All-or-nothing statistical and computational phase transitions in sparse spiked matrix estimation

Abstract

We determine statistical and computational limits for estimation of a rank-one matrix (the spike) corrupted by an additive gaussian noise matrix, in a sparse limit, where the underlying hidden vector (that constructs the rank-one matrix) has a number of non-zero components that scales sub-linearly with the total dimension of the vector, and the signal-to-noise ratio tends to infinity at an appropriate speed. We prove explicit low-dimensional variational formulas for the asymptotic mutual information between the spike and the observed noisy matrix and analyze the approximate message passing algorithm in the sparse regime. For Bernoulli and Bernoulli-Rademacher distributed vectors, and when the sparsity and signal strength satisfy an appropriate scaling relation, we find all-or-nothing phase transitions for the asymptotic minimum and algorithmic mean-square errors. These jump from their maximum possible value to zero, at well defined signal-to-noise thresholds whose asymptotic values we determine exactly. In the asymptotic regime the statistical-to-algorithmic gap diverges indicating that sparse recovery is hard for approximate message passing.

arXiv commentPart of this work (in particular the proof of Theorem 1) is already present in reference arXiv:1911.05030

References in corpus (12)

• Message Passing Algorithms for Compressed Sensing
• Optimal Errors and Phase Transitions in High-Dimensional Generalized Linear Models
• The largest eigenvalue of rank one deformation of large Wigner matrices
• Mutual information for symmetric rank-one matrix estimation: A proof of the replica formula
• Mutual Information in Rank-One Matrix Estimation
• The Mutual Information in Random Linear Estimation
• Finding One Community in a Sparse Graph
• A note on classical ground state energies
• Sum-of-Squares Lower Bounds for Sparse PCA
• Reducibility and Computational Lower Bounds for Problems with Planted Sparse Structure
• Finite Size Corrections and Likelihood Ratio Fluctuations in the Spiked Wigner Model
• Detection limits in the high-dimensional spiked rectangular model

Cited by in corpus (4)

• Computational Barriers to Estimation from Low-Degree Polynomials
• The Overlap Gap Property in Principal Submatrix Recovery
• Strong replica symmetry for high-dimensional disordered log-concave Gibbs measures
• The All-or-Nothing Phenomenon in Sparse Tensor PCA