A stable marriage of Poisson and Lebesgue (Q850966)

Let \(\Xi\) be a discrete set in the \(d\)-dimensional space whose elements are called centers. In the paper, ``fair'' allocations of the underlying space to the points of \(\Xi\) are studied in which the regions allocated to different centers have equal volumes \(\alpha\in [0,\infty]\). Assume that sites and centers both prefer to be allocated as close as possible. An allocation is called unstable if some site and center both prefer each other over their current allocation. Otherwise, it is called stable. For stable allocations, an existence result is proved for any discrete set of centers. The stable allocation is unique if the configuration \(\Xi\) is recurrent in some sense. If \(\Xi\) is a translation-invariant point process with finite intensity \(\lambda\), then there exists the unique stable allocation a.s. Furthermore, phase transition problems are studied depending on the value of \(\lambda \alpha\). Namely, the question whether all sites are claimed or all centers are sated is investigated in detail. It is shown that the region allocated to each center is a union of finitely many bounded connected components. If \(\Xi\) is a homogeneous Poisson point process and \(\lambda=\alpha=1\) (critical case), then every site is a.s. ``coveted'' by infinitely many centers and vice versa, every center is ``desired'' a.s. by an infinite volume of sites. In this case, power law lower bounds on the allocation distance of a typical site are given.