Linear bounds for Calderón-Zygmund operators with even kernel on UMD spaces (Q2253140): Difference between revisions

The paper under review studies the dependence of the norm of UMD-valued Calderón-Zygmund operators on the UMD constant. For a general Calderón-Zygmund operator on \(\mathbb{R}\) and a general UMD space, the best known result is by \textit{T. Figiel} [Lond. Math. Soc. Lect. Note Ser. 158, 95--110 (1991; Zbl 0746.47026)], which gives a quadratic dependence on the UMD constant. For specific operators, such as the Beurling-Ahlfors transform, or specific UMD spaces, a linear bound has been shown. In the article under review, the authors consider a general UMD space \(X\) and a Calderón-Zygmund operator \(T\) with even kernel \(K\) and smoothness condition \[ |K(x,y)-K(x,z)| +|K(y,x)-K(z,x)| \leq C \frac{|y-z|}{|x-y|^{1+\delta}} \] for \(|x-y|>2|y-z|\). They show that if \(\delta>1/2\) and \(T\) satisfies a weak boundedness property and the cancellation conditions \(T(1)=T^\ast(1)=0\), then the operator norm of \(T\) in \(L^p(\mathbb{R};X)\), \(1<p<\infty\), depends linearly on the UMD\({}_p\) constant. The proof uses Hytönen's representation of Calderón-Zygmund operators in terms of dyadic Haar shift operators [\textit{T. P. Hytönen}, Ann. Math. (2) 175, No. 3, 1473--1506 (2012; Zbl 1250.42036)] and Bellman function techniques. A decomposition of dyadic Haar shifts into ``slices'' allows a reduction to martingale transforms, for which the linear estimate follows by definition of the UMD constant.