Some obstacles in characterising the boundedness of bi-parameter singular integrals (Q907941)

From MaRDI portal
scientific article
Language Label Description Also known as
English
Some obstacles in characterising the boundedness of bi-parameter singular integrals
scientific article

    Statements

    Some obstacles in characterising the boundedness of bi-parameter singular integrals (English)
    0 references
    0 references
    0 references
    2 February 2016
    0 references
    The authors consider bi-parameter Calderón-Zygmund theory for the \(L^2\)-boundedness of bi-parameter singular integrals, namely, the full mixed bi-parameter paraproduct. It is known that the boundedness of the standard paraproduct is completely characterised by the product BMO condition, but the mixed one is behind the limitations of the product \(T1\) theory. In this paper, the authors show that the mixed bi-parameter paraproduct can be bounded, while being unconditionally unbounded. This implies that a BMO type condition has no chance of characterising the \(L^2\)-boundedness of the mixed paraproduct. Moreover, they construct an example showing that even the weak rectangular BMO condition is not necessary for the unconditional boundedness of the mixed paraproduct. More precisely, they define the dyadic operator by the bilinear form \[ P_{\lambda}(f,g) = \sum_{I,J}\lambda_{IJ} \big< f, h_I \otimes \frac{1_{J}}{|J|}\big> \big< g, \frac{1_{I}}{|I|} \otimes h_I \big>, \] where \(\lambda = (\lambda_{IJ})\) is an arbitrary sequence of reals indexed by dyadic rectangles \(I \times J\), and \(h_I \) denotes an \(L^2\)-normalized Haar function with zero mean. It is also pointed out that this operator can be transferred to continuous bi-parameter singular integrals via a representation theorem. For the main results they show the following. Let \(\mathcal M\) denote the space of all infinite matrices indexed by \(\mathbb N \times \mathbb N\), and let \(\Lambda\) denote the space of all sequences \((\lambda_{IJ})_{IJ}\) indexed by the dyadic rectangles \(I \times J \subset \mathbb R \times \mathbb R\). Define a mapping \(L : \mathcal M \rightarrow \Lambda\) by \[ L(A)_{IJ} = 2^{-(i+j)/2}A^{ij}, \;\text{if} \;I \times J \subset [0, 1)^2 \;\text{and} \;(|I|, |J|) = (2^{-i}, 2^{-j}). \] For dyadic rectangles \(I \times J \not \subset [0, 1)^2\) set \(L(A)_{IJ} = 0\). Theorem. The matrix \(A\) is bounded on \(\ell_{\mathbb N}^{2}\) if and only if the operator \(P_{\lambda}= P_{L(A)}\) induced by the sequence \(\lambda_{IJ} = L(A)_{IJ}\) is bounded on \(L^2({\mathbb R}^{2})\). In fact, \(\|A\|_{2 \rightarrow 2 } = \|P_{L(A)}\|_{2 \rightarrow 2 }\). As corollaries, it is stated that (1) the problem of characterising \(L^2\)-boundedness of the general paraproduct \(P_{\lambda}\) is at least as hard as the space-independent special case \(P_{L(A)}\), (2) \(P_{\lambda}\) is bounded, but not unconditionally bounded, and (3) the product BMO condition is not necessary for the unconditional boundedness of \(P_{\lambda}\).
    0 references
    0 references
    0 references
    bi-parameter singular integrals
    0 references
    paraproducts
    0 references
    unconditionality
    0 references
    0 references
    0 references
    0 references