The asymptotic behavior of the reward sequence in the optimal stopping of i.i.d. random variables (Q756259)
From MaRDI portal
scientific article
Language | Label | Description | Also known as |
---|---|---|---|
English | The asymptotic behavior of the reward sequence in the optimal stopping of i.i.d. random variables |
scientific article |
Statements
The asymptotic behavior of the reward sequence in the optimal stopping of i.i.d. random variables (English)
0 references
1991
0 references
Let \(X_ 1,X_ 2,..\). be integrable i.i.d. random variables with a distribution function F and let \(v_ n=\sup \{E(X_ T):\) T is a stopping time for \((X_ m)_{m\geq 1}\), \(T\leq n\}\), \(n=1,2,... \). It is proved that if F belongs to the domain of attraction for maxima of one of the three extreme-value laws \(G_ I\) or \(G^{\alpha}_{II}\) \((\alpha >1)\) or \(G^{\alpha}_{III}\) \((\alpha >0)\), then \(\lim_{n}n(1-F(v_ n))=1\) or \(1-\alpha^{-1}\) or \(1+\alpha^{-1}\), respectively. From this, the growth rate of \((v_ n)_{n\geq 1}\) is derived and compared to the growth rate of \((E(\max (X_ 1,...,X_ n)))_{n\geq 1}\). Under the same assumption on F, let \((u_ n)_{n\geq 1}\) be any sequence such that \(\lim_{n}n(1-F(u_ n))=\lim_{n}n(1-F(v_ n)).\) Then the sequence of stopping times \(T_ 1=1\), \(T_ n=n\wedge \inf \{k:\;k<n,\quad X_ k>u_{n-k}\}\) is ``asymptotically optimal''. Finally, for the optimal sequence \((T^*_ n)_{n\geq 1}\) (i.e., if \(v_ n\) is put instead of \(u_ n)\) the limit distribution of the optimal rewards \((X_{T^*_ n})_{n\geq 1}\) is calculated.
0 references
domain of attraction for maxima
0 references
extreme-value laws
0 references
stopping times
0 references
optimal rewards
0 references