Sparse Recovery Analysis of Generalized $J$-Minimization with Results for Sparsity Promoting Functions with Monotonic Elasticity

TL;DR

This paper provides a theoretical framework for exact sparse signal recovery using generalized nonconvex minimization, introducing the scale function concept and deriving bounds on recovery conditions, with numerical validation.

Contribution

It generalizes analysis techniques for sparsity-promoting functions via the scale function and derives explicit bounds for exact recovery conditions in sparse signal reconstruction.

Findings

Derived bounds on null space constant and restricted isometry constant

Established sufficient conditions for exact recovery

Numerical simulations confirm theoretical bounds and compare sparsity functions

Abstract

In this paper we theoretically study exact recovery of sparse vectors from compressed measurements by minimizing a general nonconvex function that can be decomposed into the sum of single variable functions belonging to a class of smooth nonconvex sparsity promoting functions. Null space property (NSP) and restricted isometry property (RIP) are used as key theoretical tools. The notion of \emph{scale function} associated to a sparsity promoting function is introduced to generalize the state-of-the-art analysis technique of the $l_p$ minimization problem. The analysis is used to derive an upper bound on the null space constant (NSC) associated to this general nonconvex minimization problem, which is further utilized to derive sufficient conditions for exact recovery as upper bounds on the restricted isometry constant (RIC), as well as bounds on optimal sparsity $K$ for which exact recovery occurs. The derived bounds are explicitly calculated when the sparsity promoting function $f$ under consideration possesses the property that the associated \emph{elasticity function}, defined as, $\psi(x)=\frac{xdf(x)/dx}{f(x)}$, is monotonic in nature. Numerical simulations are carried out to verify the efficacy of the bounds and interesting conclusions are drawn about the comparative performances of different sparsity promoting functions for the problem of $1$-sparse signal recovery.