On the convergence of an active-set method for ℓ 1 minimization
Document Type
Article
Publication Date
12-1-2012
Abstract
We analyse an abridged version of the active-set algorithm FPC-AS proposed in Wen et al. [A fast algorithm for sparse reconstruction based on shrinkage, subspace optimisation and continuation, SIAM J. Sci. Comput. 32 (2010), pp. 1832-1857] for solving the l 1-regularized problem, i.e. a weighted sum of the l 1-norm |x| 1 and a smooth function f(x). The active-set algorithm alternatively iterates between two stages. In the first non-monotone line search (NMLS) stage, an iterative first-order method based on shrinkage is used to estimate the support at the solution. In the second subspace optimization stage, a smaller smooth problem is solved to recover the magnitudes of the non-zero components of x. We show that NMLS itself is globally convergent and the convergence rate is at least R-linearly. In particular, NMLS is able to identify the zero components of a stationary point after a finite number of steps under some mild conditions. The global convergence of FPC-AS is established based on the properties of NMLS. © 2012 Copyright Taylor and Francis Group, LLC.
Publication Source (Journal or Book title)
Optimization Methods and Software
First Page
1127
Last Page
1146
Recommended Citation
Wen, Z., Yin, W., Zhang, H., & Goldfarb, D. (2012). On the convergence of an active-set method for ℓ 1 minimization. Optimization Methods and Software, 27 (6), 1127-1146. https://doi.org/10.1080/10556788.2011.591398