English

On phase retrieval via matrix completion and the estimation of low rank PSD matrices

Optimization and Control 2020-01-29 v2

Abstract

Given underdetermined measurements of a Positive Semi-Definite (PSD) matrix XX of known low rank KK, we present a new algorithm to estimate XX based on recent advances in non-convex optimization schemes. We apply this in particular to the phase retrieval problem for Fourier data, which can be formulated as a rank 1 PSD matrix recovery problem. Moreover, we provide theory for how oversampling affects the stability of the lifted inverse problem.

Keywords

Cite

@article{arxiv.1907.09537,
  title  = {On phase retrieval via matrix completion and the estimation of low rank PSD matrices},
  author = {Marcus Carlsson and Daniele Gerosa},
  journal= {arXiv preprint arXiv:1907.09537},
  year   = {2020}
}