Maximal spacings in several dimensions (Q580812)

Consider n points which are independently and uniformly distributed within a bounded subset K of \(R^ d\). The maximal spacing can be defined as the \(d^{th}\) root of the volume \(V_ n\) of the largest set of specified convex shape and orientation, contained within K, which can be centred at one of the n points without containing any other point in its interior. It is proved that \[ nV_ n-\log n-(d-1)\log \log n-\log \alpha \] tends in distribution to a random variable having the extreme value distribution function exp(-exp(-u)), where \(\alpha\) is determined by the shape of the set used to measure the spacings. Also, \(\liminf_{n\to \infty}(nV_ n-\log n)/\log \log n=d-1\) a.s. and \(\limsup_{n\to \infty}(nV_ n-\log n)/\log \log n=d+1\) a.s.