1: \begin{abstract}

2:   This paper studies function approximation for finite horizon

3:   discrete time Markov decision processes under certain convexity

4:   assumptions. Uniform convergence of these approximations on compact

5:   sets is proved under several sampling schemes for the driving random

6:   variables. Under some conditions, these approximations form a

7:   monotone sequence of lower or upper bounding functions. Numerical

8:   experiments involving piecewise linear functions demonstrate that

9:   very tight bounding functions for the fair price of a Bermudan put

10:   option can be obtained with excellent speed (fractions of a cpu

11:   second). Results in this paper can be easily adapted to minimization

12:   problems involving concave Bellman functions.

13: \end{abstract}