Optimal factorization of Muckenhoupt weights

From MaRDI portal





A non-negative weight function \(w\) on \({\mathbb R}^n\) is in \(A_p\) (or \(w\in A_p\)), \(p>1\), if for some constant \(C>0\): \[ |Q|^{-1}\Biggl(\int_Qw(x) dx \Biggr) \Biggl(\int_Qw^{-1/(p-1)}(y) dy \Biggr)^{p-1}\leq C \] for all cubes \(Q\) in \({\mathbb R}^n\) whose sides are parallel to the coordinates axes. Similarly, \(w\in A_1\) whenever \[ |Q|^{-1}\int_Qw(x) dx\leq C \inf_{y\in Q}w(y). \] The least constant \(C\) in each of the above inequalities, denoted by \(A_p(w)\), is called the \(A_p\) bound of \(w\). For \(p>1\) it is well known that \[ w=uv^{1-p}\in A_p\quad \text{whenever } u, v\in A_1 \tag \(*\) \] and moreover \(A_p(w)\leq A_1(u) [A_1(v)]^{1-p}\). A celebrated Peter Jones's factorization theorem asserts the converse, in the sense that each weight \(w\in A_p\), \(p>1\), can be written as a product of weights as in (*) for some weights \(u,v\in A_1\). The reiteration of this scheme has found many applications in analysis, as in the proof of duality of Hardy space and bounded mean oscillation space, and in extrapolation theorem for operators on weighted Lebesgue spaces. The aim in this paper is to sharpen the above Peter Jones's theorem for weights with bounds near \(1\). Precisely, a real \(\varepsilon_0>0\) exists such that, if \(w\in A_p\) with \(A_p(w)=1+\varepsilon<1+\varepsilon_0\) then the factorization \((*)\) holds with \(A_1(u),A_1(v)\leq 1+C\sqrt{\varepsilon}\) for some constant \(C>0\) depending only on the dimension \(n\) and the index \(p\). To establish such a result, the author first supplements the original argument of Peter Jones in the dyadic model with some sharp estimates he has performed in his thesis. The passage to the general case is then done by adapting the averaging method of Garnett Jones.











This page was built for publication: Optimal factorization of Muckenhoupt weights

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q4505919)