Matlab software implementing the algorithms described in these papers:

[1] W. Shi, G. Wahba, S. J. Wright, K. Lee, R. Klein, and B. Klein, "LASSO-Patternsearch Algorithm with Application to Ophthalmology and Genomic Data," Statistics and its Interface 1 (2008), pp. 137-153.

[2] S. J. Wright, "Accelerated Block-Coordinate Relaxation for Regularized Optimization," Technical Report, August 2010. Revised September 2011.

The code does l1-regularized linear logistic regression with on data with Bernoulli outcomes (indicated by +/-1). The algorithm uses a gradient projection / iterative shrinkage approach, with gradient sampling and a modified reduced Newton scaling technique on the space of nonzero variables.

The codes were written initially by W. Shi and S. Wright in 2006-2008 and rewritten for distribution in 2008-2011 by S. Wright.

Updated 9/1/2011

Code and Small Test Data Sets

These zip and tarballs contain Matlab code in subdirectory "code" and test data sets in subdirectory "data." See README.txt in the main directory for a short description of contents.

Larger Test Data Sets

Unzip, and edit the calling programs (TestBig or TestTables in LPS v2.1) to point to the location of the file on your system.


See the log of changes for notes on the various releases.


This research was supported by National Science Foundation Grants DMS-0505636, DMS-0604572, SCI-0330538, DMS-0427689, CCF-0430504, DMS-0914524, and DMS-0906818; NIH Grant EY09946; and DOE Grant DE-FG02-04ER25627. Any opinions, findings, and conclusions or recommendations expressed in this material are those of the authors and do not necessarily reflect the views of the National Science Foundation.