Weighted estimates for maximal bilinear rough singular integrals via sparse dominations (Q2075479)

This paper is to show that the maximal truncated bilinear singular integrals with rough kernel \(K(x_1, x_2)=\Omega(x/\left|x\right|)\left|x\right|^{-2n}\) satisfy a sparse bound by \((p, p, p)\)-averages for all \(p > 1\), where \(\Omega\in L^\infty(\mathbb{S}^{2n-1})\) and \(\int_{\mathbb{S}^{2n-1}}\Omega=0\). As consequence, some quantitative weighted estimates for these rough singular integrals are obtained. Besides, a pointwise sparse domination for commutators of bilinear rough singular integrals, which can be used to establish some weighted inequalities, is also established. In the bilinear case, \textit{L. Grafakos} et al. [Adv. Math. 326, 54--78 (2018; Zbl 1402.42016)] proved that the rough bilinear singular integral, which was introduced by \textit{R. R. Coifman} and \textit{Y. Meyer} [Trans. Am. Math. Soc. 212, 315--331 (1975; Zbl 0324.44005)], is bounded from \(L^{p_1}(\mathbb {R}^n) \times L^{p_2}(\mathbb {R}^n)\) to \(L^{p}(\mathbb {R}^n)\) when \(\Omega\in L^\infty(\mathbb{S}^{2n-1})\). Subsequently, they also gave some more detalied results. For the rough maximal bilinear singular integral \(T_\Omega^\ast\), when \(\Omega\in L^\infty(\mathbb{S}^{2n-1})\), based heavily on a tensor-type wavelet decomposition, \textit{E. Buriánková} and \textit{P. Honzík} [Collect. Math. 70, No. 3, 431--446 (2019; Zbl 1421.42008)] gave the following result. Theorem A: Let \(\Omega\in L^\infty(\mathbb{S}^{2n-1})\) and \(\int_{\mathbb{S}^{2n-1}} \Omega =0\). Then \(\left\|T_\Omega^\ast\right\|_{L^{p_1}(\mathbb{R}^n)\times L^{p_2}(\mathbb{R}^n) \to L^{p}(\mathbb{R}^n)} \le C\left\|\Omega\right\|_\infty\) whenever \(1<p_1\), \(p_2<\infty\) and \(\frac{1}{p}=\frac{1}{p_1}+\frac{1}{p_2}\). This paper is to obtain sparse domination for the maximal truncated bilinear integrals with rough kernels. Theorem 1.1: Let \(p>1\) and \(T_\Omega^\ast\) be the maximal truncated rough bilinear singular integral with \(\Omega\in L^\infty(\mathbb{S}^{2n-1})\) and \(\int_{\mathbb{S}^{2n-1}} \Omega =0\). Then there is a constant \(C=C_{p,n}\) such that \[ \left|\langle T_\Omega^\ast(f_1, f_2), f_3 \rangle \right| \le C\left\|\Omega\right\|_{L^\infty(\mathbb{S}^{2n-1})} \sup_{\mathcal{S}} \text{PSF}_{\mathcal{S}}^{(p,p,p)} (f_1, f_2, f_3). \] As an application of the sparse domination, some weighted estimates for \(T_\Omega^\ast\) are obtained. Definition 1.1: Let \(A_{\vec{p},\vec{r}}\) be a weight class, and \(m\ge 2\) be an integer, \(\vec{p}=(p_1, \dots, p_m)\) with \(1<p_1, \dots, p_m<\infty\) and \(\vec{r}=(r_1, \dots, r_{m+1})\) with \(1<r_1, \dots, r_{m}+1<\infty\). \(1/p=\sum_{k=1}^m 1/p_k\). Set \(w=\prod_{k=1}^m w_k^{p/p_k}\). We say that \(\vec{w}\in A_{\vec{p},\vec{r}}\) if \(0<w_i<\infty\), \(i=1,\dots,<m\) and \([w]_{A_{\vec{p},\vec{r}}}<\infty\) with \[ [\vec{w}]_{A_{\vec{p}, \vec{r}}}=\sup _{Q}\left(\frac{1}{|Q|} \int_{Q} w^{\frac{r_{m+1}^{\prime}}{r_{m+1}^{\prime}-p}}(x) \mathrm{d} x\right)^{1 / p-1 / r_{m+1}^{\prime}} \prod_{k=1}^{m}\left(\frac{1}{|Q|} \int_{Q} w_{k}^{-\frac{1}{\frac{p_k}{r_{k}}-1}}(x) \mathrm{d} x\right)^{1 / r_{k}-1 / p_{k}}. \] As corollaries of Theorem 1.1: Corollary 1.2: Suppose \(\vec{q}=\left(q_{1}, q_{2}\right)\), \(\vec{p}=\left(p_{1}, p_{2}, p_{3}\right)\) with \(\vec{p}\prec \vec{q}\) and \(p_i>1\), (\(i=1,2,3\)), \(q_3=q^\prime\). \([\vec{v}]_{\vec{q}, \vec{p}}\) is defined as above with \(v_{\vec{w}}=\prod_{k=1}^{m} v_{k}^{q / q_{k}}\) and \(\frac{1}{q}=\frac{1}{q_{1}}+\frac{1}{q_{2}}\). Then for \(\Omega \in L^{\infty}\left(\mathbb{S}^{2 n-1}\right)\) with vanishing integral, we have \[ \left\|T_{\Omega}^{\ast}(f, g)\right\|_{L^{q}\left(v_{\bar{w}}\right)} \leq C_{n \vec{p}, \vec{q}}[\vec{v}]_{A_{\vec{q} \vec{p}}}^{\max _{1 \leq i \leq 3} \frac{p_{i}}{q_{i}-p_{i}}}\|f\|_{L^{q 1}\left(v_{1}\right)}\|g\|_{L^{q 2\left(v_{2}\right)}} . \] Corollary 1.3: Suppose \(\Omega \in L^{\infty}\left(\mathbb{S}^{2n-1}\right)\) with vanishing integral. Then for \(\vec{w}=\left(w_{1}, w_{2}\right) \in A_{\vec{q}}\) with \(v_{\vec{w}}=\prod_{k=1}^{m} v_{k}^{q / q_{k}}\) and \(\frac{1}{q}=\frac{1}{q_{1}}+\frac{1}{q_{2}}\), there exists a constant \(C_n, q_1, q_2\) such that \[ \left\|T_{\Omega}^{\ast}\left(f_{1}, f_{2}\right)\right\|_{L^{q}\left(v_{\vec{w}}\right)} \leq C_{n, q_{1}, q_{2}}\left\|f_{1}\right\|_{L^{q_{1}}\left(w_{1}\right)} \left\|f_{2}\right\|_{L^{q_{2}}\left(w_{2}\right)}. \] The bilinear maximal singular integrals are defined by \[ T^{\ast}\left(f_{1}, f_{2}\right)(x)=\sup _{\delta>0}\left|\int_{\left|x-y_{1}\right|>\delta} \int_{\left|x-y_{2}\right|>\delta} K\left(y_{1}, y_{2}\right) \prod_{i=1}^{2} f_{i}\left(x-y_{i}\right) d y_{i}\right|. \] The commutators of bilinear maximal singular integrals are defined by \[ T_{\vec{b}}^{\ast}\left(f_{1}, f_{2}\right)(x)=\sup _{\delta>0}\left|\int_{\left|x-y_{1}\right|>\delta} \int_{\left|x-y_{2}\right|>\delta} K\left(y_{1}, y_{2}\right) \prod_{i=1}^{2}\left(b_{i}(x)-b_{i}\left(y_{i}\right)\right) f_{i}\left(x-y_{i}\right) d y_{i}\right|. \] Definition 5.1: For \(1 \leq r<\infty\), the kernel \(K\left(x, y_{1}, y_{2}\right)\) is called to satisfy the bilinear \(L^{r}\)-Hörmander condition if \[ K_{r}:=\sup _{Q} \sup _{x, z \in \frac{1}{2} Q} \sum_{k=1}^{\infty}\left|2^{k} Q\right|^{\frac{2}{r}}\left(\int_{\left(2^{k} Q\right)^{2} \backslash\left(2^{k-1} Q\right)^{2}}\left|K\left(x, y_{1}, y_{2}\right)-K\left(x^{\prime}, y_{1}, y_{2}\right)\right|^{r^{\prime}} d \vec{y}\right)^{\frac{1}{r^{\prime}}}<\infty. \] By a multilinear analog of the grand maximal truncated operator, the authors obtained sparse domination for the commutators of bilinear maximal singular integrals. Theorem 5.3: For \(1 \leq r<\infty\) , let the bilinear singular integral \(T^{\ast}\) be bounded from \(L^{r} \times L^{r} \rightarrow L^{r / 2, \infty}\) and the kernel satisfy the bilinear \(L^{r}\)-Hörmander condition. Then for any compactly supported function \(f_{i} \in L^{r}\left(\mathbb{R}^{n}\right)\), \(i=1,2 \), there exist \(3^{n}\) sparse families \( \mathcal{S}_{j}\) such that for a.e. \(x \in \mathbb{R}^{n}\) , it holds that \[ T_{\vec{b}}^{\ast}\left(f_{1}, f_{2}\right)(x) \leq C \sum_{j=1}^{3^{n}} \sum_{\tau \subset\{1,2\}}\left(\sum_{Q \in \mathcal{S}_{j}} \prod_{i \in \tau}\left\langle\left(b_{i}-b_{i, Q}\right) f_{i}\right\rangle_{r, Q} \prod_{j \in \tau_{m} \backslash \tau}\left\langle f_{j}\right\rangle_{r, Q}\left|b_{j}(x)-b_{j, Q}\right| \chi_{Q}\right) . \]