Stability Problems for Stochastic Models

Klappentext

The subject of this book is a new direction in the field ofprobability theory and mathematical statistics which can becalled "stability theory": it deals with evaluating theeffects of perturbing initial probabilistic models andembraces quite varied subtopics: limit theorems, queueingmodels, statistical inference, probability metrics, etc. Thecontributions are original research articles developing newideas and methods of stability analysis.

Inhalt
On periodic distribution of waiting-time process.- Asymptotic behaviour of some random fields.- Normality of a multidimensional infinitely divisible distribution which coincides with a normal distribution in a cone.- Partial sufficiency of linear forms.- A refinement of Darmois-Skitovitch and Heyde theorems.- The rate of convergence of Erlang distribution to the degenerate one.- Metric estimates of the first occurence time in regenerative processes.- Stability of the lack-of-memory property in a finite number of points.- The central limit theorem and weak convergence of maxima of sums of independent random variables.- Compound metrics with fixed minimal ones.- On limit theorems for the number of Markov renewals.- Some characterization of bivariate distributions based on hazard rate properties.- Confidence intervals for the parameters of strongly stable laws.- Generalization of the Titchmarsh convolution theorem and the complex-valued measures uniquely determined by their restrictions to a half-line.- Limit theorems for extreme order statistics under nonlinear normalization.- A characterization of Gaussian distribution by the linearity of conditional moments.- Extreme functionals in the space of probability measures.- On the ratio of independent stable random variables.- On the Berry-Esseen bound for Student's statistic.- Some new results in a theorem due to Renyi.- Stability for continuous-time nonhomogeneous Markov chains.- Rate of convergence in limit theorems for the max-scheme.- On compound probabilistic metrics with given minimal metrics.