The asymptotic behavior of extrema of compound Cox processes with nonzero means (Q1586595): Difference between revisions

Let \(N_1(t)\) be a standard Poisson process with intensity \(1\), \(\Lambda(t)\) be a random process with non-decreasing right-continuous trajectories, \(\Lambda(0)=0\), \({\mathbf P}(\Lambda(t)<\infty)=1\) for any \(t\); \(\Lambda\) and \(N_1\) are independent. The process \(N(t)=N_1(\Lambda(t))\) is called a doubly Poisson (Cox) process. The author considers a compound doubly stochastic Poisson (Cox) process \(S(t)=\sum_{k=1}^{N(t)}X_k\), where \(X_k\) are i.i.d. r.v. independent of \(N\) with \({\mathbf E}X_k=a>0\), \({\mathbf D}X_k=\sigma>0\). The main result concernes the process of extrema \(\overline S(t)=\max_{0\leq\tau\leq t}S(\tau)\). It is proven that \[ \frac{\overline S(t)-C(t)}{D(t)}\Longrightarrow Z, \quad t\to\infty , \] with conditions on the real functions \(C(t)\) and \(D(t)\), and r.v.\ \(Z\) given. It is assumed that the control process \(\Lambda(t)\to\infty\) in \({\mathbf P}\) but no assumptions on \({\mathbf E}\Lambda(t)\) are imposed. The rate of convergence is also estimated. See also \textit{V. E. Bening} and the author [J. Math. Sci., New York 92, No. 3, 3836-3856 (1998; Zbl 0919.60054)] and \textit{B. V. Gnedenko} and the author [``Random summation. Limit theorems and applications'' (1996; Zbl 0857.60002)].