Limit theorems for the population size of a birth and death process allowing catastrophes (Q1101372): Difference between revisions

A quite general linear birth and death process with catastrophes \(X_ t(BDCP)\), which is a continuous-time right-continuous random walk on \(\{\) 0,1,2,...\(\}\) with instantaneous jump rates proportional to \(X_ t\), is considered. \textit{I. I. Ezhov} and \textit{V. N. Reshetnyak} [Ukr. Mat. Zh. 35, No.1, 31-36 (1983; Zbl 0531.60081); English translation in Ukr. Math. J. 35, 28-33 (1983)] formulated that process and studied the tail behaviour of the extinction time T, obtaining, in the way, an identity relating \(P(T>t| X_ 0=1)\) to a similar quantity for a certain Markov branching process (MBP) \(\hat X_ t\). The connection is shown to be deeper, the transition probabilities of the two processes also being closely related. Such a relation is used to obtain new properties of the BDCP or similar properties to those obtained by Ezhov and Reshetnyak under weaker conditions. In particular, \(E(X_ t| X_ 0=i)\), \(P(X_ t>0| X_ 0=i)\) are studied, along with their asymptotic behaviour as \(t\to \infty\). The limiting distribution of \(X_ t\) is obtained, in some cases conditioned on non-extinction. A Q-process corresponding to the BDCP is shown to exist under mild regularity assumptions and its limiting behaviour is studied.