Gould series distributions with applications to fluctuations of sums of random variables (Q1078939): Difference between revisions

Gould series distributions (GSD) have probability functions of the form \[ p(x;s,r)=[A(s,r)]^{-1}a(x;r)s(s+rx-1)_{x-1},\quad x=0,1,..., \] where \(c_{-1}=(c+1)^{-1}\), \(c_ x=c(c-1)...(c-x+1)\), \(A(s,r)=\sum a(x;r)s(s+rx-1)_{x-1}\), and s, r are parameters. GSD's come up in fluctuations of sums of exchangeable random variables and this class includes generalized binomial (GB) [\textit{G. C. Jain} and \textit{P. C. Consul}, SIAM J. Appl. Math. 21, 501-513 (1971; Zbl 0234.60010)] and generalized negative binomial (GNB) [\textit{P. C. Consul} and \textit{L. R. Shenton}, ibid. 23, 239-248 (1972; Zbl 0226.60017)] distributions. In this paper closed form expressions for the probability generating function and the factorial moments of GSD are derived. Under certain conditions, it is shown that among GSD's only GB and GNB distributions are closed under convolution.