Sharp Lorentz space estimates for rough operators (Q5947137): Difference between revisions

Let \(K(x)\) be a convolution kernel on the Euclidean space \(\mathbb R^n\) and assume that \(K\) is homogeneous of degree \(-n\) and the restriction \(\Omega\) to the unit sphere is integrable and has mean zero. The singular integral \(T_\Omega\) is defined by \[ T_\Omega f(x)={\text{p.v.}}\int\Omega(y/|y|)|y|^{-n}f(x-y) dy. \] If \(\Omega\) satisfies an appropriate \(L^1\) Dini condition, then the standard Calderón-Zygmund theory shows that \(T_\Omega\) is bounded on all \(L^p\) spaces \((1<p<\infty)\), is of weak type \((1,1)\), and maps \(H^1\) to \(L^1\). If no regularlity is assumed, but \(\Omega\) is \(L\log L\), it is bounded on all \(L^p\) spaces \((1<p<\infty)\) (Calderón and Zygmund), and is of weak type \((1, 1)\) [\textit{A. Seeger}, J. Am. Math. Soc. 9, No.1, 95-105 (1996; Zbl 0858.42008)]. The authors show the following: If \(\Omega\) is \(L\log^2 L\), then \(T_\Omega\) maps \(H^1\) to the Hardy-Lorentz space \(H^{1,2}\) and also to the Lorentz space \(L^{1,2}\). The index 2 is optimal as was shown by M. Christ. They also consider a maximal operator related to \(T_\Omega\). Similar considerations are done to the Hilbert transform \(Hf\) and its maximal function along the curve \((t, |t|^m)\) \((m>1)\) in \(\mathbb R^2\), worked with the product Hardy space \(H_{\text{prod}}^1\) and the product Hardy-Lorentz space \(H_{\text{prod}}^{1, 2}\).