On approximative solutions of multistopping problems

DOI10.1214/10-AAP747MaRDI QIDQ655588zbMATH OpenOpenAlexFDO

Authors Andreas Faller, Ludger Rüschendorf

Publication date 4 January 2012

Published in The Annals of Applied Probability (Search for Journal in Brave)

Full work available at URL https://arxiv.org/abs/1201.0083

zbMATH Keywords

best choice problem extreme values optimal multiple stopping optimal stopping Poisson process

Mathematics Subject Classification ID

Stopping times; optimal stopping problems; gambling theory (60G40) Optimal stopping in statistics (62L15)

Abstract: In this paper, we consider multistopping problems for finite discrete time sequences

X_{1}, ..., X_{n}

.

m

-stops are allowed and the aim is to maximize the expected value of the best of these

m

stops. The random variables are neither assumed to be independent not to be identically distributed. The basic assumption is convergence of a related imbedded point process to a continuous time Poisson process in the plane, which serves as a limiting model for the stopping problem. The optimal

m

-stopping curves for this limiting model are determined by differential equations of first order. A general approximation result is established which ensures convergence of the finite discrete time

m

-stopping problem to that in the limit model. This allows the construction of approximative solutions of the discrete time

m

-stopping problem. In detail, the case of i.i.d. sequences with discount and observation costs is discussed and explicit results are obtained.

Review text

This paper deals with multiple stopping problems for finite, discrete time, random sequences without independence or identical distributions assumption. \(m\)-stops are allowed and the aim is to maximize the expected value of the best of these \(m\) stops, i. e., the sequence of stopping times \(1\leq \tau_1^\star\leq\ldots\leq\tau_m^\star\leq n\) with respect to the natural filtration related to the observed sequence \(\{X_n\}_{i=1}^n\) such that \[ {\mathbf E}[\displaystyle{\max_{1\leq i\leq m}}X_{\tau_i^\star}]=\sup_{1\leq \tau_1\leq\ldots\leq\tau_m\leq n}{\mathbf E}[\max_{1\leq i\leq m}X_{\tau_i}]. \] The basic assumption is convergence of a related embedded point process to a continuous time Poisson process in the plane, which serves as a limiting model for the stopping problem. The optimal \(m\)-stopping curves for this limiting model are determined by differential equations of first order. A general approximation result is established which ensures convergence of the finite discrete time \(m\)-stopping problem to that in the limit model. This allows the construction of approximative solutions of the discrete time m-stopping problem. The case of i.i.d. sequences with discount and observation costs is discussed and explicit results are obtained in details. The related results for optimal stopping problem have been obtained by \textit{R. Kühne} and \textit{L. Rüschendorf} [Theory Probab. Appl. 48, No. 3, 465--480 (2003) and Teor. Veroyatn. Primen. 48, No. 3, 557--575 (2003; Zbl 1079.60043)] and the authors [Adv. Appl. Probab. 43, No. 4, 1086--1108 (2011; Zbl 1235.60038)].

Recommendations

Cites work

Cited in

(8)

This page was built for publication: On approximative solutions of multistopping problems

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q655588)