A probabilistic proof of Kruglov's theorem on the tails of infinitely divisible distributions

From MaRDI portal





Let \(G\) denote the distribution function of an infinitely divisible law on the real line, let \(L\) and \(R\) denote the distribution functions of the Lévy measure restricted to \((-\infty, 0)\) and \((0, \infty)\), respectively. Let \(\mathcal {K}\) denote the set of nonnegative locally bounded Borel functions which are submultiplicative or subadditive (called Kruglov class). It is well known that generalized moments exist for \(G\) iff truncated generalized moments (outside a neighbourhood) exist for the Lévy measure, i.e. for \(h \in \mathcal {K}\) \[ \int_\mathbb{R} h \,dG < \infty \quad \text{iff} \quad \int_{(-\infty, -\varepsilon)} h \,dL + \int_{(\varepsilon, \infty)} h \,dR < \infty \quad (\varepsilon >0).\tag \(*\) \] The aim of the paper is to obtain a new proof of this result \((*)\) based only on probabilistic methods: Avoiding the use of characteristic functions, the author uses a probabilistic interpretation of the Lévy-Khinchin formula as in [\textit{S. Csörgő} and \textit{D. M. Mason}, in: Sums, Trimmed sums and extremes. Prog. Probab. 23, 317--335 (1991; Zbl 0728.60019)]. Reviewer's remark: Reading this paper the reviewer learned that the well-known beautiful characterization \((*)\) is originally due to \textit{V. M. Kruglov} [Theory Probab. Appl. 17, 685-694 (1972); translation from Teor. Veroyatn. Primen. 17, 723--732 (1972; Zbl 0279.60035)]. It seems worth to mention that in the past this result was generalized into different directions, in particular it was proved for wide classes of topological groups.











This page was built for publication: A probabilistic proof of Kruglov's theorem on the tails of infinitely divisible distributions

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q5696192)