1: \begin{abstract}

2: We investigate the nonparametric estimation for regression in a

3: fixed-design setting when the errors are given by a field of

4: dependent random variables. Sufficient conditions for kernel

5: estimators to converge uniformly are obtained. These estimators

6: can attain the optimal rates of uniform convergence and the

7: results apply to a large class of random fields which contains

8: martingale-difference random fields and mixing random fields.

9: \vspace{8cm}\\

10: \hspace{-0.7cm}{\em AMS Subject Classifications} (2000): 60G60, 62G08\\

11: {\em Key words and phrases:} nonparametric regression estimation,

12: kernel estimators, strong consistency, fixed-design, exponential

13: inequalities, martingale difference random fields, mixing, Orlicz spaces.\\

14: {\em Short title:} Nonparametric regression in a fixed design.

15: \end{abstract}

16: