An optimal stopping problem with finite horizon for sums of I.I.D. random variables
The problem of selling a commodity optimally at one of n successive time instants leads to the optimal stopping problem for the finite sequence ((n-j)lSj)1[less-than-or-equals, slant]j[less-than-or-equals, slant]n, where Sj=U1 + ... + Uj, U1, U2,... are i.i.d., E(U1) = 0 and E(U21) = 1. The optimal stopping time [pi]n is seen to be of the form [tau]n = inf{jj = n or j < n, Sj[greater-or-equal, slanted]clj,n}, where c1j,1>...>cln-1,n = 0 , if is the solution of the equation . For the value vln we have . vl is explicitly computed. In the normal case we also obtain results on the speed of convergence of and .
Year of publication: |
1987
|
---|---|
Authors: | Stadje, Wolfgang |
Published in: |
Stochastic Processes and their Applications. - Elsevier, ISSN 0304-4149. - Vol. 26.1987, p. 107-121
|
Publisher: |
Elsevier |
Keywords: | optimal stopping salesman problem finite horizon |
Saved in:
Saved in favorites
Similar items by person
-
Some remarks on the average sample number of sequential and non-sequential tests
Stadje, Wolfgang, (1982)
-
ON THE THEORY OF ROTATIONALLY SYMMETRIC RANDOM VECTORS
Stadje, Wolfgang, (1984)
-
Annuities with controlled random interest rates
Perry, David, (2003)
- More ...