English

Approximation by log-concave distributions, with applications to regression

Statistics Theory 2011-10-17 v4 Probability Methodology Statistics Theory

Abstract

We study the approximation of arbitrary distributions PP on dd-dimensional space by distributions with log-concave density. Approximation means minimizing a Kullback--Leibler-type functional. We show that such an approximation exists if and only if PP has finite first moments and is not supported by some hyperplane. Furthermore we show that this approximation depends continuously on PP with respect to Mallows distance D1(,)D_1(\cdot,\cdot). This result implies consistency of the maximum likelihood estimator of a log-concave density under fairly general conditions. It also allows us to prove existence and consistency of estimators in regression models with a response Y=μ(X)+ϵY=\mu(X)+\epsilon, where XX and ϵ\epsilon are independent, μ()\mu(\cdot) belongs to a certain class of regression functions while ϵ\epsilon is a random error with log-concave density and mean zero.

Keywords

Cite

@article{arxiv.1002.3448,
  title  = {Approximation by log-concave distributions, with applications to regression},
  author = {Lutz Duembgen and Richard Samworth and Dominic Schuhmacher},
  journal= {arXiv preprint arXiv:1002.3448},
  year   = {2011}
}

Comments

Version 3 is the technical report cited in the published paper. Published in at http://dx.doi.org/10.1214/10-AOS853 the Annals of Statistics (http://www.imstat.org/aos/) by the Institute of Mathematical Statistics (http://www.imstat.org)

R2 v1 2026-06-21T14:48:20.236Z