Gelfand pairs, \(K\)-spherical means and injectivity on the Heisenberg group (Q1808880)

Let \(\mathbb{H}_n\) denote the Heisenberg group which is \(\mathbb{C}^n \times\mathbb{R}\) with the group law \[ (z,t)(w,s)= (z+w,t+s+ \textstyle{1\over 2} \text{Im} z\overline w). \] The study of injectivity of the spherical mean value operators on \(L^p(\mathbb{H}_n)\) has been carried out by \textit{S. Thangavelu} [J. Anal. Math. 63, 255-286 (1994; Zbl 0822.43001)] for the case \(1\leq p< \infty\) and by \textit{M. Agranovsky}, \textit{C. Berenstein}, \textit{D.-C. Chang} and \textit{D. Pascuas} [J. Anal. Math. 63, 131-173 (1994; Zbl 0808.43002)] for the case \(p=\infty\). The authors study the above injectivity properties in the more general setup of Gelfand pairs. Let \(G\) be a nilpotent Lie group and \(K\) a compact subgroup of \(\Aut(G)\). Recall that the pair \((G,K)\) is a Gelfand pair if the subalgebra \(L^1_K(G)\) of \(K\)-invariant functions in \(L^1(G)\) is commutative with respect to the usual convolution \(f*g\). Let \((\mathbb{H}_n,K)\) be a Gelfand pair. Let \(\mu_{z_0,t_0}\) denote the normalized surface measure on the \(K\)-orbit of the point \((z_0,t_0) \in\mathbb{H}_n\). Given a continuous function \(f\) on \(\mathbb{H}_n\), the spherical mean over the \(K\)-orbit of \((z_0,t_0) \in\mathbb{H}_n\) is defined as the convolution operator \(f*\mu_{z_0,t_0}\). The injectivity theorem for \(K\)-spherical means for the case \(1\leq p<\infty\) is the following. Theorem. If \(f\in L^p(\mathbb{H}_n)\), \(1\leq p<\infty\), satisfies \(f*\mu_{z_0,t_0}=0\), then \(f\equiv 0\). -- The proof of the theorem is based on harmonic analysis on \(\mathbb{H}_n\) due to \textit{R. Strichartz} [J. Funct. Anal. 96, 350-406 (1991; Zbl 0734.43004)]. For \(p=\infty\) the above spherical mean value operators are not injective. Nevertheless, if the spherical means \(\{f*\mu_{z_i, t_i}\}\) over \(K\)-orbits of sufficiently many points \(\{(z_i,t_i)\}\) in \(\mathbb{H}_n\) vanish, the authors give a necessary and sufficient condition on the points \(\{(z_i,t_i)\}\) which guarantees \(f\equiv 0\). For \(K=U(n)\) this is equivalent to the condition for the two-radius theorem in \textit{M. Agranovsky}, \textit{C. Berenstein}, \textit{D.-C. Chang} and \textit{D. Pascuas} [op. cit.].