ResearchTrend.AI
  • Papers
  • Communities
  • Events
  • Blog
  • Pricing
Papers
Communities
Social Events
Terms and Conditions
Pricing
Parameter LabParameter LabTwitterGitHubLinkedInBlueskyYoutube

© 2025 ResearchTrend.AI, All rights reserved.

  1. Home
  2. Papers
  3. 0807.2096
85
49

A Bernstein-Von Mises Theorem for discrete probability distributions

14 July 2008
S. Boucheron
Elisabeth Gassiat
ArXivPDFHTML
Abstract

We investigate the asymptotic normality of the posterior distribution in the discrete setting, when model dimension increases with sample size. We consider a probability mass function θ0\theta_0θ0​ on \mathbbmN∖{0}\mathbbm{N}\setminus \{0\}\mathbbmN∖{0} and a sequence of truncation levels (kn)n(k_n)_n(kn​)n​ satisfying kn3≤ninf⁡i≤knθ0(i).k_n^3\leq n\inf_{i\leq k_n}\theta_0(i).kn3​≤ninfi≤kn​​θ0​(i). Let θ^\hat{\theta}θ^ denote the maximum likelihood estimate of (θ0(i))i≤kn(\theta_0(i))_{i\leq k_n}(θ0​(i))i≤kn​​ and let Δn(θ0)\Delta_n(\theta_0)Δn​(θ0​) denote the knk_nkn​-dimensional vector which iii-th coordinate is defined by \sqrt{n} (\hat{\theta}_n(i)-\theta_0(i)) for 1≤i≤kn.1\leq i\leq k_n.1≤i≤kn​. We check that under mild conditions on θ0\theta_0θ0​ and on the sequence of prior probabilities on the knk_nkn​-dimensional simplices, after centering and rescaling, the variation distance between the posterior distribution recentered around θ^n\hat{\theta}_nθ^n​ and rescaled by n\sqrt{n}n​ and the knk_nkn​-dimensional Gaussian distribution N(Δn(θ0),I−1(θ0))\mathcal{N}(\Delta_n(\theta_0),I^{-1}(\theta_0))N(Δn​(θ0​),I−1(θ0​)) converges in probability to 0.0.0. This theorem can be used to prove the asymptotic normality of Bayesian estimators of Shannon and R\'{e}nyi entropies. The proofs are based on concentration inequalities for centered and non-centered Chi-square (Pearson) statistics. The latter allow to establish posterior concentration rates with respect to Fisher distance rather than with respect to the Hellinger distance as it is commonplace in non-parametric Bayesian statistics.

View on arXiv
Comments on this paper