A probabilistic approach to domains of partial attraction (Q749993)

The paper is a continuation of an earlier work by the author, \textit{E. Haeusler} and \textit{D. M. Mason} [ibid. 9, No.3, 259-333 (1988; Zbl 0657.60029)] completing the theory of ``probabilistic approach'' to the problem of asymptotic distribution of sums of i.i.d. r.v.'s \(X_ 1,X_ 2,..\). with a common non-degenerate distribution function F. For each \(n\geq 1\), let \(X_{n,1}\leq...\leq X_{n,n}\) denote the order statistics based on \(X_ 1,...,X_ n\) and let \(S_ n(m,k)=\sum^{k}_{j=m+1}X_{j,n}.\) The author investigates the limiting behaviour of normalied trimmed sums \(S_ n(m,n-k)\), \(S_ n(m,\ell_ n)\), \(S_ n(\ell_ n,r_ n)\), \(S_ n(r_ n,n-r_ n)\), \(S_ n(n-r_ n,n-\ell_ n)\) and \(S_ n(n-\ell_ n,n-k)\) for subsequences \(\{n'\}\) of \(\{\) \(n\}\), where \(\ell_{n'}\to \infty\), \(\ell_{n'}/r_{n'}\to 0\), \(r_{n'}/n'\to 0\), and in this way identifies the portions that contribute to the ingredients of the limiting infinitely divisible law. In particular, there are given necessary and sufficient conditions ensuring that F is in the domain of partial attraction of an infinitely divisible distribution. All the analytic conditions in the presented theorems and corollaries are expressed in terms of the quantile function \(Q(s)=\inf \{x:\;F(x)\geq s\},\) \(0<s\leq 1\), \(Q(0)=Q(0+)\) and the truncated variance \[ \sigma^ 2(s)=\int^{1-s}_{s}\int^{1-s}_{s}[\min (u,v)-u\cdot v]dQ(u)dQ(v),\quad 0<s<1/2; \] furthermore, centering and normalizing constants are certain functions of n, Q and \(\sigma\). The purely probabilistic approach proposed by the author enables him to prove also the well-known results such as Gnedenko's transitivity theorem and the fact that each F belongs to the domain of partial attraction of one or nondenumerably many (in fact continuum) different types of laws, or else does not belong to any domain of partial attraction at all. The key part of this approach is an integral representation of any infinitely divisible law by means of two independent Poisson processes.