1: \begin{abstract}

2: Optimal sampling of non band-limited functions is an issue of great

3: importance that has attracted considerable attention. We propose to

4: tackle this problem through the use of a frequency warping: First, by

5: a non-linear shrinking of frequencies, the function is transformed

6: into a band-limited one.  One may then perform a decomposition in

7: Fourier series. This process gives rise to new orthonormal bases of

8: the Sobolev spaces ${\mathcal H}^\alpha$. When $\alpha$ is an integer,

9: these orthonormal bases can be expressed in terms of Laguerre

10: functions.  We study the reconstruction and speed of convergence

11: properties of the warping-based sampling. Besides theoretical

12: considerations, numerical experiments are performed.

13: \end{abstract}

14: