1: \begin{abstract}%

2: We find the minimax rate of convergence 

3: in Hausdorff distance

4: for

5: estimating a manifold $M$ of dimension $d$

6: embedded in $\mathbb{R}^D$

7: given a noisy sample from the manifold.

8: Under certain conditions,

9: we show that the optimal rate of convergence is

10: $n^{-2/(2+d)}$.

11: Thus, the minimax rate depends only

12: on the dimension of the manifold, not on the

13: dimension of the space in which $M$ is embedded.

14: \end{abstract}

15: