Probabilités de présence dans un processus de branchement spatial markovien. (Presence probability in a spatial Markov branching process)

Large deviations (60F10) Point processes (e.g., Poisson, Cox, Hawkes processes) (60G55) Branching processes (Galton-Watson, birth-and-death, etc.) (60J80)

Review text

Given the point processes \(\{M_{t,x}:\) \(t\geq 0\), \(-\infty <x<\infty \}\), a sequence of spatial branching processes, indexed by a small parameter \(\epsilon\) is constructed in the following way. Let \(S_{\epsilon}M\) be obtained by contracting the point process M by \(\epsilon\) ; so, if m is a point of M, \(\epsilon\) m is a point of \(S_{\epsilon}M\). Now if a kth generation person is at x, his children's positions, relative to his own, are given by an independent copy of \(S_{\epsilon}M_{k\epsilon,x}.\) Let \(P^{\epsilon}_{k,x}\) be the law of the process resulting from a single kth generation person at x. If \(\zeta_ n\) is the point process of the nth generation people then the main objective of this paper is to establish lower bounds on \[ \liminf \epsilon \log (P^{\epsilon}_{0,x}(\zeta_ n[y-u,y+u]>0), \] as \(\epsilon\to 0\), where \(n=[t/\epsilon]\) and \(u=\sigma \sqrt{\epsilon}\). Obviously interest centres on y values for which the probability here is small (and these, in a sense, constitute the subcritical region of the process). Techniques from branching processes and large deviation theory are used; the lower bounds obtained arise from 'action integrals' defined through the Cramér transform of the intensity measure of \(M_{t,x}\).

Recommendations

Cited in

(8)

This page was built for publication: Probabilités de présence dans un processus de branchement spatial markovien. (Presence probability in a spatial Markov branching process)

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q1087250)