job_adpWBPDN

 Comparison of various solvers for adaptive reweighting
 with largescale examples

 Solves the following basis pursuit denoising (BPDN) problem
 min_x  \Sum \w_i |x_i| + 1/2*||y-Ax||_2^2

 while adaptively selecting the weights w_i

 Written by: Salman Asif, Georgia Tech
 Email: sasif@gatech.edu
 Created: June 16, 2011

 Reference: 
 "Fast and accurate algorithms for re-weighted L1 norm minimization," by 
 M. Salman Asif and Justin Romberg
 
 To reproduce experiments in the paper, use this scrip with 
 
 rseed = 2012;
 rand('state',rseed);
 randn('state',rseed);
  
 before running each simulation in 
 script_simulation_adpWBPDN

0001 % Comparison of various solvers for adaptive reweighting
0002 % with largescale examples
0003 %
0004 % Solves the following basis pursuit denoising (BPDN) problem
0005 % min_x  \Sum \w_i |x_i| + 1/2*||y-Ax||_2^2
0006 %
0007 % while adaptively selecting the weights w_i
0008 %
0009 % Written by: Salman Asif, Georgia Tech
0010 % Email: sasif@gatech.edu
0011 % Created: June 16, 2011
0012 %
0013 % Reference:
0014 % "Fast and accurate algorithms for re-weighted L1 norm minimization," by
0015 % M. Salman Asif and Justin Romberg
0016 %
0017 % To reproduce experiments in the paper, use this scrip with
0018 %
0019 % rseed = 2012;
0020 % rand('state',rseed);
0021 % randn('state',rseed);
0022 %
0023 % before running each simulation in
0024 % script_simulation_adpWBPDN
0025 
0026 clear
0027 % close all force
0028 
0029 %% parfor setup
0030 % numCores = str2double(getenv('NUMBER_OF_PROCESSORS'));
0031 % mpSize = numCores-1;
0032 % if matlabpool('size') ~= mpSize
0033 %     if matlabpool('size')~=0
0034 %         matlabpool close;
0035 %     else
0036 %         matlabpool('open', mpSize);
0037 %     end
0038 % end
0039 
0040 %% Setup path
0041 mname = mfilename;
0042 mpath = mfilename('fullpath');
0043 mdir = mpath(1:end-length(mname));
0044 cd(mdir);
0045 
0046 addpath ../Pursuits_Homotopy/
0047 addpath ../utils/
0048 addpath ../utils/utils_Wavelet/
0049 addpath ../utils/utils_meas/
0050 addpath ../solvers/
0051 addpath src/
0052 
0053 disp(['--------------------',datestr(now),'-------------------------'])
0054 
0055 % simulation parameters
0056 rwt_mode = 5;
0057 lambda = 0;
0058 
0059 SAVE_RESULTS = false; 
0060 
0061 % SAVE_RESULTS = true;  diary(sprintf('%s-reproduce.txt',mname));
0062 
0063 largescale = 1;
0064 
0065 for SNR = [40]
0066     for M = [25000 30000] % [2 3]
0067         IMG_LIST = {'barbara','boats', 'cameraman','house','peppers','shapes','lena','airplane','baboon','sailboat','tiffany'};
0068         for img = 1:5; % length(IMG_LIST)
0069             sType = IMG_LIST{img};
0070             
0071             if largescale
0072                 mType = 'noiselets';
0073                 N = (256)^2;
0074                 % M = round(N/R);
0075                 T = N;
0076                 str0 = sprintf('mType-%s, sType-%s, SNR = %d, (N,M) = %d, %d, rwt_mode-%d, lambda%3.4g.', mType, sType, SNR, N, M, rwt_mode, lambda);
0077                 disp(str0);
0078             else
0079                 mType = 'randn';
0080                 sType = 'HeaviSine';
0081                 N = 512;   % signal length
0082                 M = round(N/2);    % no. of measurements
0083                 T = round(M/3);    % sparsity level
0084                 str0 = sprintf('mType-%s, sType-%s, SNR = %d, (N,M,T) = %d, %d, %d, rwt_mode-%d, lambda%3.4g.', mType, sType, SNR, N, M, T, rwt_mode, lambda);
0085                 disp(str0);
0086             end
0087             
0088             % rank-1 update mode
0089             delx_mode = 'mil'; % mil or qr
0090             
0091             %% Simulation
0092             maxsim = 10;
0093             script_simulation_adpWBPDN            
0094             
0095         end
0096     end
0097 end
0098 diary off;