Rate of convergence for predictive distributions of exchangeable indicators
Patrizia Berti, Irene Crimaldi, Luca Pratelli, Pietro Rigo · AMS Acta (University of Bologna) · 2008
take values in an arbitrarymeasurable space, is dealt with in Subsection 4.2.(ii) In Bayesian statistics, π is the prior distribution. And priors are typicallyassumed absolutely continuous with respect to Lebesgue measure (possibly, withsmooth densities). The results mentioned above, thus, apply to most Bayesianproblems.(iii) Let p > 1 and c > 0. Those π which are absolutely continuous with respectto Lebesgue measure, with a density f such that R