Existence of joint moments of stable random variables (Q917137)

The authors consider general \(\alpha\)-stable random vectors \((X_ 1,...,X_ n)\). Their main result, which generalizes a result by \textit{G. Miller} [J. Multivariate Anal. 8, 346-360 (1978; Zbl 0394.62034)], reads \[ E | X_ 1|^{p_ 1}...| X_ n|^{p_ n}<\infty \quad iff\quad p_{k_ 1}+...+p_{k_ m}<\alpha \] for every maximal subset \(\{k_ 1,...,k_ m\}\) of \(\{\) 1,2,...,n\(\}\). Here maximal subsets are defined in terms of quantities occurring in the integral representation of \((X_ 1,...,X_ m)\). If the \(X_ j\) are independent, the condition reduces to \(p_ j<\alpha\) \((j=1,2,...,n)\). The authors also give some other examples. Proofs are based on the integral representation [see the first author, Stochastic Processes Appl. 30, No.1, 17-39 (1988; Zbl 0656.60029)].