Limiting weak-type behaviors for Riesz transforms and maximal operators in Bessel setting (Q2316941)

In this paper, one deals with limiting weak-type estimates for the Riesz transforms associated to the Bessel operators on \(\mathbb{R}^{+}\) and for the fractional maximal operator in Bessel setting. For \(0<\lambda <\infty\), the Bessel operator \(\Delta_{\lambda}\) is defined by \[ \Delta_{\lambda}f(x) = -\frac{\mathrm{d}^{2}}{\mathrm{d}x^{2}}f(x)-\frac{2\lambda}{x}\frac{\mathrm{d}}{\mathrm{d}x} f(x),\quad f\in C^{2} (\mathbb{R}^{+}), \] and the Riesz transform \(R_{\Delta_{\lambda}}\) associated to \(\Delta_{\lambda}\) is \[ R_{\Delta_{\lambda}}f(x)=\partial_{x}(\Delta_{\lambda})^{-1/2} f(x). \] Writing \(\mathrm{d}m_{\lambda}(y)\) for the measure \(y^{2\lambda}\,dy\) it is known that \(R_{\Delta_{\lambda}}\) is bounded on \(L^{p}(\mathbb{R}^{+},\mathrm{d}m_{\lambda})\) for \(1 < p <\infty\) and from \(L^{1}(\mathbb{R}^{+},\mathrm{d}m_{\lambda})\) to \(L^{1,\infty}(\mathbb{R}^{+},\mathrm{d}m_{\lambda})\). We recall as well that for \(0\le \alpha<1\) and \(f\in L^{1}_{\text{loc}}(\mathbb{R}^{+},\mathrm{d}m_{\lambda})\), the fractional maximal operator \(M^{\alpha}_{\lambda}\) is defined by \[ M_{\lambda}^{\alpha}f(x)=\sup_{x,r\in \mathbb{R}^{+}}\frac{1}{m_{\lambda}(I(x,r))^{1-\alpha}} \int_{I(x,r)} |f(y)|\,\mathrm{d}m_{\lambda}(y), \] where \(I(x,r)=(x-r,x+r)\cap\mathbb{R}^{+}\). It is known that for \[ 1 < p < \frac{1}{\alpha}\quad\text{and}\quad \frac{1}{q}=\frac{1}{p}-\alpha, \] \(M^{\alpha}_{\lambda}\) is bounded from \(L^{p}(\mathbb{R}^{+},\mathrm{d}m_{\lambda})\) to \(L^{q}(\mathbb{R}^{+},\mathrm{d}m_{\lambda})\) and from \(L^{1}(\mathbb{R}^{+},\mathrm{d}m_{\lambda})\) to \(L^{1/1-\alpha,\infty}(\mathbb{R}^{+},\mathrm{d}m_{\lambda})\). The limiting weak-type estimates obtained here for the operators \(R_{\Delta_{\lambda}}\) and \(M^{\alpha}_{\lambda}\) are the following ones: \begin{itemize} \item Let \(f\in L^{1}(\mathbb{R}^{+},\mathrm{d}m_{\lambda})\) and \(f\ge 0\). Then \[ \begin{aligned} &\lim_{\gamma\to 0}\gamma m_{\lambda}\Bigl(\Bigl\{x\in\mathbb{R}^{+}: |R_{\Delta_{\lambda}}f(x)|>\gamma\Bigr\}\Bigr)=\frac{2\lambda}{\sqrt{\pi}(2\lambda+1)} \frac{\Gamma(\lambda)}{\Gamma(\lambda +\frac{1}{2})}\|f\|_{L^{1}(\mathbb{R}^{+},\mathrm{d}m_{\lambda})}; \\ &\lim_{\gamma\to 0}\gamma m_{\lambda}\Bigl(\Bigl\{x\in \mathbb{R}^{+}:|R_{\Delta_{\lambda}}f(x)-\mathscr{R}_{\Delta_{\lambda}}(x,0)\|f\|_{L^{1}(\mathbb{R}^{+},\mathrm{d}m_{\lambda})}|>\gamma\Bigr\}\Bigr)=0, \end{aligned} \] where \[ \mathscr{R}_{\Delta_{\lambda}}(x,0)=-\frac{2\lambda}{\pi} \int^{\tau}_{0}\frac{1}{x^{2\lambda+1}}(\sin\theta)^{2\lambda-1}\,d\theta. \] \item Let \(0\le \alpha<1\) and \(q=1/(1-\alpha)\). Then, for \(f\in L^{1}( \mathbb{R}^{+},\mathrm{d}m_{\lambda})\), \[ \begin{aligned} &\lim_{\gamma\to 0}\gamma^{q} m_{\lambda}\Bigl(\Bigl\{x\in\mathbb{R}^{+}: M_{\lambda}^{\alpha}f(x)>\gamma\Bigr\}\Bigr)= \frac{1}{2^{2\lambda+1}} \|f\|^{q}_{L^{1}(\mathbb{R}^{+},\mathrm{d}m_{\lambda})}; \\ &\lim_{\gamma\to 0}\gamma^{q}m_{\lambda}\Bigl(\Bigl\{ x\in\mathbb{R}^{+}: \Bigl|M_{\lambda}^{\alpha}f(x)-\frac{\|f\|_{L^{1}(\mathbb{R}^{+},\mathrm{d}m_{\lambda})}}{m_{\lambda}(I(x,x))^{1/q}}\Bigr|>\gamma\Bigr\}\Bigr)=0. \end{aligned} \] \end{itemize}