New properties and representations for members of the power-variance family. I (Q1936260)

This article is devoted to a detailed study of an important class of exponential dispersion models, EDMs, introduced in 1984 by M.C.K. Tweedy and now named after him. Earlier investigations of the authors are carried further. In the introduction,an outline of earlier results is given, dealing with the relevance of the power parameter \(p \in (-\infty, 0]\cup (2, \infty)\) and relations to extreme stable distributions with an index of stability \(\alpha\in (0 ,1)\cup(1, 2]\). The authors' main aim is to fill a gap, namely to closely analyze the relationship between TWEEDY exponential dispersion models and the Wright function \[ \phi(\rho, \delta, z) := \sum_{k=0}^\infty {\frac{z^k}{k! \Gamma(k \rho + \delta)}} \] with \(\rho \in (-1, 0)\cup (0, \infty)\), \(\delta\) a complex number. As for the class of functions of interest, we have \(\delta=0\), it suffices to use the special Wright function \[ \phi(\rho, 0, z) := \sum_{k=1}^\infty {\frac{z^k}{k! \Gamma(k \rho)}}. \] Leaning on the authors' abstract, the essential can be shortly characterized as follows. New Wright function representations for the densities of generating measures of the power variance family are derived. New saddle-point type approximations are constructed, with the aid of newly invented ``exponentially small'' Poincaré series for a subclass of Wright functions expressed via Zolotarev polynomials.