Phase Retrieval by Alternating Minimization With Random Initialization

DOI10.1109/TIT.2020.2971211MaRDI QIDQ5124448zbMATH OpenOpenAlexFDO

Authors Teng Zhang

Publication date 29 September 2020

Published in IEEE Transactions on Information Theory (Search for Journal in Brave)

Full work available at URL https://arxiv.org/abs/1812.01255

Mathematics Subject Classification ID

Numerical mathematical programming methods (65K05) Signal theory (characterization, reconstruction, filtering, etc.) (94A12)

Abstract: We consider a phase retrieval problem, where the goal is to reconstruct a $n$-dimensional complex vector from its phaseless scalar products with $m$ sensing vectors, independently sampled from complex normal distributions. We show that, with a random initialization, the classical algorithm of alternating minimization succeeds with high probability as $n,mightarrowinfty$ when $m/lo{g}^{3}mgeqM{n}^{3/2}lo{g}^{1/2}n$ for some $M>0$. This is a step toward proving the conjecture in cite{Waldspurger2016}, which conjectures that the algorithm succeeds when $m=O(n)$. The analysis depends on an approach that enables the decoupling of the dependency between the algorithmic iterates and the sensing vectors.

Cited in

(8)

• Gradient descent with random initialization: fast global convergence for nonconvex phase retrieval
• scientific article; zbMATH DE number 7295466 (Why is no real title available?)
• Phase Retrieval Algorithm via Nonconvex Minimization Using a Smoothing Function
• scientific article; zbMATH DE number 7255152 (Why is no real title available?)
• Solving phase retrieval via graph projection splitting
• Convolutional Phase Retrieval via Gradient Descent
• Optimization-Based AMP for Phase Retrieval: The Impact of Initialization and $\ell_{2}$ Regularization
• Near-optimal bounds for generalized orthogonal Procrustes problem via generalized power method

This page was built for publication: Phase Retrieval by Alternating Minimization With Random Initialization

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q5124448)