The recoverability limit for superresolution via sparsity
Abstract
We consider the problem of robustly recovering a -sparse coefficient vector from the Fourier series that it generates, restricted to the interval . The difficulty of this problem is linked to the superresolution factor SRF, equal to the ratio of the Rayleigh length (inverse of ) by the spacing of the grid supporting the sparse vector. In the presence of additive deterministic noise of norm , we show upper and lower bounds on the minimax error rate that both scale like , providing a partial answer to a question posed by Donoho in 1992. The scaling arises from comparing the noise level to a restricted isometry constant at sparsity , or equivalently from comparing to the so-called -spark of the Fourier system. The proof involves new bounds on the singular values of restricted Fourier matrices, obtained in part from old techniques in complex analysis.
Keywords
Cite
@article{arxiv.1502.01385,
title = {The recoverability limit for superresolution via sparsity},
author = {Laurent Demanet and Nam Nguyen},
journal= {arXiv preprint arXiv:1502.01385},
year = {2015}
}
Comments
19 pages