Weighted weak-type \((1, 1)\) estimates via Rubio de Francia extrapolation (Q2355435)

Partly based on quotations from the authors' abstract: The classical extrapolation theorem of \textit{J. L. Rubio de Francia} [Am. J. Math. 106, 533--547 (1984: Zbl 0558.42012)] asserts that if an operator \(T : L^{p_0}(u) \rightarrow L^{p_0, \infty}(u)\) is bounded for some \(p_0 >1\) and every \(u \in A_{p_0}\), then for every \(1<p<\infty\) and every \(u \in A_{p}\), \(T : L^{p}(u) \rightarrow L^{p, \infty}(u)\) is bounded. However, there are examples showing that it is not possible to extrapolate to the end-point \(p=1\). The authors prove that there exists a class of weights, slightly larger than \(A_p\), with the following property: if an operator \(T : L^{p_0}(u) \rightarrow L^{p_0, \infty}(u)\) is bounded for some \(p_0 >1\) and every \(u\) in this class, then for every \(u \in A_{1}\), \(T\) is of restricted weak-type \((1,1)\). They also consider \(L \log L\) type estimates. They define \[ \hat{A}_p = \big\{ u : \exists f \in L_{loc}^1 \;\text{and} \;\exists u_1 \in A_1 \;\text{such that} \;u = (Mf)^{1-p}u_1 \big\}. \] They prove the following. Let \(T\) be an operator satisfying that, for some \(p_0>1\) and every \(u \in \hat{A}_{p_0}\), \[ T: L^{p_0,1}(u) \rightarrow L^{p_0, \infty}(u) \] is bounded, with constant less than or equal to \(\varphi_{p_0}(\| u \|_{ \hat{A}_{p_0}})\), with \(\varphi_{p_0}\) an increasing function on \((0, \infty)\). Then \(T\) is of restricted weak-type \((1,1)\), that is, for every measurable set \(E\), \[ \| T \chi_E \|_{L^{1, \infty}(u)} \leq C \| u \|_{A_1}^{1 - 1/p_0} \varphi_{p_0}( \| u \|_{A_1}^{1/p_0}) u(E). \] Next they consider the following problem: when does the implication \[ \| T \chi_E \|_{L^{1,\infty}(u)} \leq C u(E) \;\Longrightarrow \;T:L^1(u) \rightarrow L^{1, \infty}(u) \] hold for every \(u \in A_1\). It is known that this is not true in general. However, \textit{M. J. Carro} [J. Lond. Math. Soc., II. Ser. 70, No. 3, 750--762 (2004; Zbl 1071.47012)] proved that for a quite big class of operators the above implication is true in the case \(u=1\). The authors also prove that this implication is true when \(u \in A_1\) for a Calderón-Zygmund maximal operator, the \(g\)-function and the intrinsic square function.