1: \begin{abstract}

2: An algorithm to generate random points inside an arbitrary

3: $d$--dimensional convex body $X$ with respect to the flat (Lebesgue) measure

4: is proposed. It can be considered as an

5: iterated functions system (IFS) with an infinite number of functions

6: acting on $X$.  We analyze the corresponding Markov operator which acts on the

7: probability measures, and show that any initial

8: measure converges exponentially to the uniform measure.

9: Estimations for the convergence rate are derived in terms

10: of the dimension $d$ and  the ratio between the radius of the sphere

11: inscribed in $X$ and the radius of the outscribed sphere.

12: Concrete estimations are provided for the Birkhoff polytope

13: containing bistochastic matrices, the set of quantum states

14: acting on $N$--dimensional Hilbert space and its subset

15: consisting of states with positive partial transpose.

16: \end{abstract}

17: