June, 2011

This subdirectory contains the following Matlab source codes:

cond_adm_bd.m         The main code that minimizes condition number using ADM.

update_X_adm_bd.m
update_Yt_adm_bd.m    These codes solve the ADM subproblems by spectral projected gradient method.

fval_cond_bd.m        This code computes the function value for update_Yt_adm_bd.m.

grad_cond_bd.m        This code computes the gradient for update_Yt_adm_bd.m.

simplex_proj.m        This code computes the projection onto simplex.

cond_se_bd_dual2.m     This code writes input to SeDuMi for minimizing condition number.

cond_test_bd2.m        This is the run code.

Implementation and numerical experience with the above codes are described in the paper: 
    Zhaosong Lu and Ting Kei Pong
    "Minimizing Condition Number via Convex Programming",
    Submitted.
This code was last updated on June 22, 2011.

Questions/comments/suggestions about the codes are welcome.  

Zhaosong Lu,   zhaosong@sfu.ca
Ting Kei Pong, tkpong@uw.edu