A sharper Lyapunov-Katz central limit error bound for i.i.d. summands Zolotarev-close to normal
Lena Jonas, Lutz Mattner
TL;DR
This work sharpens finite-sample central limit bounds for i.i.d. sums with moment order $r=2+\delta$, $\delta\in[0,1]$, by replacing moment-based error terms with a weak distance to normality measured by Zolotarev–Senatov norms $\zeta_{m,g}$. It establishes a Senatov–Zolotarev type bound, $\zeta_1(\widetilde{P^{\ast n}}-\mathrm{N}) \le n^{-\delta/2} \xi_{\delta}( \zeta_1(\widetilde{P}-\mathrm{N}), \zeta_{2,\delta}(\widetilde{P}-\mathrm{N}) )$, and then derives the Lyapunov–Katz bound in the Kolmogorov metric, $\|\widetilde{P^{\ast n}}-\mathrm{N}\|_{\mathrm{K}} \le \frac{c}{n^{\delta/2}} ( \zeta_1 \vee \zeta_{2,\delta} )(\widetilde{P}-\mathrm{N})$, for $n\ge 2$, with a universal constant $c$ (e.g., $c=48$). The proofs combine a refined smoothing argument, a convolution inequality, and a careful analysis of the $\zeta_{m,g}$ norms, relying on a partial generalisation of Senatov–Zolotarev theory and clarifying historical attributions and gaps (including errata to Mattner2024). The results extend Mattner (2024) by treating the full range $\delta\in[0,1]$ and improving the dependence on normal closeness, with explicit constants and a groundwork of auxiliary lemmas for the $\zeta_{m,g}$ framework.
Abstract
We prove a central limit error bound for convolution powers of laws with finite moments of order $r \in \mathopen]2,3\mathclose]$, taking a closeness of the laws to normality into account. Up to a universal constant, this generalises the case of $r=3$ of the sharpening of the Berry (1941) - Esseen (1942) theorem obtained by Mattner (2024), namely by sharpening here the Katz (1963) error bound for the i.i.d. case of Lyapunov's (1901) theorem. Our proof uses a partial generalisation of the theorem of Senatov and Zolotarev (1986) used for the earlier special case. A result more general than our main one could be obtained by using instead another theorem of Senatov (1980), but unfortunately an auxiliary inequality used in the latter's proof is wrong.
