Non-homogeneous \( T1\) theorem for bi-parameter singular integrals (Q2453573)

This is the first paper which deals with ``bi-parameter'' and ``non-homogeneous'' at the same time, establishing a \(T1\) theorem on the product domain \(\mathbb{R}^n \times\mathbb{R}^m\) equipped with a product measure \(\mu=\mu_n\times\mu_m\) with \(\mu_n\) and \(\mu_m\) satisfying only the \textit{upper doubling condition} (first introduced by \textit{T. Hytönen} [Publ. Mat., Barc. 54, No. 2, 485--504 (2010; Zbl 1246.30087)]). After the first \(T1\) theorem by \textit{G. David} and \textit{J.-L. Journé} [Ann. Math. (2) 120, 371--397 (1984; Zbl 0567.47025)] and the \(Tb\) theorem by \textit{G. David} et al. [Rev. Mat. Iberoam. 1, No. 4, 1--56 (1985; Zbl 0604.42014)], the first \(T1\) type theorem for product spaces was proved by \textit{J.-L. Journé} [Rev. Mat. Iberoam. 1, No. 3, 55--91 (1985; Zbl 0634.42015)] where he formulated the concept of vector-valued Calderón-Zygmund operators (the so-called Journé class). Recently \textit{S. Pott} and \textit{P. Villarroya} [``A T(1) theorem on product spaces'', Preprint (2011), \url{arXiv:1105.2516}] introduced an alternative framework of product \(T1\) theorems bypassing the vector-valued assumptions. In the current paper, the authors formulate the class of product type Calderón-Zygmund operators \(T\) by assuming the full standard kernel estimates (size and smoothness) as well as the vector-valued kernel estimates (size and smoothness for the variables on \(\mathbb {R}^n\) and on \(\mathbb {R}^m\)) on \(T\), \(T^*\) (the adjoint), and \(T_1\) and \(T_1^*\) (partial adjoints), in terms of the dominating functions \(\lambda_n\) and \(\lambda_m\) of \(\mu_n\) and \(\mu_m\), respectively. They also consider Journé's lemma, product BMO, product \(H^1\)-BMO type dual inequalities and multi-parameter para-products with respect to the non-doubling measures, which extends all the previous related results to a more general non-homogeneous setting. The main techniques they apply are the remarkable Nazarov-Treil-Volberg methods, using random dyadic cubes to analyze the good parts and bad parts of functions.