Domains of semi-stable attraction of nonnormal semi-stable laws (Q1340304): Difference between revisions

Operator-limit distributions theory deals with sequences of the form \((*)\) \(A_ n (X_ 1 + X_ 2 + \cdots + X_{k_ n}) + a_ n\), where \(X_ j\) are \(\mathbb{R}^ d\)-valued random variables, \(a_ n \in \mathbb{R}^ d\) and \(A_ j\) are linear operator (matrices) on \(\mathbb{R}^ d\). If \(X_ j\) are i.i.d. and \(k_{n + 1}/ k_ n \to 1\), then limits of \((*)\) are called operator-stable measures. In case \(k_{n + 1}/k_ n \to c > 1\) we have so-called operator-semi-stable measures. For fixed (operator- semi-stable) \(\nu\) we say that the random variable \(Y\) belongs to the \({\mathcal D}\)-domain of \(\nu\) if there exist \(k_ n \in \mathbb{N}\) such that \(k_{n + 1} /k_ n \to c > 1\) and diagonal matrices \(D_ n\) and vectors \(v_ n \in \mathbb{R}^ d\) such that \(D_ n (Y_ 1 + Y_ 2 + \cdots + Y_{kn}) + v_ n \Rightarrow \nu\) (weak convergence), where \(Y_ j\) are i.i.d. copies of \(Y\). The author characterizes (partially) \({\mathcal D}\)-domains of \(\nu\) and then describes more completely moments of measures from \({\mathcal D}\)-domains.