Title:Precise Performance Analysis of the LASSO under Matrix Uncertainties

Abstract:In this paper, we consider the problem of recovering an unknown sparse signal $\xv_0 \in \mathbb{R}^n$ from noisy linear measurements $\yv = \Hm \xv_0+ \zv \in \mathbb{R}^m$. A popular approach is to solve the $\ell_1$-norm regularized least squares problem which is known as the LASSO. In many practical situations, the measurement matrix $\Hm$ is not perfectely known and we only have a noisy version of it. We assume that the entries of the measurement matrix $\Hm$ and of the noise vector $\zv$ are iid Gaussian with zero mean and variances $1/n$ and $\sigma_{\zv}^2$. In this work, an imperfect measurement matrix is considered under which we precisely characterize the limiting behavior of the mean squared error and the probability of support recovery of the LASSO. The analysis is performed when the problem dimensions grow simultaneously to infinity at fixed rates. Numerical simulations validate the theoretical predictions derived in this paper.