1: \begin{abstract}

2: The paper considers functional linear regression, where scalar

3: responses $Y_1,\ldots,Y_n$ are modeled in dependence of i.i.d. random functions

4: $X_1,\ldots, X_n$.

5: We study

6: a generalization of the classical functional linear regression model.

7: It is assumed that there exists an unknown number of ``points of

8: impact,'' that is, discrete observation times where the corresponding

9: functional values possess significant influences on the response

10: variable. In addition to estimating

11: a functional slope parameter, the problem then is to determine the

12: number and locations of points of impact

13: as well as corresponding regression coefficients.

14: Identifiability of the generalized model is considered in detail. It is

15: shown that points of impact are

16: identifiable if the underlying process generating $X_1,\ldots,X_n$

17: possesses ``specific local variation.''

18: Examples are well-known processes like the Brownian motion, fractional

19: Brownian motion or the

20: Ornstein--Uhlenbeck process.

21: The paper then proposes an easily implementable method for estimating

22: the number and locations of points of impact. It is shown that this

23: number can be estimated consistently. Furthermore, rates of convergence for

24: location estimates, regression coefficients and the slope parameter are derived.

25: Finally, some simulation results as well as a real data application are

26: presented.

27: \end{abstract}