1: \begin{abstract}

2: We consider the model of nonregular nonparametric regression where

3: smoothness constraints are imposed on the regression function $f$ and

4: the regression errors are assumed to decay with some sharpness level at

5: their endpoints. The aim of this paper is to construct an adaptive

6: estimator for the \mbox{regression} function $f$. In contrast to the standard

7: model where local averaging is fruitful, the nonregular conditions

8: require a substantial different treatment based on local extreme

9: values. We study this model under the realistic setting in which both

10: the smoothness degree $\beta> 0$ and the sharpness degree $\mathfrak

11: {a}\in(0, \infty)$ are unknown in advance. We construct adaptation

12: procedures applying a nested version of Lepski's method and the

13: negative Hill estimator which show no loss in the convergence rates

14: with respect to the general \mbox{$L_q$-}risk and a logarithmic loss with

15: respect to the pointwise risk. Optimality of these rates is proved for

16: $\mathfrak{a}\in(0, \infty)$. Some numerical simulations and an

17: application to real data are provided.

18: \end{abstract}