Maximal estimates for a generalized spherical mean Radon transform acting on radial functions (Q778058)

The paper is a natural continuation of an earlier research by the authors [SIAM J. Math. Anal. 49, No. 6, 4402--4439 (2017; Zbl 1377.44002)], where two-weight \(L^p\)-\(L^q(L^r_t)\) estimates for a family of integral transforms \(\{M^{\alpha,\beta}_t\colon t>0\}\), called the \textit{generalized spherical means}, were studied. The paper under rewiev treats the case of the maximal operator \(f\mapsto M^{\alpha,\beta}_*f:=\sup_{t>0}|M^{\alpha,\beta}_tf|\), \(f\) being a function on \((0,\infty)\), and the obtained results are strongly related to \(L^p\) bounds for the maximal (Euclidean) maximal operator \(M^{\beta}_*f:=\sup_{t>0}|M^{\beta}f(\cdot,t)|\). Here \(M^\beta\) denotes the (Euclidean) generalized spherical means transformation \(M^\beta f(x,t)=\mathcal F^{-1}(m_\beta(t|\cdot|)\mathcal F f)(x)\), where \(\mathcal F\) is the Fourier transform in \(\mathbb R^n\), \(n\ge2\), and the radial multiplier is given by \[ m_\beta(s)=2^{\beta+n/2-1}\Gamma(\beta+n/2)J_{\beta+n/2-1}(s)/s^{\beta+n/2-1}, \qquad s>0. \] For \(\beta=0\) one recovers the classical spherical means \(M^0f(x,t)=\int_{S^{n-1}}f(x-ty)\,d\sigma(y)\), \((x,t)\in\mathbb R^n\times(0,\infty)\). \(L^p\) bounds for \(M^{\beta}_*\) were investigated among others by Stein, Bourgain, Mockenhaupt, Seeger and Sogge, and recently in [\textit{C. Miao} et al., Proc. Am. Math. Soc. 145, 4269--4282 (2017; Zbl 1379.42008)]. Altogether, these results, for given \(n\ge2\) and \(2 < p < \infty\), provide sufficient conditions on \(\beta\) for \(M_*^\beta\) to be bounded on \(L^p(\mathbb R^n)\). It is not known if these conditions are optimal. Another line of research was conducted for \(M^{\beta}_*\) in the restricted setting of radially symmetric functions, but with radial power weights involved. Recently in [\textit{J. Duoandikoetxea} et al., J. Math. Anal. Appl. 387, No. 2, 655--666 (2012; Zbl 1235.42013)] and [\textit{C. Cattaneo}, C. R. Acad. Sci., Paris 246, 431--433 (1958; Zbl 1339.35135)], for given \(n\ge2\) and \(1 < p <\infty\) (\(2 < p < \infty\) for \(n=2\) in the first paper), sufficient conditions on \(\gamma\) were provided for the boundedness of \(M^{0}_*\) or \(M^{(3-n)/2}_*\), respectively, on \(L^p_{\mathrm{rad}}(\mathbb R^n,|x|^\gamma\, dx)\). For a radial \(f=f_0(|\cdot|)\in L^2(\mathbb R^n)\), \(n\ge2\), with profile \(f_0\in L^2(\mathbb R_+, d\mu_{n/2-1})\), \(d\mu_{\alpha}(x)=x^{2\alpha+1}dx\), for each \(t>0\), \(M^\beta f(x,t)\) is radial in \(x\in\mathbb R^n\) with profile given by \(M^{n/2-1,\beta} f_0\), and the maximal operators \(M^{\beta}_*\) and \(M^{n/2-1,\beta}_*\) are connected in the same way. Therefore, \(M^{\beta}_*\) is bounded on \(L^p_{\mathrm{rad}}(\mathbb R^n,|x|^\gamma\, dx)\) if and only if \(M^{n/2-1,\beta}_*\) is bounded on \(L^p(\mathbb R_+,x^\gamma\, d\mu_{n/2-1})\) and this explains the motivation for investigating the broader family \(M^{\alpha,\beta}_*\), in which \(M^{n/2-1,\beta}_*\) is embedded. The main result of the reviewed paper, Theorem 1.5, provides for given \(\alpha>-1\), \(\alpha+\beta>-1/2\), \(1 < p <\infty\) and \(\delta\in\mathbb R\), under the assumption \(\frac1p<\alpha+\beta+1/2\), sufficient conditions on \(\delta\) for \(M^{\alpha,\beta}_*\) to be bounded on \(L^p(\mathbb R_+,s^\delta\,ds)\). The authors presume the result is sharp. In particular, in Proposition 1.6 the assumption \(\frac1p<\alpha+\beta+1/2\) is shown to be necessary and, moreover, the sharpness of the conditions imposed on \(\delta\) is verified for most choices of the parameters \(\alpha\) and \(\beta\). The proof of Theorem 1.5 relies on very delicate pointwise estimates of the integral kernels corresponding to the operators \(M^{\alpha,\beta}_t\) which brings sufficient control, in terms of a number of special operators with (essentially) known mapping properties, of the resulting maximal operators. Due to the connection mentioned above, for \(\alpha=n/2-1\) and appropriately specified values of \(\beta\) and \(\delta\), Theorem 1.5 implies the weighted results of both papers of \textit{J. Duoandikoetxea} et al. [J. Math. Anal. Appl. 387, No. 2, 655--666 (2012; Zbl 1235.42013); Proc. Am. Math. Soc. 144, No. 4, 1543--1552 (2016; Zbl 1339.35153)]. Also results in the unweighted case are implied and resulting sufficient conditions for the boundedness of \(M^\beta_*\) on \(L^p_{\mathrm{rad}}(\mathbb R^n)\) occur to be less restrictive than those in the general (non-radial) case. This gives rise to a challenging conjecture stated by the authors, that the obtained sufficient conditions for the boundedness of \(M^\beta_*\) on \(L^p_{\mathrm{rad}}(\mathbb R^n)\) remain sufficient also in the general case.