{"id":"arxiv/2610.10022","url":"https://arxiv.org/abs/2610.10022","tags":["math.PR","math.ST"],"score":0,"title":"Improved Berry-Esseen bounds for multivariate nonlinear statistics in convex distance","domain":"math.PR","source":"arxiv","authors":["Zhi-Jun Cai","Jing-Cai Yang","Zhuo-Song Zhang"],"reviews":0,"subject":"math.PR","summary":"In this paper, we establish two nonasymptotic Berry--Esseen bounds over convex sets for the Gaussian approximation of multivariate nonlinear statistics. The statistics of interest can be written as a sum of independent centered random vectors plus a remainder that may depend on all observations. The first bound retains the classical factor $d^{1/4}$ in the contribution of the independent sum, where $d$ is the dimension, while controlling the remainder through its size and its sensitivity to replacing a single observation. The second bound expresses the contribution of the independent sum in…","category":"research","comments":0,"disputed":false,"division":"math","mentions":0,"discussed":0,"published":"2026-10-07","reactions":{}}