Point processes competing for runs: A new tool for their investigation (Q1807740): Difference between revisions

A right-continuous semi-Markov process \(\xi(t)\) \((t\geq 0)\) with a finite set of states is considered. The set consists of two non-empty subsets \(S_1\) and \(S_2\). Let \((\tau_k)\) \((k\geq 1)\) be a sequence of jump times of the process. The number \(k> r_1\) is said to be the end of a 1-run iff \(\xi(\tau_k)\in S_2\) and \((\forall i:1\leq i\leq r_1)\) \(\xi(\tau_i)\in S_1\). Similarly, the end of 2-run is defined with respect to a series of \(r_2\) points of \(S_2\). Let \(R_1\) and \(R_2\) be (random) sets of all ends of 1-runs and 2-runs, respectively; \(N_1= \min\{R_1\cup R_2\}\); \(\sigma_1= \tau_{N_1}\); \(\sigma_n\) be the \(n\)th iteration of \(\sigma_1\). The embedded semi-Markov process corresponding to the Markov chain \((\sigma_k,\xi(\sigma_k))\) \((k\geq 1)\) is investigated. In terms of the initial semi-Markov process the semi-Markov transition function of the embedded process is derived when the initial process is a stepped Markov process with a given matrix of intensities. A vector of stationary probabilities of the embedded process is obtained. Some descriptors of the stationary process are determined, such as durations of a typical 1-run, 1-circle, 2-runs in a 1-circle, and their distributions and moments. The peculiarity of the paper is its purely matrix language of all the presentation. Results of numerical modelling are proposed in form of graphics and tables.