Random measures and measure-valued processes are probability models describing systems involving complex spatial and temporal structures. They arise in many different branches of science. In Bayesian statistics they serve as prior distributions. A fundamental structure unifying different subjects is the underlying property of exchangeability. Our research focuses on the asymptotic behaviour of various statistics and their application in forecasting and predictions.
Impact
In species sampling problem, an important statistics is the number of distinct species in a random sample. Depending on the correlation structures in the model, the large sample approximation may lead to different variables, usually called diversity variables. Parameters in these variables correspond to different mechanism in the system. We stablished a series of limit theorems that serves as the basis for statistical estimation inference on these parameters.
Student Experience
One Ph.D student (graduated in 2014) was involved in the project . Another Ph.D. student is currently working on a sub-project.
