A note on the finite-dimensional Dirichlet prior
Xia Yemao, Gou Jianwei · Communication in Statistics- Theory and Methods · 2016
As an approximation to the Dirichlet process which involves the infinite-dimensional distribution, finite-dimensional Dirichlet prior is a widely appreciated method to model the underlying distribution in non parametric Bayesian analysis. In this short note, we present some key characteristics of finite-dimensional Dirichlet process and exploit some important sampling properties which are very useful in Bayesian non parametric/semiparametric analysis.