1: \begin{abstract}

2: Accurate reconstruction of piecewise-smooth functions from a finite

3: number of Fourier coefficients is an important problem in various

4: applications. This probelm exhibits an inherent inaccuracy, in particular

5: the Gibbs phenomenon, and it is being intensively investigated during

6: the last decades. Several nonlinear reconstruction methods have been

7: proposed in the literature, and it is by now well-established that

8: the ``classical'' convergence order can be completely restored up

9: to the discontinuities. Still, the maximal accuracy of determining

10: the positions of these discontinuities remains an open question.

11: 

12: In this paper we prove that the locations of the jumps (and subsequently

13: the pointwise values of the function) can be reconstructed with at

14: least ``half the classical accuracy''. In particular, we develop

15: a constructive approximation procedure which, given the first $k$

16: Fourier coefficients of a piecewise-$C^{2d+1}$ function, recovers

17: the locations of the jumps with accuracy $\sim k^{-\left(d+2\right)}$,

18: and the values of the function between the jumps with accuracy $\sim k^{-\left(d+1\right)}$

19: (similar estimates are obtained for the associated jump magnitudes).

20: A key ingredient of the algorithm is to start with the case of a single

21: discontinuity, where a modified version of one of the existing algebraic

22: methods (due to K.Eckhoff) may be applied. It turns out that the additional

23: orders of smoothness produce highly correlated error terms in the

24: Fourier coefficients, which eventually cancel out in the corresponding

25: algebraic equations. To handle more than one jump, we apply a localization

26: procedure via a convolution in the Fourier domain, which eventually

27: preserves the accuracy estimates obtained for the single jump. We

28: provide some numerical results which support the theoretical predictions.

29: \end{abstract}

30: