Greedy Matching Pursuit algorithms¶
Greedy Pursuit algorithms solve an approximate problem
of \(\text{P}_0\) problem of a system of linear equations
where \(k\) is the maximum number of non-zero real-valued coefficients (atoms) of \(\vec{x}\).
Each algorithm returns a Solution which is a namedtuple with the following attributes:
.x - \(\vec{x}_{\text{sparsest}}\) solution to (1)
.support - a list of atoms (non-zero elements of \(\vec{x}\))
.residuals - a list of residual vectors after each iteration
|
Orthogonal Matching Pursuit (OMP) algorithm finds an approximate solution to (1) s.t. |
|
(Weak) Matching Pursuit (MP, WMP) algorithms find an approximate solution to (1). |
|
Thresholding algorithm is the fastest and least accurate among greedy pursuit algorithms of finding an approximate solution to (1). |