Skip to content

Matlab code for the reproduction of certain results from two papers by Jeremy Vila and Philip Schniter

Notifications You must be signed in to change notification settings

ThomasA/reprod-vila-gamp

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

15 Commits
 
 
 
 
 
 

Repository files navigation

reprod-vila-gamp

This repository contains Matlab code for the reproduction of certain results from the following paper by Jeremy Vila and Philip Schniter:

  1. J. P. Vila and P. Schniter, "Expectation-Maximization Bernoulli-Gaussian Approximate Message Passing," Proc. Asilomar Conf. on Signals, Systems, and Computers (Pacific Grove, CA), Nov. 2011.

Specifically, the code seeks to reproduce Figure 1 from this paper.

Dependencies

In order to run this code, you must have the gampmatlab toolbox for Matlab available (as well as Matlab). Matlab must be set up to be able to find the following folders from this toolbox: trunk/code/main/, trunk/code/EMGMAMP, trunk/code/EMGMAMPnew/.

The results stored in this repository were produced using revision 562 of the 'gampmatlab' toolbox from the (Subversion) repository svn://svn.code.sf.net/p/gampmatlab/code.

Variants

The folder reprod-vila2011 contains the main files of the project. Files are provided for a total of four combinations of two features of the algorithms: There is an 'old' and a 'new' version of the EMGMAMP algorithm in the 'gampmatlab' toolbox, trunk/code/EMGMAMP and trunk/code/EMGMAMPnew/ respectively. Furthermore, the algorithms have been run with and without the uniformVariance option. These four combinations have been configured in these four files in the folder reprod-vila2011:

  • phasetrans_new_not_unif.m
  • phasetrans_new_unif.m
  • phasetrans_old_not_unif.m
  • phasetrans_old_unif.m

These four files each contain the main algorithm and the simulation framework set up around it to estimate its reconstruction capability phase transition through Monte Carlo simulations over the undersampling/sparsity parameter plane. The file 'regression_plot.m' is a helper file for generating plots of the simulation results. Running any of the above scripts will automatically call 'regression_plot.m' to plot and store figures and append the estimated phase transition curves to the result .MAT files.

Generating main results

For convenience, the script 'run_main.m' in the root folder can be called to automatically run the main simulations from this repository (i.e. 'phasetrans_old_unif.m') in reprod-vila2011 and copy the results to the root folder.

About

Matlab code for the reproduction of certain results from two papers by Jeremy Vila and Philip Schniter

Resources

Stars

Watchers

Forks

Packages

No packages published

Languages