Spectral proximal method for solving large scale sparse optimization

In this paper, we propose to use spectral proximal method solve sparse optimization problems. Sparse refers an problem involving the ι 0 -norm in objective or constraints. The previous research showed that gradient is outperformed other standard unconstrained methods. This due replaced full rank matrix by a diagonal and memory decreased from Ο(n 2 ) Ο(n). Since term nonconvex non-smooth, it cannot be solved algorithm. We will with underdetermined system as its constraint considered. Using Lagrange method, transformed into problem. A new called proposed, which combination of method. then applied programming code written Python compare efficiency proposed some existing benchmarks comparison are based on number iterations, functions call computational time. Theoretically, requires less storage