On functions on the sphere with zero integrals over geodesics (Q2254115)

Given a constant \(\delta<\frac{\pi}{2}\), the authors consider the open set of the \(2\)-sphere containing the equator \(E_0=\{\xi: d(o,\xi)=\frac{\pi}{2}\}\), \[ K_{\delta}=\{\xi\in \mathbb{S}^2: \delta <d(o,\xi)<\pi-\delta\}, \] where \(o=(0,0,1)\) and \(d\) is the intrinsic distance of \( \mathbb{S}^2\). The main objective is to describe the continuous functions on \(K_{\delta}\) with vanishing integrals over any closed geodesic contained in \(K_{\delta}\). For each nonnegative integer \(M\in \mathbb{Z}_+\), they consider the classes of functions: \[ \mathcal{R}_M(K_{\delta})=\{f \in C(K_{\delta}): \int_{E_0} f(\tau\xi)\mathcal{P}_M(\xi)\,dl(\xi)=0, \forall{\tau}\in O(3): \tau E_0\subset K_{\delta}\}, \] \[ C_M(K_{\delta})=\{f\in C(K_{\delta}): f(-\xi)=(-1)^Mf(\xi), \forall \xi \in K_{\delta}\}, \] where \(\mathcal{P}_M(\xi)=(\xi_1+i\xi_2)^M\). Denoting by \(\mathcal{H}_l\) the space of spherical harmonics of degree \(l\) and by \({H}_M =\sum_{n=0}^{[M/2]-1} \mathcal{H}_{2n+2[M/2]}\), the authors prove that \(\mathcal{R}_M(S^2)=C_{M+1}(S^2)+ {H}_M\). For \(M=0\) this gives the classical Minkowski theorem, and as consequences they claim that given two nonnegative integers \(M_1<M_2\), \(\mathcal{R}_{M_1}(S^2)\cap \mathcal{R}_{M_2}(S^2)\) is just \(C_{M_1+1}(S^2)+ {H}_{M_1}\) for \(M_1+M_2\) even, and \({H}_{M_1}+ {H}_{M_2}\), for \(M_1+M_2\) odd. When restricted to \(K_{\delta}\), for \(0\leq \delta<\frac{\pi}{2}\), they obtain a space decomposition \[ \mathcal{R}_M(K_{\delta})=C_{M+1}(K_{\delta})+{H}_M|_{K_{\delta}} +\mathcal{G}_M(K_{\delta}), \] where \(\mathcal{G}_M(K_{\delta})\) is a suitable space of functions with vanishing \(k\) and \(-k\) terms of the Fourier series, for \(|k|\leq M+1\), and claim that the unique function \(f\in \mathcal{R}_M(K_{\delta})\cap C_M(K_{\delta})\) with bounded \(f(\xi)(\xi_3-\cos r)^{-m}\) in a neighbourhood of some fixed circle of radius \(r\), \(S_r\subset K_{\delta}\), for all positive integer \(m\), is the zero function, and counterexamples exist when the conditions are not satisfied. This generalizes a classical vanishing theorem when \(M=0\) and \(S_r=E_0\). As a consequence, they obtain vanishing results for functions \(f\in \mathcal{R}_M(K_{\delta})\cap C_M(K_{\delta})\) with zeros on a certain infinite number of circles, or with certain type of growth near the north pole \(\xi=o\), if \(\delta=0\).