Dyadic structure theorems for multiparameter function spaces (Q888919)

In this paper, the authors prove that the multiparameter (product) space \(\mathrm{BMO}\) of functions of bounded mean oscillation is an intersection of finitely many dyadic product \(\mathrm{BMO}\) spaces with equivalent norms. They also show that the corresponding intersection results hold for the space VMO of functions of vanishing mean oscillation, for \(A_p\) weights, for reverse-Hölder weights and for doubling weights. The authors prove the equivalences of several definitions of VMO, in the continuous, dyadic, one-parameter and product cases. Moreover, for any product \(A_\infty\) weight \(\omega\), the authors show that the weighted product Hardy space \(H_\omega^1(\mathbb{R}\otimes \mathbb{R})\) is the sum of finitely many translates of dyadic weighted Hardy spaces, and the weighted strong maximal function \(M_{s,\omega}\) is pointwise comparable to the sum of finitely many dyadic weighted strong maximal functions, for each doubling weight \(\omega\). Their results hold in both the compact and non-compact cases. To be precise, a function \(f\in L^1_{\mathrm{loc}}(\mathbb{R}\otimes \mathbb{R})\) is said to belong to the product BMO space \(\mathrm{BMO}(\mathbb{R}\otimes \mathbb{R})\) if there exists a positive constant \(C\) such that, for any open set \(\Omega\subset\mathbb{R}\otimes\mathbb{R}\) with finite measure, \[ \int\int_{T(\Omega)}|f\ast\psi_{y_1}\psi_{y_2}(t_1,\,t_2)|^2\, \frac{dt_1\,dy_1\,dt_2\,dy_2}{y_1y_2}\leq C|\Omega|, \] where \(T(\Omega):=\{(t_1,\,y_1,\,t_2,\,y_2):\;t_1,\,t_2\in\mathbb{R},\, y_1,\,y_2\in(0,\,\infty),\,(t_1-y_1,\,t_1+y_1)\times (t_2-y_2,\,t_2+y_2)\subset\Omega\}\) is the Carleson tent on \(\Omega\) and \(\psi_{y_1}\psi_{y_2}(t_1,\,t_2):=y_1^{-1}y_2^{-1}\psi(t_1/y_1)\psi(t_2/y_2)\) with \(\psi\in C_c^\infty(\mathbb{R})\) satisfying \(\int_{\mathbb{R}}\psi(t)\,dt=0\). Let \(\mathcal{D}\) be the grid of dyadic integrals in \(\mathbb{R}\) and \(\mathcal{D}^\delta\) be the translate of \(\mathcal{D}\) for \(\delta\in\mathbb{R}\). A function \(f\in L^1_{\mathrm{loc}}(\mathbb{R}\otimes \mathbb{R})\) is said to belong to the dyadic product BMO space \(\mathrm{BMO}_{d,d}(\mathbb{R}\otimes \mathbb{R})\) if there exists a positive constant \(C\) such that, for any open set \(\Omega\subset\mathbb{R}\otimes\mathbb{R}\) with finite measure, \[ \sum_{R:=I\times J\in \mathcal{D}\times\mathcal{D},\,R\subset\Omega}(f,\,h_R)^2\leq C|\Omega|, \] where \(h_R:=h_I\times h_J\), and \(h_I\) and \(h_J\) are the Haar functions on the intervals \(I,\,J\in\mathcal{D}\). \(\mathrm{BMO}_{d,\delta}(\mathbb{R}\otimes \mathbb{R})\), \(\mathrm{BMO}_{\delta,d}(\mathbb{R}\otimes \mathbb{R})\), and \(\mathrm{BMO}_{\delta,\delta}(\mathbb{R}\otimes \mathbb{R})\) are defined similarly. They differ only in which of the dyadic grids \(\mathcal{D}\) and \(\mathcal{D}^\delta\) is used in each variable. In this paper, the authors prove the following dyadic structure theorems: \[ \mathrm{BMO}(\mathbb{R}\otimes\mathbb{R}) =\mathrm{BMO}_{d,d}(\mathbb{R}\otimes\mathbb{R}) \cap \mathrm{BMO}_{d,\delta}(\mathbb{R}\otimes\mathbb{R}) \cap \mathrm{BMO}_{\delta,d}(\mathbb{R}\otimes\mathbb{R}) \cap \mathrm{BMO}_{\delta,\delta}(\mathbb{R}\otimes\mathbb{R}), \] with equivalent norms, for each \(\delta\in\mathbb{R}\) that is far from dyadic rationals. If \(\omega\) is a product \(A_\infty\) weight, then it holds true that \[ H^1_{\omega}(\mathbb{R}\otimes\mathbb{R})=H^1_{d,d,\omega}(\mathbb{R}\otimes\mathbb{R}) +H^1_{d,\delta,\omega}(\mathbb{R}\otimes\mathbb{R}) +H^1_{\delta,d,\omega}(\mathbb{R}\otimes\mathbb{R}) +H^1_{\delta,\delta,\omega}(\mathbb{R}\otimes\mathbb{R}), \] with equivalent norms, for each \(\delta\in\mathbb{R}\) that is far from dyadic rationals. If \(\omega\) is a product doubling weight, then it holds true that \[ M_{s,\omega}(f)\sim M_s^{d,d,\omega}(f)+M_s^{d,\delta,\omega}(f) +M_s^{\delta,d,\omega}(f)+M_s^{\delta,\delta,\omega}(f), \] with implicit constants independent of \(f\in L_{\mathrm{loc}}^1(\mathbb{R}\otimes\mathbb{R})\), for each \(\delta\in\mathbb{R}\) that is far from dyadic rationals.