References & Citations
Mathematics > Statistics Theory
Title: Nonparametric Bayesian posterior contraction rates for scalar diffusions with high-frequency data
(Submitted on 15 Feb 2018 (this version), latest version 23 Aug 2018 (v2))
Abstract: We consider inference in the scalar diffusion model $dX_t=b(X_t)dt+\sigma(X_t)dW_t$ with discrete data $(X_{j\Delta_n})_{0\leq j \leq n}$, $n\to \infty,~\Delta_n\to 0$ and periodic coefficients. For $\sigma$ given, we prove a general theorem detailing conditions under which Bayesian posteriors will contract in $L^2$-distance around the true drift function $b_0$ at the frequentist minimax rate (up to logarithmic factors) over Besov smoothness classes. We exhibit natural nonparametric priors which satisfy our conditions. Our results show that the Bayesian method adapts both to an unknown sampling regime and to unknown smoothness.
Submission history
From: Kweku Abraham [view email][v1] Thu, 15 Feb 2018 16:11:54 GMT (49kb)
[v2] Thu, 23 Aug 2018 14:27:30 GMT (81kb)
Link back to: arXiv, form interface, contact.