Moments of suprema of random variables |
| |
| Authors: | Stuart Scott |
| |
| Affiliation: | Department of Applied Mathematics , University of Western Ontario , London, Ontario, N6A 5B9 |
| |
| Abstract: | The supremum of random variables representing a sequence of rewards is of interest in establishing the existence of optimal stopping rules. Necessary and sufficient conditions are given for existence of moments of supn(Xn ? cn) and supn(Sn ? cn) where X1, X2, … are i.i.d. random variables, Sn = X1 + … + Xn, and cn = (nL(n))1/r, 0 < r < 2, L = 1, L = log, and L = log log. Following Cohn (1974), “rates of convergence” results are used in the proof. |
| |
| Keywords: | rates of convergence optimal stopping largest excess |
|
|
