1: \begin{abstract}

2: We study in this paper a smoothness regularization method for

3: functional linear regression and provide a unified treatment for both

4: the prediction and estimation problems. By

5: developing a tool on simultaneous diagonalization of two positive

6: definite kernels, we obtain shaper results on the minimax rates of

7: convergence and show that smoothness regularized estimators

8: achieve the optimal rates of convergence for both prediction and

9: estimation under conditions weaker than those for the

10: functional principal components based methods developed in the literature.

11: Despite the generality of the method of regularization, we show that

12: the procedure is easily implementable. Numerical results are obtained to

13: illustrate the merits of the method and to demonstrate the theoretical

14: developments.

15: \end{abstract}