On two-weight norm estimates for multilinear fractional maximal function (Q1743693)

In the paper [Stud. Math. 75, 1--11 (1982; Zbl 0508.42023)], \textit{E. T. Sawyer} proved a general criterium for the boundedness of the fractional maximal function frow \(L^p(\sigma)\) to \(L^q(\omega)\), \( 1< p \leq q<\infty\). In the present work, it is proved a multilinear analogue by considering the multilinear fractional maximal functions associated to cubes in \(\mathcal Q\) with sides parallel to the coordinate axes, defined for \(x\in \mathbb R^n\) and \(\vec{f}=(f_1,\cdots, f_m)\) by \[ \mathcal M_\alpha \vec{f}(x):=\sup_{Q\in \mathcal Q}| Q|^{\alpha/n} \prod_{i=1}^m \frac{\chi_Q (x)}{|Q|} \int_Q |f_i(y)| dy, \] where the \(f_i\)'s are measurable functions and \(0\leq \alpha< mn\). The main results consist in providing sufficient conditions for \({\mathcal M}_{\alpha}\) to be bounded from \(L^{p_1}(\sigma_1)\times \cdots \times L^{p_m}(\sigma_m) \) to \(L^q(\omega)\), the conditions obtained on the weights are close to the \(A_p\) characterization of Muckenhoupt. Once the problem is reduced to the corresponding dyadic maximal function, one of the main tools used in the proofs is the extension of Carlesons embedding theorem and its multilinear analogue.