings

• An affirmative answer to Question A would imply a dimension-free version of the John-Nirenberg inequality with explicit constants depending on $\tau$ and $s$.
• For each fixed dimension $d$, constants $\tau \in (0,1/2)$ and $s > 0$ exist satisfying the geometric condition, but $s$ depends on $d$.
• With $\tau = \sqrt{2} - 1$, the best available $s = 2^{-d}(3 - 2\sqrt{2})$ still depends on $d$, showing current limitations.
• The geometric condition is equivalent to a comparison between $BMO$ norms of $f$ and its decreasing rearrangement $f^*$, as formalized in Theorems 8.3 and 8.5.
• The function $s \mapsto \mathbf{J}(f,Q,s)$ is non-increasing and left-continuous, a key technical property for the analysis.
• The problem reduces to the dyadic case: if Question A holds for finite unions of dyadic cubes, it holds in general.