Asymptotic distribution of coordinates on high dimensional spheres (Q2461030)

For a random vector \(X^{(n)} =(X_1, \ldots, X_n)\) with independent components \(X_i\) with densities \(f_p (t) \sim e^{- | t| ^p / p}\), the normalized random vector \(U^{(n)} = n^{1/p} X^{(n)}/\|X^{(n)}\|_p\) is uniformly distributed on the \(p\)-sphere \(S_{p,n} (n^{1/p})\) of radius \(n^{{1/p}}\) in \(\mathbb{R}^n\) as was shown by \textit{S. T. Rachev} and \textit{L. Rüschendorf} [Ann. Probab. 19, No. 3, 1311--1337 (1991; Zbl 0732.62014)]. Extending previous work of \textit{P. Diaconis} and \textit{D. Freedman} [Ann. Inst. Henri Poincaré, Probab. Stat. 23, Suppl., 397--423 (1987; Zbl 0619.60039)] who dealt with the case \(p=1,2\) it was shown in that paper that the first \(k\) coordinates of \(U^{(n)}\) are asymptotically independent with distribution \(\sim f_p\) and the convergence rate (w.r.t. total variation) is of order \(1/{\sqrt{n}}\). In the present paper a Brownian bridge approximation of the corresponding empirical process of the components of a uniformly distributed point on \(S_{p,n} (n^{1/p})\) ist derived. The proof uses essentially the representation of \(U^{(n)}\) as described above which allows to relate this empirical process to that of an independent sample. As typical in similar approximations besides the Browian bridge a second simple normal process arises.