The uniform norm of hyperinterpolation on the unit sphere in an arbitrary number of dimensions (Q5937040)

The authors study the hyperinterpolation method to approximate a continuous function \(f\) on the unit sphere \(S^{r-1} \subset {\mathbb R}^r\) by the set \(P_n^r(S^{r-1})\) of spherical polynomials of degree \(\leq n\). Let \(\int_{S^{r-1}} f \approx \sum_{k=1}^{m} w_kf(t_k)\) be an \(m\)-point quadrature rule with \(\{t_1,\ldots,t_m\} \subset S^{r-1}\) and \(w_k > 0\), \(k=1, \ldots,m\), which is exact on all spherical polynomials of degree \(\leq 2n\). Let \(\{p_1,\ldots,p_d\}\) be an orthonormal basis of \(P_n^r(S^{r-1})\). The hyperinterpolation operator is defined as \( L_nf = \sum_{j=1}^d\langle f,p_j\rangle_m p_j,\) where \(\langle f,g\rangle_m = \sum_{k=1}^{m}w_kf(t_k)g(t_k).\) In a more general situation, hyperinterpolation was introduced by \textit{I. H. Sloan} [J. Approximation Theory 83, No. 2, 238-254 (1995; Zbl 0839.41006)]. The authors prove that under an additional ``quadrature regularity'' assumption on the quadrature rule \(\|L_n\|_{C(S^{r-1}) \to C(S^{r-1})} = O(n^{r/2-1})\), which is the optimal order among all linear projections onto \(P_n^r(S^{r-1})\). This extends the \(r=3\) result of \textit{I. H. Sloan} and \textit{R. S. Womersley} [J. Approximation Theory 103, No. 1, 91-118 (2000; Zbl 0946.41007)].