Improved Bounds on Restricted Isometry Constants for Gaussian Matrices

Abstract

The Restricted Isometry Constants (RIC) of a matrix $A$ measures how close to an isometry is the action of $A$ on vectors with few nonzero entries, measured in the ${\ell }^2$ norm. Specifically, the upper and lower RIC of a matrix $A$ of size $n\times N$ is the maximum and the minimum deviation from unity (one) of the largest and smallest, respectively, square of singular values of all $\left(\genfrac{}{}{0ex}{}{N}{k}\right)$ matrices formed by taking $k$ columns from $A$. Calculation of the RIC is intractable for most matrices due to its combinatorial nature; however, many random matrices typically have bounded RIC in some range of problem sizes $(k,n,N)$. We provide the best known bound on the RIC for Gaussian matrices, which is also the smallest known bound on the RIC for any large rectangular matrix. Improvements over prior bounds are achieved by exploiting similarity of singular values for matrices which share a substantial number of columns.

16 pages, 8 figures

References in corpus (3)

• Generalized power method for sparse principal component analysis
• Subspace Pursuit for Compressive Sensing Signal Reconstruction
• Phase Transitions for Greedy Sparse Approximation Algorithms

Cited by in corpus (4)

• Proof of Convergence and Performance Analysis for Sparse Recovery via Zero-point Attracting Projection
• Restricted Isometry Property in Quantized Network Coding of Sparse Messages
• Deterministic Bounds for Restricted Isometry of Compressed Sensing Matrices
• Recovering Compressively Sampled Signals Using Partial Support Information