Are random permutations spherically uniform? (Q2184629)

The author of the paper studies the following problem. For any non-zero real vector of length \(q\), \(\boldsymbol{x}\in\mathbb{R}^q\), \(q\ge 2\), let \(\Pi (\boldsymbol{x})\) be the set of all permutations of \(\boldsymbol{x}\). Consider then the \((q-2)\)-sphere which contains \(\Pi (\boldsymbol{x})\). Do there exist sequences \(\{\boldsymbol{x}_q \in \mathbb{R}^q\}\) such that \(\Pi (\boldsymbol{x}_q)\) approaches spherical uniformity as \(q\rightarrow\infty\) ? The deviation from spherical uniformity is, in this paper, measured by the normalized spherical cap discrepancy or the largest empty cap discrepancy. In the paper it is shown that the natural choice of \(\boldsymbol{x}_q=(1,2,\ldots,q)^\top\) considerably deviates from spherical uniformity. Furthermore, also other, more favorable choices of \(\boldsymbol{x}_q\) are considered, which are in a certain sense better than \((1,2,\ldots,q)^\top\), but unfortunately also do not approach spherical uniformity. Whether or not there exist sequences \(\{\boldsymbol{x}_q \in \mathbb{R}^q\}\) that can approach spherical uniformity remains an open question.