Uniformly distributed points on the sphere (Q2566919)

Uniformly distributed sequences of deterministic points \(P= \{ X_1, \ldots, X_N \}\) on the unit sphere are considered for an integration by the quasi-Monte Carlo method, where the integral is approximated by the mean over these points. It is shown that the property of sequences being uniformly distributed is equivalent to requiring the sequences to have a low discrepancy. The \textit{discrepancy} measures the deviation of a sequence from a uniform distribution. Different ways of defining the discrepancy (i.~e., \textit{star discrepancy} and a general \(L^p\) discrepancy), as well as the error bound in the sense of Hardy and Krause [see \textit{H. Niederreiter}, Random number generation and quasi-Monte Carlo methods. SIAM (1992; Zbl 0761.65002)] are briefly reviewed for the euclidian case of the \(s\)-dimensional cube \(I^s=[0,1)^s\). The error bound for the case of unit sphere is derived in terms of the \(L^p\) discrepancy, which is related to Beltrami operator and the \(s\)-th iterated Greens's function of the unit sphere. Finally, several examples of sequences (Reuter, Brand, Hammersley, Halton) on the sphere are considered and illustrated.