Once you get the hang of it, turboGAMP is quite a simple algorithm to use, making it super-easy to quickly try a variety of different probabilistic models of signal structure with just a few keystrokes. Here we provide links to several documents that can assist you in understanding how EMturboGAMP (the MATLAB software package) works. The code itself is extensively documented and well-commented, and provides a highly detailed description of each file.

  • Paper: "A Generalized Framework for Learning and Recovery of Structured Sparse Signals," J. Ziniel, S. Rangan, and P. Schniter, Statistical Signal Processing Workshop, (Ann Arbor, MI), Aug. 2012.
    • Describes the mathematics underpinning the turboGAMP framework
    • Outlines the MATLAB software implementation
  • Class Diagram
    • A diagram that describes the relationships between the various classes that make up the bulk of the EMturboGAMP package
  • Background: Since turboGAMP is an extension of GAMP, it is helpful to understand how the gampmatlab code works. A users' guide for gampmatlab can be found here.