Explicit formulae for stationary distributions of stress release processes
A general stress release model assumes linear build-up of stress and its stochastic release in ``events occurring with a rate depending on the current stress event. For a general model see M. H. A. Davis (l984). The version of the stress release process considered by the authors can be described as follows. The current stress level is represented by a real-valued process \(\{X(t)\}\). Due to the tectonic forces it increases between the events with a constant rate \(\lambda>0\). The occurrence of ``events is specified by the exponential risk function \(\psi(x)= \exp(\beta (x-x_0))\), \(x\in\mathbb{R}\), \(\beta\) and \(x_0\) being the parameters of the model. To model the physical situation, the risk function should rise rapidly and nonlinearly as the stress approaches a critical breaking level. The amounts of stress \(\xi_j>0\) released in consequent events are assumed to be independent and identically distributed r.v.s following a common distribution function \(J\). The function \(J\) is truncated Pareto distribution with \[ Jdx= \begin{cases} Cx^{-\omega-1} dx\text{ if }0< C_1<x <C_2< \infty, \\0 \text{ otherwise}, \end{cases} \] with \(\omega\) close to 1, see Gutenberg and Richter (1954), Kagan (1997). Thus \(X(t)\) is a Markov process with the generator \[ Ah(x)=\lambda h'(x)+ \psi(x)\bigl( Eh(x-\xi) -h(x) \bigr). \] \(X (t)\) is ergodic Harris process. Also there exists an invariant distribution \(F\) such that the distribution of \(X(t)\) converges to \(F\) in total variation as \(t \to \infty\) whatever the initial state \(X(0)\). From the point of view of applications it is of interest to know the properties even the form of stationary distribution \(F\). The dependence structure of the stationary process is of interest as well. In this note both problems are solved. The authors introduce the following non-restrictive technical assumption: for some \(\alpha> 0\), \(J(x)= O(x^\alpha)\) as \(x\to 0\). Also the authors assume that the distribution \(J\) has a finite mean and set \[ \mu=\int^\infty_0 xJ(dx), \quad \chi(s)=\int^\infty_0 e^{-isx}J(dx),\quad \gamma(s)= \bigl(1-\chi (s) \bigr)/i \mu s. \] Furthermore, the authors prove two interesting theorems. For other details see the authors' references.
- Dynamics of the time to the most recent common ancestor in a large branching population
- TCP and iso-stationary transformations
- On Rice's formula for stationary multivariate piecewise smooth processes
- Hitting times and the running maximum of Markovian growth-collapse processes
- On level crossings for a general class of piecewise-deterministic Markov processes
- Ergodicity of a stress release point process seismic model with aftershocks
- scientific article; zbMATH DE number 4107861 (Why is no real title available?)
- Simulation and estimation procedures for stress release models
- Ergodicity properties of stress release, repairable system and workload models
- Exact simulation of extrinsic stress-release processes
- Ergodic theorems for stress release processes
This page was built for publication: Explicit formulae for stationary distributions of stress release processes
Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q4519096)
