Off-diagonal extrapolation on mixed variable Lebesgue spaces and its applications to strong fractional maximal operators (Q2662074)

The paper under review gives a slight extension of the Rubio de Francia extrapolation theorem that covers mixed norm \(L^p\) spaces with variable exponents \(L^{p(.)}(L^{q(.)})\). The case \(p(.)=q(.)\) was treated in [\textit{D. Cruz-Uribe} et al., Ann. Acad. Sci. Fenn., Math. 31, No. 1, 239--264 (2006; Zbl 1100.42012)], and the proof of the mixed norm case is similar. One can use duality, as in [\textit{K.-P. Ho}, Ann. Univ. Ferrara, Sez. VII, Sci. Mat. 62, No. 2, 275--291 (2016; Zbl 1354.42031)], to reduce the problem to estimating quantities of the form \[ \int \limits _{\mathbb{R}^{n}\times \mathbb{R}^{m}} |f(x,y)|^{q_{0}} \mathcal{R}_{1}h_{1}(x) \mathcal{R}_{2}h_{2}(y)dxdy, \] where \(h_{1}\), \(h_{2}\) are arbitrary elements in the unit balls of variable \(L^p\) spaces, and \(\mathcal{R}_{1}\), \(\mathcal{R}_{2}\) are operators involving normalised sums of iterates of the Hardy-Littlewood maximal function acting on these variable \(L^p\) spaces. The dualised quantities are then estimated using a Rubio de Francia extrapolation argument for standard (i.e. fixed exponent) \(L^{p}\) spaces. Here this needs to be done for weights in the product domain, but this can be done (in Theorem 2.1) using the usual proof given in [\textit{D. V. Cruz-Uribe} et al., Weights, extrapolation and the theory of Rubio de Francia. Basel: Birkhäuser (2011; Zbl 1234.46003)]. Using this extrapolation result, the author can then bootstrap the boundedness of certain sublinear operators from the standard weighted setting to the setting of mixed norms \(L^{p}\) spaces with variable exponents. This is done for strong fractional maximal operators in Section 3 (using a control by the strong maximal function), and for Littlewood-Paley square functions in Section 4.