1: \begin{abstract}

2: In this article we propose an algorithm, \textsc{nesta}-\textsc{lasso}, for the \textsc{lasso} problem (i.e., an underdetermined linear least-squares problem with a one-norm constraint on the solution) that exhibits linear convergence under the \textit{restricted isometry property} (\textsc{rip}) and some other reasonable assumptions. Inspired by the state-of-the-art sparse recovery method, \textsc{nesta}, we rely on an accelerated proximal gradient method proposed by Nesterov in 1983 that takes $O(\sqrt{1/\varepsilon})$ iterations to come within $\varepsilon > 0$ of the optimal value. We introduce a modification to Nesterov's method that regularly updates the prox-center in a provably optimal manner, resulting in the linear convergence of \textsc{nesta}-\textsc{lasso} under reasonable assumptions.

3: 

4: Our work is motivated by recent advances in solving the basis pursuit denoising (\textsc{bpdn}) problem (i.e., approximating the minimum one-norm solution to an underdetermined least squares problem). Thus, one of our goals is to show that \textsc{nesta}-\textsc{lasso} can be used to solve the \textsc{bpdn} problem. We use \textsc{nesta}-\textsc{lasso} to solve a subproblem within the Pareto root-finding method used by the state-of-the-art \textsc{bpdn} solver \textsc{spgl1}. The resulting algorithm is called \textsc{parnes}, and we show, experimentally, that it is comparable to currently available solvers.  

5: 

6: %Insert your abstract here. Include keywords, PACS and mathematical

7: %subject classification numbers as needed.

8: \keywords{basis pursuit \and Newton's method \and Pareto curve \and Nesterov's method \and compressed sensing \and convex minimization \and duality}

9: % \PACS{PACS code1 \and PACS code2 \and more}

10: % \subclass{MSC code1 \and MSC code2 \and more}

11: \end{abstract}

12: