Optimal stopping in a cumulative damage model (Q2640434): Difference between revisions
From MaRDI portal
Revision as of 14:30, 21 June 2024
scientific article
Language | Label | Description | Also known as |
---|---|---|---|
English | Optimal stopping in a cumulative damage model |
scientific article |
Statements
Optimal stopping in a cumulative damage model (English)
0 references
1991
0 references
This paper is concerned with the optimal stopping problem in the following cumulative damage model: let \(\xi_ i\) be the i-th shock amount of damage where shocks occur at random points of time, and it is assumed that \(\xi_ i\), \(i=1,2,..\). are independent and identically distributed non-negative random variables with the common distribution function F, and that the sequence \(\{\xi_ i\}\) and the process counting the number of shocks are independent. Let the cumulative damage process be \(S_ n=\xi_ 1+...+\xi_ n\), the tolerance level a fixed constant \(\beta >0\), \({\mathcal F}_ n\) the \(\sigma\)-field generated by \(\xi_ 1,...,\xi_ n\), and let \({\mathcal F}_ 0\) be the trivial \(\sigma\)-field. Also denote by \({\mathcal I}\) the class of all stopping times \(\tau\) with respect to (\({\mathcal F}_ n)_{n\geq 0}\) such that \(P(\tau <\infty)=1\), and for some \(\epsilon\) in [0,1) set \({\mathcal I}_{\epsilon}=\{\tau \in {\mathcal I}|\) \(P(S_{\tau}>\beta)\leq \epsilon \}\). Then the aim is to find a stopping time in \({\mathcal I}_{\epsilon}\) which maximizes \(E[S_{\tau}| S_{\tau}\leq \beta]\). When F is continuous and \(F(\beta)>0\), it is shown that for \(\epsilon\in (1-F(\beta),1)\) a stopping time \(\sigma =\inf \{n\geq 1| S_ n\geq s_{\epsilon}\}\) is optimal, where \(s_{\epsilon}\in (0,\beta)\) is a solution to the equation \[ 1-F(\beta)+\int^{s}_{0}(1-F(\beta -t))dU(t)=\epsilon, \] letting U be the renewal measure associated to F. For other \(\epsilon\in [0,1)\), trivial optimal stopping is given. To solve this optimal stopping problem the author in general deals with the problem of the maximization of \(E[f(S_{\tau})I_{(S_{\tau}\leq \beta)}+\alpha I_{(S_{\tau}>\beta)}]\) in \({\mathcal I}\), where f is a concave, non- decreasing, non-negative function on [0,\(\beta\) ] and \(\alpha\) is a constant satisfying \(\alpha <f(\beta)\). This problem is reduced to the theory of the monotone case of \textit{Y. S. Chow, H. Robbins} and \textit{D. Siegmund} [``Great expectations: The theory of optimal stopping'' (1971; Zbl 0233.60044)], and is solved. By similar discussions the author also treats another problem: minimize \(E[g(| S_{\tau}-\beta |)]\) in \({\mathcal I}\) where g is a convex function from \([0,\infty)\) to \([0,\infty)\) such that \(g(0)=0\) and \(g(S_ n)\) is integrable for all \(n\geq 1\).
0 references
optimal stopping
0 references
cumulative damage model
0 references
tolerance level
0 references