A note on multiple stopping rules
The paper discusses some details of the multiple stopping problem as introduced by \textit{W. Stadje} [Optimization 16, 401--418 (1985; Zbl 0571.60057)]. Let \({\mathfrak F}_1,{\mathfrak F}_2,\dots\) be an increasing sequence of \(\sigma\)-fields on \(\Omega\), where \((\Omega,{\mathfrak F},\mathbf{P})\) is a probability space and \({\mathfrak F}_n\subset {\mathfrak F}\) for all \(n\in\mathbb N\). Define \((X_{n_1,n_2,\dots,n_k})_{1\leq n_1<n_2<\dots n_k}\) a family of random variables such that \(X_{n_1,n_2,\dots,n_k}\) is \({\mathfrak F}_{n_k}\)-measurable and integrable. A \(k\)-stopping rule is a sequence \((\tau_1,\tau_2,\dots,\tau_k)\) of \(k\) stopping times such that \(\tau_1<\tau_2<\dots<\tau_k\). The \(k\)-stopping problem is to find a \(k\)-stopping rule \((\sigma_1,\sigma_2,\dots,\sigma_k)\) satisfying \(\mathbf{E}X_{\sigma_1,\sigma_2,\dots,\sigma_k} = \sup_{(\tau_1,\dots,\tau_k)} \text\textbf{E}X_{\tau_1,\tau_2,\dots,\tau_k}\), where the supremum is taken over all \(k\)-stopping rules \((\tau_1,\dots,\tau_k)\) for which \(\mathbf{E}X_{\tau_1,\tau_2,\dots,\tau_k}\) exists. \(V=\mathbf{E}X_{\sigma_1,\sigma_2,\dots,\sigma_k} \) is called the value of \((X_{n_1,\dots,n_k})\). The first theorem of Stadje [loc. cit.] claims that if \(V<\infty\) and for all \(l\in\{0,1,\dots,k-1\}\) and all fixed \(n_1^0<\dots< n_l^0\) an optimal \((k-l)\)-stopping rule for \((X_{n_1^0,\dots,n_l^0,n_{l+1},\dots,n_k})_{n_l^0<n_{l+1}<\dots <n_k}\) exists, \((\sigma_1,\dots,\sigma_k)\) is well-defined, optimal and belongs to \(C=\{(\tau_1,\dots,\tau_k): \mathbf{E}X^{-}_{\tau_1,\tau_2,\dots,\tau_k}\}<\infty\). An example is shown that the identity \[ \begin{multlined} \mathop{\text{ess\;sup}}_{m\leq\tau_{k-1}} \mathbf{E}(X_{n_1,n_2,\dots,n_{k-2},\tau_{k-1},\tilde{\sigma}_{n_1,n_2,\dots ,n_{k-2},\tau_{k-1}}}| {\mathfrak F}_m) \!\!\!= \!\!\!\mathop{\text{ess\;sup}}_{m\leq\tau<\bar{\tau}} \text\textbf{E}(X_{n_1,n_2,\dots,n_{k-2},\tau,\bar{\tau}}| {\mathfrak F}_m) \end{multlined} \] for \(m\geq n_{k-2}\), used in the proof of Theorem 1 of Stadje [loc. cit.] need not hold in general and the constructed strategy need not belong to \(C\). However, it is proved that if \(\mathbf{E}X_{\sigma_1,\sigma_2,\dots,\sigma_k}\) exists, then \((\sigma_1,\dots,\sigma_k)\in C\) and this \(k\)-stopping rule is optimal. Under a suitable integrability condition, like in the proof of Theorems 2 and 3 of Stadje [loc. cit.], the mentioned problem does not appear. The reviewer would like to remark that the detailed definitions are not provided in the paper and the results of the paper can be followed having Stadje [loc. cit.].
- An optimal double stopping rule for a buying-selling problem
- A multiple optimal stopping rule for a buying-selling problem with a deterministic trend
- Construction of Nash equilibrium based on multiple stopping problem in multi-person game
- Multiple-stopping problems with random horizon
- Pareto optimality in a bicriterion optimal stopping problem
- Construction of Nash equilibrium in a game version of Elfving's multiple stopping problem
- On stopping rules in proofreading
- Analysis of the ``hiring above the median selection strategy for the hiring problem
- scientific article; zbMATH DE number 1865420 (Why is no real title available?)
- Analysis of the strategy ``hiring above the \(m\)-th best candidate
This page was built for publication: A note on multiple stopping rules
Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q4656059)
