LOGARITHMIC GROWTH FOR MATRIX MARTINGALE TRANSFORMS

From MaRDI portal





If \(W\) is a positive definite \(n\times n\) matrix of functions on the unit circle \(\mathbb{T}\), then \(L^2(W)= L^2(\mathbb{T}, W(t) dt)\) represents the space of \(\mathbb{C}^n\)-valued functions on \(\mathbb{T}\) with the norm \[ \|f\|_{2,W}^2= \int_{\mathbb{T}}(W(t) f(t),f(t))_n dt. \] If \({\mathcal D}\) denotes the class of dyadic intervals of \(\mathbb{T}\) and \({\mathcal D}^0={\mathcal D}\cup\mathbb{T}\), then the martingale transform \(T_\sigma\) is defined by \(T_\sigma(f)= \sum\{\sigma_j(f, h_j) h_j\), \(j\in{\mathcal D}^0\}\), where \(\sigma_j= \pm 1\) for \(j\in{\mathcal D}^0\). The expression \(\langle f\rangle_J\) denotes \(\text{meas}(J)^{-1}\int_J|f(t)|dt\). In the main results of this paper, the authors show that there are constants \(0< a\), \(A< \infty\), such that for each integer \(n\geq 1\), there is an \(n\times n\) matrix \(W_n\) of functions on \(\mathbb{T}\) with \[ \|W_n\|_{A2,d}= \sup\{\|(\langle W_n\rangle^{1/2}_J\langle W^{-1}_n\rangle^{1/2}_J)\|\leq A, J\in{\mathcal D}\} \] but \(\sup\{\|T_\sigma\|_{2,W_n\to 2,W_n}\}>a\log n\). Reference is made in the introduction of the paper to the results of \textit{R. Hunt}, \textit{B. Muckenhoupt} and \textit{R. Wheeden} [Trans. Am. Math. Soc. 176, 227-251 (1973; Zbl 0262.44004)], in which it is shown that if, for some finite constant \(c\), \(\langle w\rangle_J\langle w^{-1}\rangle_J\leq c\), then \(w\) is a weight of \(L^2\)-norm estimates.











This page was built for publication: LOGARITHMIC GROWTH FOR MATRIX MARTINGALE TRANSFORMS

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q3151072)