A kernel bound for non-symmetric stable distribution and its applications (Q2173772)

Let \(\xi_{1,n},\ldots,\xi_{n,n}\) be independent random variables with \(\mathbb E[\xi_{i,n}]=0\) for \(i=1,\ldots,n\) and denote \(S_n=\sum_{i=1}^n \xi_{i,n}\).The authors prove an upper bound for the Wasserstein-1 distance of the law of \(S_n\) to an \(\alpha\)-stable distribution with \(\alpha\in(1,2)\), extending an inequality in [\textit{L. Xu}, Ann. Appl. Probab. 29, No. 1, 458--504 (2019; Zbl 1448.60062)] to the non-symmetric case. The upper bound has a similar form as in the classical Esseen inequality for the corresponding Kolmogorov distance. The authors further give some explicit examples for the upper bound in case \(\xi_{i,n}=n^{-1/\alpha}(X_i-\mathbb E[X_i])\) for an i.i.d. sequence \((X_i)\) in the domain of attraction of an \(\alpha\)-stable distribution with \(\alpha\in(1,2)\). In particular, they extend a convergence rate result in [\textit{P. Hall}, J. Lond. Math. Soc., II. Ser. 23, 179--192 (1981; Zbl 0455.60023)].