Weighted estimates for dyadic paraproducts and \(t\)-Haar multipliers with complexity (\(m,n\)) (Q2637434): Difference between revisions

For positive integer \(n\), the Euclidean \(n\)-dimensional space is denoted by \(\mathbb{R}^n\), and the Lebesgue measure of a subset \(E\subset\mathbb{R}^n\) is denoted by \(|E|\). If \(w\) is a nonnegative weight function and \(v\) is a nonnegative function, then the measure \(\omega(.)\) and averaging expression, \(m^v_E\), are defined by \[ \omega(E)= \int_W w(x)\,dx,\quad m^v_E(f)= \Biggl(\int_E v(x) f(x)\,dx\Biggr)/\Biggl(\int_E v(x)\,dx\Biggr). \] The \(L^p\)-space \(L^p(\omega)\) is defined for \(p\geq 1\), by \(L^p(w)= \{[f]:\|[f]\|_{p, \omega}=\| f\|_{p,\omega}< \infty\}\), where \([f]\) is the equivalence class of functions which differ from \(f\) on sets of measure \(0\), and \[ \| f\|_{p,\omega}= \Biggl(\int_{\mathbb{R}^n}| f(x)|^p w(x)\,dx\Biggr)^{1/p}. \] If, for integer \(K\), \({\mathcal D}_K= \{k2^{-K},(k+ 1)2^{-K}\), \(k= 0,1,2,\dots\}\), then the class of diadic intervals \(\bigcup\{{\mathcal D}_K\}\) is denoted by \({\mathcal D}\), and if \(L\) is a dyadic subinterval, then for integer \(J\), \({\mathcal D}_J(L)\) denotes \(\{I\subseteq L:|I|= 2^{-J}|L|\}\). The weight function \(w\) is said to satisfy the diadic Muckenhoupt condition, or the reverse Hölder \(RH^d_p\) condition, or the \(C^d_p\) condition if, respectively, (i) \([w]_{H,p,d}= \sup\{m_1(w)^{-1},\, I\in{\mathcal D}<\infty\), or (ii) \([w]_{H,p,d} \sup\{m_1(w^p)^{1/p} m_1(w)^{-1},\, I\in{\mathcal D}\}< \infty\), or (iii) \([w]_{C,p,d} = \sup\{m_1(w^s) m_1(w)^{-s},\, I\in{\mathcal D}\}<\infty\). The diadic paraproduct \(\pi^{m,n}_b\), the Haar shift operator \(S^{m,n}\) and the \(t\)-Haar muliplier of complexity \((m,n)\) are defined by \[ \pi^{m,n}_b(f)(x)= \sum_{{\mathcal D}} \sum_{{\mathcal D}_n\times{\mathcal D}_m} c^L_{I,J} m_1(f) (\int bh_Jdx)\, h_J(x); \] \[ S^{m,n}(f)(x)= \sum_{{\mathcal D}} \sum_{D_n\times D_m} c^L_{I,J}(\int fh_I dx)\,h_J(x), \] \[ T^{m,n}_{t,w}(f)(x)= \sum_{{\mathcal D}} \sum_{D_n(L)\times D_m(L)} c^L_{I,J}(w(x)/m_L(w))^t(\int fh_I dx)\,h_J(x), \] where the terms of the summation are defined in terms of the function \(v\). The main results derived in this paper include estimates which have the form (1) \(\|\pi^{m,n}_b(f)\|_{2, w}\leq C(m+ n+ 2)^5[w]_{A,2,d}\| b\|_{\text{BMO}}\| g\|_{2,1/w}\| f\|_{2,w}\), (2) \(\| T^{m,n}_{t,w}(f)\|_2\leq C(m+ n+ 2)^3[w]_{C,2t,d}^{1/2} [w^{2t}]^{1/2}_{A, 2,d}\| f\|_2\| g\|_2\). References to earlier papers and some special cases are contained in the paper. Results which depend on \(L^p\)-estimates for maximal functions defined in terms of \(\sup m^v_Q:Q\subset \mathbb{R}^n\), may require conditions on the functions.