Conditional distributions and characterizations of multivariate stable distribution (Q1893359)

Let \(X\) and \(Y\) be two identically distributed random variables. \textit{M. Ahsanullah} [Metrika 32, 215-218 (1985; Zbl 0609.62019)] showed that if the conditional distribution of \(X|Y=y\) is a normal distribution with mean \(ay+b\) and with variance \(\sigma^2\), then \(|a|<1\), and the joint distribution of \(X\) and \(Y\) is a bivariate normal distribution. Also in this paper Ahsanullah proposed a multivariate conjecture to his bivariate result as follows: Let \(X_1, \dots, X_n\) be \(n\) identically distributed random variables. If \(X_1 |X_2 =x_2, \dots, X_n=x_n\) has a normal distribution with mean \(a_2x_2+ \cdots+ a_nx_n +b\) and the variance \(\sigma^2\), then \(X_1, \dots, X_n\) have a joint multivariate normal distribution. We extend the result of Ahsanullah in bivariate normal distribution to a result in bivariate stable distribution; then we will use this result to characterize multivariate stable distribution. Some supplementary conditions are added to the conjecture of Ahsanullah in a conditional stable case to make the joint distribution of \(X_1, \dots, X_n\) be multivariate stable.