Large deviation asymptotics for occupancy problems. (Q1889799)

Urn occupancy problems center on the distribution of \(r\) balls in \(n\) cells, typically with each ball independently assigned to a given cell with probability \(1/n\). The present paper focuses on finite-dimensional occupancy measures in which urn occupancies above a given level are not distinguished. In this finite case, the authors provide a concise large deviations proof along with explicit, insightful and computable expressions for the rate functions and for the large deviations extremals. The rate function for the occupancy model after all the balls have been thrown is shown to have a simple and fairly explicit rate function, which can be defined in terms of relative entropy with respect to the Poisson distribution. Many different problems can be solved in this framework simply by changing the set over the rate function is minimized. The authors also present some sample path results for the evolution of the urn occupancies toward a particular event.