Marginals of multivariate Gibbs distributions with applications in Bayesian species sampling

TL;DR

This paper derives explicit formulas for marginals and moments of multivariate Gibbs distributions, enhancing Bayesian species sampling methods by providing new tools for nonparametric estimation of species probabilities.

Contribution

It introduces explicit formulas for marginals and factorial moments of multivariate Gibbs distributions, advancing Bayesian nonparametric species sampling models.

Findings

Derived explicit formulas for joint falling factorial moments.

Provided applications to Bayesian estimation of species probabilities.

Enhanced understanding of Gibbs partition models in Bayesian sampling.

Abstract

Gibbs partition models are the largest class of infinite exchangeable partitions of the positive integers generalizing the product form of the probability function of the two-parameter Poisson-Dirichlet family. Recently those models have been investigated in a Bayesian nonparametric approach to species sampling problems as alternatives to the Dirichlet and the Pitman-Yor process priors. Here we derive marginals of conditional and unconditional multivariate distributions arising from exchangeable Gibbs partitions to obtain explicit formulas for joint falling factorial moments of corresponding conditional and unconditional Gibbs sampling formulas. Our proofs rely on a known result on factorial moments of sum of non independent indicators. We provide an application to a Bayesian nonparametric estimation of the predictive probability to observe a species already observed a certain number of times.