Quantitative weighted estimates for rough homogeneous singular integrals (Q522331)

The paper deals with general Calderón-Zygmund kernels in \(\mathbb R^d\) with standard size conditions \(|K(x,y)|\leq \frac{C_K}{|x-y|^d}\) and smoothness estimates \[ |K(x,y)-K(x',y)|+|K(y,x)-K(y,x')|\leq w\Big(\frac{|x-x'|}{|x-y|}\Big) \frac{1}{|x-y|^d} \] for \(|x-y|>2|x-x'|>0\) where \(w:[0,\infty)\to [0,\infty)\) is increasing, sub-additive and \(w(0)=0\), which is usually assumed to satisfy \(\|w\|_{Dini}=\int_0^1 \frac{w(t)}{t}<\infty.\) They also deal with the particular cases \(T_\Omega\) of rough homogeneous singular kernels \(K(x,y)=K(x-y)\) where \(K(x)=\frac{\Omega(x')}{|x|^d}\) with \(\Omega\in L^\infty(S^{d-1})\) and \(\int_{S^{d-1}} \Omega d\sigma=0\). The authors consider homogeneous singular kernels whose angular part is bounded but not necessarily continuous and obtain a bound for the norm in \(L^2(w)\) of the rough singular integral operator with such kernels in terms of the weighted \(A_2\)-constant \([w]_{A_2}\). As a consequence of their results a quadratic dependence in that quantity is achieved. This apparently it is not known to be sharp but it seems to be the best estimate until now. The proof of their main theorem relies on a decomposition of the operators into smooth pieces and an elaboration of the of some recent results by \textit{M. T. Lacey} [Isr. J. Math. 217, 181--195 (2017; Zbl 1368.42016)]. Denoting \([w]_{A_\infty}= \sup_Q \frac{1}{w(Q)}\int_Q M(1_Q w)(x)dx\) in the case \(p=\infty\), where \(Mf\) stands for the Hardy-Littlewood maximal function, and for \(1<p<\infty\) of \([w]_{A_p}=\sup_Q (\frac{1}{|Q|}\int_Q w)(\frac{1}{|Q|}\int_Q w^{1-p'})^{p-1}\), \(\{w\}_{A_p}=[w]_{A_p}^{1/p}\max\{[w]^{1/p'}_{A_\infty}, [w^{1-p'}]^{1/p}_{A_\infty}\} \) and \((w)_{A_p}=\max\{[w]_{A_\infty}, [w^{1-p'}]_{A_\infty}\}\), the authors prove the estimates \[ \|T_{\#}\|_{L^p(w)\to L^p(w)}\leq C_{d,p} (\|T\|_{L^2\to L^2}+ C_K+ \|w\|_{Dini})) \{w\}_{A_p} \] where \(T_{\#}f\) is the maximal truncation operator given by \(T_{\#}f(x)=\sup_{\varepsilon>0} |T_\varepsilon f(x)|\) and \(T_\varepsilon\) is the standard truncation of \(T\) and \[ \|T_\Omega\|_{L^p(w)\to L^p(w)}\leq C_{d,p} \|\Omega\|_{L^\infty}\{w\}_{A_p}(w)_{A_p}. \] From the last result one obtains \(\|T_\Omega\|_{L^2(w)\to L^2(w)}\leq C_{d} \|\Omega\|_{L^\infty}[w]_{A_2}^{2}.\)