Paperlane

no. 16 · Tuesday 17 November 2026

Effective-dimension bounds in the central limit theorem for centered Euclidean balls

Alexey Naumov, Sergey Samsonov, Vladimir Spokoiny and Vladimir Ulyanov

math.PR · v2 · 56 pp

Abstract

We establish a nonasymptotic Gaussian approximation bound over centered Euclidean balls for sums of independent mean zero random vectors in $\mathbb R^p$ that need not be identically distributed. Let $S_n$ denote the sum, $Σ=\operatorname{Var}(S_n)\ne0$, and let $G\sim N(0,Σ)$. Define the effective rank and the normalized squared spectral mass by \[ r(Σ)=\frac{\operatorname{tr}Σ}{\|Σ\|}, \qquad v(Σ)=\frac{\operatorname{tr}(Σ^2)}{\|Σ\|^2}. \] Under a uniform directional fourth-moment condition, covariance-balance assumptions, and the spectral conditions $r(Σ)\le C_Σv(Σ)$ and $v(Σ)\ge6$, we prove \[ \sup_{z\ge0} \left| \mathbb P\{\|S_n\|^2\le z\} - \mathbb P\{\|G\|^2\le z\} \right| \le C\,\frac{r(Σ)}{n}. \] The constant depends only on the parameters in the assumptions and is independent of the ambient dimension and individual small eigenvalues of the covariance matrix. The approximation error thus tends to zero when $r(Σ)=o(n)$, even if $p$ grows faster than $n$. We also provide an example showing that the order $r(Σ)/n$ is sharp in general. The same estimate extends to random elements of a separable Hilbert space with trace class covariance. The proof combines Esseen smoothing, truncation, covariance partitioning, moment matching, frame/lattice estimates, a second-order endpoint expansion and comparison estimates expressed in terms of the covariance matrices.

comments
56 pages; We improved the dependence on the effective dimension to optimal
fields
Probability, Statistics Theory
pdf ↗ page arXiv:2610.12722 ↗