Rearrangements of Gaussian fields (Q719776)

The authors define the notion of rearrangement for multiparameter functions defined on a compact set of \(\mathbb{R}^d\). For a function defined on the real line, a monotone rearrangement is a non-decreasing map which preserves the distribution of the original map. Monotone rearrangements find applications in Economics and are related to the optimal transport problem. Another useful result, is to determine the convex rearrangement of a Gaussian process as the limit in distribution of convex rearrangements of regular discrete approximations of the process. For instance, it has been shown in the literature that this limit is independent of the chosen approximation (piece-wise linear interpolation, approximation by convolution, etc.) for a certain class of Gaussian processes. In this paper, the authors prove this is no longer the case for anisotropic multivariate random fields. In Section 1 the authors provide a review of rearrangements for univariate functions and their links to Economics, optimal transport and to the regular approximations aforementioned. Then the results about the multivariate framework are presented in Section 2. Finally, Section 3 is devoted to the particular case of the fractional Brownian field and of the Brownian sheet.