Quasi-stationary phenomena in nonlinearity perturbed stochastic systems. (Q951132)

The book of Mats Gyllenberg and Dmitrii S. Silvestrov is devoted to problems of quasi-stationarity for processes perturbed by a small parameter. The authors deal with renewal, regenerating and semi-Markov processes. In every these classes of processes there exists a lot of practical examples of application. In the book some of them are treated, for example, queueing theory, population dynamic, risk processes. The idea to learn perturbed processes relative to an initial one leads to necessity to investigate a family of processes of given type, differing from the initial one by their vectors of parameters, belonging to a small neighborhood of the initial vector. Passage to the family of processes of similar type opens lots of new problems, e.g. that of asymptotic character under simultaneous tendency \(t\to\infty\) and \(\varepsilon\to 0\), where \(t\) is time and \(\varepsilon\) is a small parameter, characterizing closeness to the initial process. The authors have success to get non-trivial results in the problem of quasi-stationarity, i.e. to find an index of exponent for a non-degenerated limit of a degenerating one-dimensional distribution, normalized with an exponentially decreasing factor, to exist. Actually it concerns about a proof of an ergodic theorem for the conditional distribution with respect to a condition with probability, converging to zero. The authors' results permit to take into account a rate of coming together parameters of the perturbed and initial processes, which is given by nonlinear function of the parameter \(\varepsilon\). Besides of an original material the book contains a reach reference information on processes what are dealt with, and also a wide bibliography (1094 items) on affected theme.