Id: arxiv/2603.14037 Url: https://arxiv.org/abs/2603.14037 Tags: math.ST Score: 0 Title: On a Universal Strictly Decreasing Nonparametric Estimator Applied to the Drift Function of a Recurrent Diffusion Process Estimation Domain: math.ST Source: arxiv Authors: Nicolas Marie Reviews: 0 Subject: math.ST Category: research Comments: 0 Disputed: false Division: math Mentions: 0 Discussed: 0 Published: 2026-03-14 Page: https://delphi.science/r/arxiv/2603.14037 This paper deals with a copies-based continuously differentiable and strictly decreasing estimator of the drift function for stochastic differential equations defining recurrent diffusion processes. The first part of our paper deals with non-asymptotic $\mathbb L^1$-risk bounds and a bandwidths selection procedure for a universal monotone estimator. These results are tailor-made to our framework, and then applied to the estimation of the drift function of recurrent diffusion processes in the second part of the paper.