[Paper Review] Iterated commutators under a joint condition on the tuple of multiplying functions

Tuomas Hytönen, Kangwei Li+1 more|arXiv (Cornell University)|Jan 1, 2019

Advanced Harmonic Analysis Research16 references4 citations

TL;DR

This paper introduces a new joint condition on two functions $b_1, b_2$ that characterizes the $L^2$ boundedness of the iterated commutator $[b_2, [b_1, T]]$ with a Calderón-Zygmund operator $T$, using two bisublinear mean oscillation norms $S_p$ and $T_p$. It proves that the commutator is bounded on $L^2$ if and only if $S_2(b_1,b_2) + T_2(b_1,b_2) < ∞$, and shows that this condition is sharp, improving upon the classical $BMO$ condition.

ABSTRACT

We present a pair of joint conditions on the two functions $b_1,b_2$ strictly weaker than $b_1,b_2\\in \\operatorname{BMO}$ that almost characterize the $L^2$ boundedness of the iterated commutator $[b_2,[b_1,T]]$ of these functions and a Calder\\'on-Zygmund operator $T.$ Namely, we sandwich this boundedness between two bisublinear mean oscillation conditions of which one is a slightly bumped up version of the other.

Motivation & Objective

• To identify a joint condition on $b_1, b_2$ weaker than individual $BMO$ membership that characterizes the $L^2$ boundedness of the iterated commutator $[b_2, [b_1, T]]$.
• To extend the sparse domination principle to iterated commutators under joint conditions, improving on classical $BMO$-based bounds.
• To demonstrate that the $S_2 + T_2$ condition is both necessary and sufficient for $L^2$ boundedness, using sharp examples and weighted norm estimates.
• To show that stronger conditions like $S_{2+\varepsilon} + T_{2+\varepsilon}$ are not necessary, and to propose a conjecture involving Young functions for a full characterization.

Proposed method

• Introduces two joint mean oscillation norms: $S_p(b_1,b_2) = \sup_Q \left( \frac{1}{|Q|}\int_Q |b_1 - \langle b_1\rangle_Q|^p \right)^{1/p} \left( \frac{1}{|Q|}\int_Q |b_2 - \langle b_2\rangle_Q|^p \right)^{1/p}$ and $T_p(b_1,b_2) = \sup_Q \left( \frac{1}{|Q|}\int_Q |b_1 - \langle b_1\rangle_Q|^p |b_2 - \langle b_2\rangle_Q|^p \right)^{1/p}$.
• Establishes two-sided estimates: $C_d(S_2(b_1,b_2) + T_2(b_1,b_2)) \lesssim \|[b_2,[b_1,T]]\|_{L^2 \to L^2} \lesssim C_{T,\varepsilon}(S_{2+\varepsilon}(b_1,b_2) + T_{2+\varepsilon}(b_1,b_2))$ for Calderón-Zygmund operators.
• Uses the sparse domination principle from Lerner [9] to control the commutator via dyadic sparse operators, enabling sharp weighted estimates.
• Constructs explicit counterexamples using odd functions $b_1, b_2$ with logarithmic growth to show that $S_{2+\varepsilon} + T_{2+\varepsilon}$ fails to characterize boundedness for any $\varepsilon > 0$, proving optimality of $S_2 + T_2$.
• Introduces a conjecture involving Young functions $A, B, C$ with $\bar{A}, \bar{B}, \bar{C} \in B_2$ such that $S_{A,B}(b_1,b_2) + T_C(b_1,b_2) < \infty$ characterizes $L^2$ boundedness.

Experimental results

Research questions

• RQ1 Is there a joint condition on $b_1, b_2$ weaker than $b_1, b_2 \in \operatorname{BMO}$ that characterizes the $L^2$ boundedness of the iterated commutator $[b_2, [b_1, T]]$?
• RQ2 Can the classical $BMO$ condition be replaced by a more precise joint condition that captures the true threshold for boundedness?
• RQ3 Why do $S_{2+\varepsilon} + T_{2+\varepsilon}$ conditions fail to characterize boundedness, even though they are sufficient?
• RQ4 Can the sparse domination principle be extended to handle iterated commutators under joint conditions rather than individual $BMO$ bounds?
• RQ5 Is there a full characterization of $L^2$ boundedness using Young functions that interpolate between $S_2 + T_2$ and $S_{2+\varepsilon} + T_{2+\varepsilon}$?

Key findings

• The $L^2$ operator norm of the iterated commutator $[b_2, [b_1, T]]$ is bounded below by a constant multiple of $S_2(b_1,b_2) + T_2(b_1,b_2)$, proving necessity of this joint condition.
• The $L^2$ norm is bounded above by $C_{T,\varepsilon}(S_{2+\varepsilon}(b_1,b_2) + T_{2+\varepsilon}(b_1,b_2))$ for any $\varepsilon > 0$, showing sufficiency under slightly stronger conditions.
• The authors construct a pair of functions $b_1, b_2$ such that $S_2(b_1,b_2) + T_2(b_1,b_2) < \infty$ but $S_{2+\varepsilon}(b_1,b_2) + T_{2+\varepsilon}(b_1,b_2) = \infty$ for all $\varepsilon > 0$, proving that $S_2 + T_2$ is strictly weaker than any $S_{2+\varepsilon} + T_{2+\varepsilon}$.
• The iterated commutator $[b_2, [b_1, H]]$ is bounded on $L^2$ but unbounded on $L^p$ for $p \neq 2$, showing that the $L^2$ boundedness is sharp and not extendable to other $L^p$ spaces.
• The paper proves that $S_{2+\varepsilon} + T_{2+\varepsilon}$ conditions are not sufficient to characterize $L^2$ boundedness, as demonstrated by a counterexample where the commutator is bounded but the norms diverge.
• The authors conjecture that $L^2$ boundedness is equivalent to the finiteness of $S_{A,B}(b_1,b_2) + T_C(b_1,b_2)$ for some Young functions $A,B,C$ with $\bar{A}, \bar{B}, \bar{C} \in B_2$, suggesting a full characterization via logarithmic bumps.