文件列表:

cs-code\benchmark.m (1562, 2008-10-06)
cs-code\benchmark_plot.m (581, 2008-10-09)
cs-code\demo.m (1416, 2013-07-03)
cs-code\gen_matrix.m (2777, 2008-10-07)
cs-code\gen_signal.m (1012, 2008-10-01)
cs-code\gen_sparse_signal.m (490, 2008-01-16)
cs-code\Images\boat.jpg (27090, 2007-11-24)
cs-code\Images\Thumbs.db (7680, 2014-03-10)
cs-code\image_experiment.asv (3364, 2012-09-15)
cs-code\image_experiment.m (3289, 2008-10-06)
cs-code\image_test.m (858, 2008-10-06)
cs-code\init.m (136, 2008-10-04)
cs-code\load_image.m (920, 2008-01-16)
cs-code\Matrices\At_fw.m (699, 2008-01-04)
cs-code\Matrices\A_fw.m (604, 2008-01-04)
cs-code\Matrices\gen_matrix_countmin.m (1494, 2008-10-06)
cs-code\Matrices\gen_matrix_countmin_implicit_twowise.m (1455, 2008-10-06)
cs-code\Matrices\gen_matrix_countmin_threewise.m (1771, 2008-10-06)
cs-code\Matrices\gen_matrix_countmin_twowise.m (1729, 2008-10-06)
cs-code\Matrices\gen_matrix_fourier.m (445, 2008-09-23)
cs-code\Matrices\gen_matrix_gaussian.m (296, 2008-03-08)
cs-code\Matrices\gen_matrix_hadamard.m (570, 2008-03-08)
cs-code\Matrices\gen_matrix_sparse.m (1187, 2008-10-10)
cs-code\Matrices\gen_matrix_sparseplusminus.m (806, 2008-03-08)
cs-code\Matrices\gen_matrix_std.m (1284, 2008-10-06)
cs-code\Matrices\randint.m (283, 2008-10-01)
cs-code\recovery.asv (5530, 2012-09-15)
cs-code\recovery.m (5607, 2013-10-03)
cs-code\recover_countmin_positive.m (769, 2008-10-01)
cs-code\smp.m (1132, 2008-10-10)
cs-code\sparse_experiments.m (2174, 2008-10-09)
cs-code\sparse_experiments_distributed.m (2038, 2008-10-06)
cs-code\sparse_experiments_distributed_helper.m (577, 2008-10-06)
cs-code\sparse_experiments_plot.m (1224, 2008-09-19)
cs-code\testSign.m (38, 2013-10-03)
cs-code\Util\absvalheap.h (3537, 2009-07-03)
cs-code\Util\binsparsemul.c (1619, 2008-02-22)
cs-code\Util\binsparsemul.dll (20480, 2012-09-14)
cs-code\Util\binsparsemul.mexa64 (11022, 2008-02-22)
... ...

CODE FOR SPARSE RECOVERY EXPERIMENTS Introduction This package contains the framework used to perform sparse recovery experiments like those in papers BI08, BGIKS08, BIR08. The base functions are: gen_matrix.m - generates a given type of measurement matrix gen_signal.m - generates a test signal experiment.m - performs a sparse recovery experiment A small demo of how to use these functions is shown in demo.m. (Note: you might need to compile some of the mex files for your platform, see below; also for the image part of the demo you need to have Matlab's Wavelet Toolbox installed). Important high level programs are: sparse_experiments.m - Used to generate a probability of exact recovery plot. The range of parameters (N, Ms, Ks, attempts) is set from inside this file. sparse_experiments_distributed.m - Matlab DCT cluster version of the above. benchmark.m - Performs a runtime benchmark of a set of algorithms. sparse_experiments_plot.m, benchmark_plot.m - Generate plots from the experiment data. There are also programs to perform image experiments; see load_image.m, image_experiment.m, and image_test.m . All .m files contain documentation, which can be seen by viewing the file or using the Matlab command help (e.g. "help benchmark"). Note: init.m adds relevant subdirectories to the Matlab path. It is ran inside most programs, but it can also be called manually at the start of the Matlab session for safety. Installing solvers for LP experiments To perform LP experiments, l1magic must be installed. Download the l1magic archive from http://www.acm.caltech.edu/l1magic/ . Unpack the archive so that the l1magic directory is at the root of the code installation (i.e. l1magic is a directory along with Matrices, Util, etc.) To use GPSR, the GPSR_BB.m program must be placed in the root directory of the code installation. The GPSR website is http://www.lx.it.pt/~mtf/GPSR/ . Compiling the C programs (MEX) The important (bottleneck) parts are implemented as C programs which Matlab calls as MEX files. These files are in Util. Compiled binaries are included for Windows (32-bit Matlab) and Linux (***-bit). For other platforms, one needs to compile them by running "mex ". The mex script comes with Matlab (it should be in the run path). The Util directory contains compile.bat and compile.sh scripts to call mex on all the c files there. Authors Radu Berinde, MIT, texel@mit.edu Piotr Indyk, MIT, indyk@mit.edu References BI08 R. Berinde and P. Indyk, Sparse recovery using sparse random matrices. ''MIT-CSAIL Technical Report'', 2008. BGIKS08 R. Berinde, A. Gilbert, P. Indyk, H. Karloff, and M. Strauss. Combining geometry and combinatorics: a unified approach to sparse signal recovery, ''Allerton'', 2008. BIR08 R. Berinde, P.Indyk, and M. Ruic. Practical Near-optimal Sparse Recovery in the L1 Norm, ''Allerton'', 2008

近期下载者：

相关文件：

评论：[我要评论] [举报此文件]

收藏者：