Haar multipliers meet Bellman functions (Q2655931)

Let \(\mathcal D\)~be the tree of dyadic intervals on~\(\mathbb R\). Let \(\sigma\)~be a dyadic doubling measure on~\(\mathbb R\), in the sense that \(\sigma(\tilde I)/\sigma(I)\) admits a least upper bound \(D(\sigma)\) for~\(I\in\mathcal D\), where \(\tilde I\)~is the parent interval of \(I\). Let \(\{h_I^\sigma\}_{I\in\mathcal D}\) be the weighted Haar system for~\(L^2(d\sigma)\). The central theorem of this article states that if a positive function~\(v\) admits a least upper bound \([v]_{A_2^d(d\sigma)}\) of \((\sigma(I)^{-1}\int_Ivd\sigma)(\sigma(I)^{-1}\int_Iv^{-1}d\sigma)\) for~\(I\in\mathcal D\), then the Haar multiplier given by \(h_I^\sigma\mapsto(v/(\sigma(I)^{-1}\int_Ivd\sigma))^{1/2}h_I^\sigma\) extends to a bounded linear operator \((T_v^\sigma)^{1/2}\) on~\(L^2(d\sigma)\) and \(\|(T_v^\sigma)^{1/2}\|\leq CD(\sigma)^{3/2}[v]_{A_2^d(d\sigma)}^{1/2}\). As corollaries, the author gets the bound \(\|(T_v^\sigma)^{-1/2}\|\leq CD(\sigma)^{3/2}[v]_{A_2^d(d\sigma)}\) and also direct and inverse estimates for the \(\sigma\)-dyadic square function~\(S_\sigma^d\). Her strategy is to reduce this theorem to a weighted Carleson Embedding Lemma estimate, which is established along the lines of an argument by \textit{S. Petermichl} and \textit{S. Pott} [Trans. Am. Math. Soc. 354, No. 4, 1699--1703 (2002; Zbl 1005.42002)], through a Sawyer type estimate and a Bellman function argument. The simpler case when \(d\sigma\)~is Lebesgue measure is treated in detail in the beginning of the article; in this case, the author proves also that \(\|T_v^\sigma\|\leq CD(v)[v]_{RH_2^d}^2\) if \(v\) admits a least upper bound~\([v]_{RH_2^d}\) of \((|I|^{-1}\int_Iw^2)^{1/2}/(|I|^{-1}\int_Iw)\) for~\(I\in\mathcal D\).