Paperlane

no. 17 · Wednesday 18 November 2026

A counterexample to the one-supremum conjecture and sharp logarithmic weighted bounds for Calderón–Zygmund operators

Xi Cen

math.CA

Abstract

We give a negative answer to the one-supremum conjecture of Hytönen and Lacey~\cite[Remark~1.3]{HL}, in the strong-type form stated by Lerner and Moen~\cite[Conjecture~1.3, formula~(9)]{LM}. The conjecture asks whether the mixed $A_p$--$A_\infty$ characteristics $K_p(w)$ and $K_{p'}(σ)$, with both factors evaluated on the same cube, control Calderón–Zygmund operators without a logarithmic loss. For every $1<p<\infty$, we construct explicit weights for which $K_p(w)+K_{p'}(σ)\eqsim_p m$, but one fixed Hilbert transform has weighted norm at least $m(1+\log m)^{γ_p}$, where $γ_p=\max\{1/p,1/p'\}$. Thus the proposed bound fails even for a fixed singular integral. The counterexample also extends to a fixed Riesz transform in each dimension $n\ge2$. We further establish a logarithmically corrected bound for Calderón–Zygmund operators, with powers $1/p'$ and $1/p$ on the respective one-supremum terms. Together, the lower and upper bounds identify $γ_p$ as the optimal common logarithmic exponent and give the sharp worst-case growth $(1+\log A)^{γ_p}$ under the constraint $[w]_{A_p}\le A$. The results show why information from any one cube cannot fully capture the accumulation of a singular integral across many scales.

comments
Submitted; Comments are welcome!
fields
Classical Analysis and ODEs, Functional Analysis
pdf ↗ page arXiv:2611.22097 ↗