Regular simplices and Gaussian samples (Q1317873): Difference between revisions

The authors consider a randomly rotated spherico-regular simplex in \(\mathbb{R}^ n\) determined by the set of \(m\leq n+1\) points having the same interpoint distances and lying on a sphere centered at the origin. It is shown that the projections of the simplex's vertices onto a fixed subspace are affine equivalent in distribution to a standard Gaussian point sample in this subspace. The affine transformation can be taken so that (i) it is orientation-preserving and (ii) its linear and translation parts and the random rotation are three mutually stochastically independent actions. Conversely, only the vertices of such a simplex have this property. This theorem generalizes substantially a result of \textit{F. Affentranger} and \textit{R. Schneider} [Discrete Comput. Geom. 7, No. 3, 219-226 (1992; Zbl 0751.52002)]. The proofs are based on factorizations of random matrices. Similar results hold for random orthogonal transformations of a spherico- regular simplex. As corollaries, the authors give limit results for the expected number of \(k\)-dimensional faces of the convex hull of a standard Gaussian sample.