Dyadic \(BMO\) on the bidisk (Q2583388)

There are several possible definitions of \(BMO\) on the bicircle, that in the case of one variable coincide. As a dual of the dyadic Hady space, it is the space \(BMO_{prod}^d\) of all \(\varphi\in L^2({\mathbf T}^2)\) such that \(\| \varphi\| _{prod}:=\sup_\Omega | \Omega| ^{-1/2}\| P_\Omega\varphi\| _2<\infty\), where \(P_\Omega \) denotes the projection \(P_\Omega f=\sum_{R\subset \Omega} (f,h_R)_2 h_R\) when \((h_R)\) is the Haar basis associated to the familly of all dyadic rectangles \(R\subset {\mathbf T}^2\). Similarly, \(BMO_{vect,p}^d\) is defined by the condition \(\sup_R | R| ^{-1/p}\| P_R\varphi\| _p<\infty\), and \(b\in BMO_{mult}^d\) means that \((P_R b)_R\) is a Haar multiplier on \(L^2({\mathbf T}^2)\). The paper under review explores the connection between these spaces, and shows how estimates of paraproducts turn out to be useful to obtain strict inclusions such as \(BMO_{mult}^d\subsetneq \bigcap_{p\geq 1} BMO_{prod}^d\), and to prove that \(BMO_{prod}^d=BMO_{mult}^d\).