1: \begin{abstract}

2: This work provides theoretical foundations for kernel methods in the hyperspherical context. 

3: Specifically, we characterise the native spaces (reproducing kernel Hilbert spaces) and the Sobolev spaces associated with kernels defined over hyperspheres. 

4: Our results have direct consequences for kernel cubature, determining the rate of convergence of the worst case error, and expanding the applicability of cubature algorithms based on Stein's method. 

5: We first introduce a suitable characterisation on Sobolev spaces on the $d$-dimensional hypersphere embedded in $(d+1)$-dimensional Euclidean spaces. Our characterisation is based on the Fourier--Schoenberg sequences associated with a given kernel. Such sequences are hard (if not impossible) to compute analytically on $d$-dimensional spheres, but often feasible over Hilbert spheres. We circumvent this problem by finding a projection operator that allows to Fourier mapping from Hilbert into finite dimensional hyperspheres. We illustrate our findings through some parametric families of kernels.  \\

6: 

7: \vspace{0.3cm}

8: \noindent 

9: 

10: {\em Keywords: Discrepancies; Kernel Cubature; Native Spaces; Sobolev Spaces.

11: %\textcolor{green}{[I understood ``keywords'' to mean relevant search terms that are not already in the title (since the title is automatically indexed by search engines)?]}

12: }

13: \end{abstract}

14: