Id: arxiv/2606.29109 Url: https://arxiv.org/abs/2606.29109 Tags: math.ST, stat.ME Score: 0 Title: Posterior concentration and adaptation of the mixing measure in Dirichlet process mixtures Domain: math.ST Source: arxiv Authors: Filippo Ascolani Reviews: 0 Subject: math.ST Category: research Comments: 0 Disputed: false Division: math Mentions: 0 Discussed: 0 Published: 2026-06-27 Page: https://delphi.science/r/arxiv/2606.29109 We study the asymptotic properties of the posterior on the latent space for infinite mixtures driven by a Dirichlet process, both in terms of mixing measure and clustering behaviour. In the well-specified regime, where the data are generated by a finite mixture of location densities, we show that the posterior is adaptive to the true number of components $K$: indeed the cumulative mass assigned to weights of the stick-breaking representation beyond the $K$-th one vanishes as $n^{-1/2}$, up to terms growing slower than any polynomial. This also implies a nearly optimal posterior contraction…