No label defined (Q3979564)

The problem of sequential estimation in a class of random processes having exponential type likelihood functions is considered. Basing on modifications of the characterization theorems of \textit{P. Diaconis} and \textit{D. Ylvisaker} [Ann. Stat. 7, 269-281 (1979; Zbl 0405.62011)], the properties of conjugate priors given by \textit{C. N. Morris} [ibid. 11, 515-529 (1983; Zbl 0521.62014)] and the results concerning the moments of first passage times for such processes obtained by \textit{V. T. Stefanov} [Stochastics 19, 207-220 (1986; Zbl 0608.60046)], the form of admissible estimators, under quadratic loss function, for a quadratic parameter function is derived. The results generalize those presented by the author in Sequential Anal. 6, 207-218 (1987; Zbl 0634.62077)]. The outcomes are applied to construct admissible estimators of linear parameter functions for a compound Poisson process, the multinomial process, and a Markov branching process. For a linear birth-death process the form of admissible sequential estimators for quadratic functions of these parameters is given.