YALL1: Your ALgorithms for L1

Yin Zhang
Department of CAAM
Rice University, Houston, Texas 77005

(Number of access to this page is since 05/13/2009)


This Matlab code can currently be applied to the following eight (8) L1-minimization problems:
   (BP)          min ||Wx||w,1 s.t. Ax = b
   (L1/L1)       min ||Wx||w,1 + (1/&nu)||Ax - b||1
   (L1/L2)       min ||Wx||w,1 + (1/2&rho)||Ax - b||22
   (L1/L2con)    min ||Wx||w,1, s.t. ||Ax - b||2 <= &delta
   (BP+)         min ||x||w,1 s.t. Ax = b          and  x >= 0
   (L1/L1+)      min ||x||w,1 + (1/&nu)||Ax - b||1   s.t. x >= 0
   (L1/L2+)      min ||x||w,1 + (1/2&rho)||Ax - b||22 s.t. x >= 0
   (L1/L2con+)   min ||x||w,1,  s.t. ||Ax - b||2 <= &delta,  x >= 0
where A is m by n with m less than n, and the solution x (or its representation Wx) is supposed to be (approximately) sparse. The data (A,b) can be real or complex, and the signal x can also be complex in the cases of no nonnegativity constraint. A unitary sparsifying basis W is allowed and the 1-norm for x (or Wx) can be weighted by a vector w >= 0.

The capacity for solving models (L1/L2con) and (L1/L2con+) has been added to version beta-6 for A*A' = I.

News:

Downloading:


Acknowledgment: The author's work was supported in part by NSF DMS-0811188 and ONR N00014-08-1-1101.

YALL1 (YALL-one) Copyright (C) 2009 Yin Zhang