English

Super-resolution limit of the ESPRIT algorithm

Information Theory 2019-10-23 v3 math.IT

Abstract

The problem of imaging point objects can be formulated as estimation of an unknown atomic measure from its M+1M+1 consecutive noisy Fourier coefficients. The standard resolution of this inverse problem is 1/M1/M and super-resolution refers to the capability of resolving atoms at a higher resolution. When any two atoms are less than 1/M1/M apart, this recovery problem is highly challenging and many existing algorithms either cannot deal with this situation or require restrictive assumptions on the sign of the measure. ESPRIT is an efficient method that does not depend on the sign of the measure. This paper provides an explicit error bound on the support matching distance of ESPRIT in terms of the minimum singular value of Vandermonde matrices. When the support consists of multiple well-separated clumps and noise is sufficiently small, the support error by ESPRIT scales like SRF2λ+2×Noise{\rm SRF}^{2\lambda+2} \times {\rm Noise}, where the Super-Resolution Factor (SRF{\rm SRF}) governs the difficulty of the problem and λ\lambda is the cardinality of the largest clump. {If the support contains one clump of closely spaced atoms, the min-max error is SRF2λ+2×Noise/M{\rm SRF}^{2\lambda +2} \times {\rm Noise}/M. Our error bound matches the min-max rate up to a factor of MM in the small noise regime. Our results therefore establishes the near-optimality of ESPRIT,} and our theory is validated by numerical experiments.

Keywords

Cite

@article{arxiv.1905.03782,
  title  = {Super-resolution limit of the ESPRIT algorithm},
  author = {Weilin Li and Wenjing Liao and Albert Fannjiang},
  journal= {arXiv preprint arXiv:1905.03782},
  year   = {2019}
}

Comments

30 pages

R2 v1 2026-06-23T09:02:04.746Z