OMP-variations

This repo contains variations of Orthogonal Matching Pursuit (OMP) Algorithm family.

OMP_MMV (OMP with Multiple Measurement Vector):

A fast implementation of OMP with Multiple Measurement Vector (to recover jointly row-sparse matrix). However, it works for sparse vectors as well

POMP (Piecewise Orthogonal Matching Pursuit):

This variation of OMP algorithm finds piecewise sparse vector. Especially handy when you find indices of joint codebooks/sensing matrix. Details here: https://ieeexplore.ieee.org/document/7472550

DGOMP (Detector based OMP):

DGOMP is the variation of OMP which doesn't require sparsity (number of non zero elements in the vector) as a prior. It performs statistical hypothesis testing on each iteration on the residual to determine algorithm stopping criterion. Details here: https://link.springer.com/article/10.1186/1687-6180-2014-178

BSBL (Block Sparse Bayesial Learning):

Not necessarily a matching pursuit based sparse recovery algorithm, however, it is a faster approximation of a block-sparse recovery algorithm by Block Sparse Bayesian Learning.