English

Goodness-of-Fit Testing for H\"older-Continuous Densities: Sharp Local Minimax Rates

Statistics Theory 2023-03-21 v2 Information Theory math.IT Statistics Theory

Abstract

We consider the goodness-of fit testing problem for H\"older smooth densities over Rd\mathbb{R}^d: given nn iid observations with unknown density pp and given a known density p0p_0, we investigate how large ρ\rho should be to distinguish, with high probability, the case p=p0p=p_0 from the composite alternative of all H\"older-smooth densities pp such that pp0tρ\|p-p_0\|_t \geq \rho where t[1,2]t \in [1,2]. The densities are assumed to be defined over Rd\mathbb{R}^d and to have H\"older smoothness parameter α>0\alpha>0. In the present work, we solve the case α1\alpha \leq 1 and handle the case α>1\alpha>1 using an additional technical restriction on the densities. We identify matching upper and lower bounds on the local minimax rates of testing, given explicitly in terms of p0p_0. We propose novel test statistics which we believe could be of independent interest. We also establish the first definition of an explicit cutoff uBu_B allowing us to split Rd\mathbb{R}^d into a bulk part (defined as the subset of Rd\mathbb{R}^d where p0p_0 takes only values greater than or equal to uBu_B) and a tail part (defined as the complementary of the bulk), each part involving fundamentally different contributions to the local minimax rates of testing.

Keywords

Cite

@article{arxiv.2109.04346,
  title  = {Goodness-of-Fit Testing for H\"older-Continuous Densities: Sharp Local Minimax Rates},
  author = {Julien Chhor and Alexandra Carpentier},
  journal= {arXiv preprint arXiv:2109.04346},
  year   = {2023}
}

Comments

79 pages