Shelf life of candidates in the generalized secretary problem (Q1694784): Difference between revisions
From MaRDI portal
Added link to MaRDI item. |
Set profile property. |
||
Property / MaRDI profile type | |||
Property / MaRDI profile type: MaRDI publication profile / rank | |||
Normal rank |
Revision as of 04:21, 5 March 2024
scientific article
Language | Label | Description | Also known as |
---|---|---|---|
English | Shelf life of candidates in the generalized secretary problem |
scientific article |
Statements
Shelf life of candidates in the generalized secretary problem (English)
0 references
6 February 2018
0 references
This paper presents a survey on the duration problems (see [\textit{T. S. Ferguson} et al., Contemp. Math. 125, 37--57 (1992; Zbl 0745.62079)]) and a new solution to a problem of this type. In the classical duration problem, a decision maker observes a sequence of observations that can be ranked with respect to the observations that have been seen so far. Only one object can be taken. The payoff of the decision maker is the time for which the chosen object remains the best seen so far (from the moment when it is chosen). The authors present the solution to the problem in which the payoff of the decision maker is the time for which the object chosen remains one of the best two observations seen so far. A precise mathematical model is constructed and a complete solution is provided for the problem. The optimal strategy is of threshold type, i.e. it immediately selects the best candidate if it appears after or on time \(k_1^*\) and it immediately selects the second best candidate if it appears after time \(k_2^*>k_1^*\). For \(n\rightarrow\infty\), \(\frac{k_1^*}{n}\) converges to \(0.120381\dots\) and \(\frac{k_2^*}{n}\) to \(0.417188\dots\). The asymptotic mean time of shelf life of the relatively best or the second best object is \(0.403827\dots N\).
0 references
best-choice duration model
0 references
no-information case
0 references
secretary problem
0 references
relative best rank
0 references
recalls of preceding observations
0 references