Pointwise ergodic theorems for radial averages on the Heisenberg group (Q1369695)

Let \(H=H^n =\mathbb{C}\times \mathbb{R}\) denote the Heisenberg group, \(G_n= U(n) \ltimes H^n\) the Heisenberg motion group, and let \(\sigma_r\) (resp. \(\sigma_r')\) denote the normalized Lebesgue measure on the sphere \(\{(z,0): |z|=r\}\) of \(H^n\) (resp. the measure \(1_{U(n)}* \sigma_r* 1_{U(n)}\) on \(G_n)\). Let \((X, {\mathcal B}, m)\) be a standard Borel probability space on which \(H_n\) or \(G_n\) act measurably and ergodically by measure preserving transformations. The authors prove maximal and pointwise ergodic theorems in \(L^p\). They show that the spherical means \(\sigma_r\) on \(H_n\) and \(\sigma_r'\) on \(G_n\) are pointwise ergodic families in \(L^p(X)\) provided \((2n-1)/(2n-2)<p <\infty\) and \(n>1\). The proof of this result depends on the fact that under this condition on \(p\), for any non-negative \(f\in L^p(X)\), the maximal function \({\mathcal M}_\sigma f\) is measurable, and the strong \(L^p\) maximal inequalities \(|{\mathcal M}_\sigma f|_p\leq C_p(n) |f|_p\) hold for the families \(\sigma_r\) on \(H^n\) and \(\sigma_r'\) on \(G_n\). Similar results are shown for the reduced Heisenberg groups, where they are best possible. The proof uses the spectral theory of the Banach algebra of radial measures on \(H^n\) and decay estimates for its characters to establish maximal inequalities using spectral methods, in particular Littlewood-Paley-Stein square functions and analytic interpolation.