Rates of contraction with respect to \(L_2\)-distance for Bayesian nonparametric regression (Q2326065)

This paper is about the problem of rates of convergence in nonparametric regression, in a version that applies to posterior distributions. The point of departure is the work [Ann. Stat. 28, No. 2, 500--531 (2000; Zbl 1105.62315)] by \textit{S. Ghosal} et al., where a framework of prior concentration, sieves and test functions is set up in order to approach this problem. According to the authors of the paper under review, the ``major contribution of this work \(\ldots\) is that we impose certain structural assumption(s) on the sieves to construct the desired test functions. By doing so, we are able to make the framework more concrete and instructive for a variety of nonparametric regression priors \(\dots\)'' (p. 3489). It is also emphasized that the space of regression functions is allowed to be unbounded.