Optimal Greedy Sparse Solver

There are 10s of greedy algorithms for solving y=Ax where x is sparse. But what do they actually solve?

OMP, StOMP, SWOMP, ROMP, CoSAMP, Bayesian OMP, OLS, etc. are few of the popular solvers for sparse recovery: y=Ax where x is sparse.

In this work, we re-engineer our way back from the algorithms to a unifying objective. And then directly solve the objective. This is more optimal than all greedy algorithms in a sense.