1: \begin{abstract}

2: On \cite{NeVe1} some consequences of the Restricted Isometry

3: Property (RIP) of matrices have been applied to develop a greedy

4: algorithm called ``ROMP" (Regularized Orthogonal Matching Pursuit)

5: to recover sparse signals and to approximate non-sparse ones. These

6: consequences were subsequently applied to other greedy and

7: thresholding algorithms like ``SThresh", ``CoSaMP", ``StOMP" and

8: ``SWCGP". In this paper, we find another consequence of the RIP

9: property and use it to analyze the approximation to $k$-sparse

10: signals with \textit{Stagewise Weak} versions of Gradient Pursuit

11: (SWGP),

12:  Matching Pursuit (SWMP) and Orthogonal Matching Pursuit

13: (SWOMP) algorithms described in in \cite{BlDa2}.

14: %obtaining better convergence rates for this kind of matrices than those

15: %already known for moderate values of $k$.

16: %originally obtained (\cite{BlDa2},\cite{CoDaDe2}, \cite{GrVa}).

17: We combine the above mentioned

18: algorithms with another selection rule similar to the ones that appeared in \cite{CoDaDe2}  and  \cite{DTDS}

19: showing that results are obtained with less restrictions in the RIP constant, but we need a smaller threshold parameter for

20: the coefficients. The results of some experiments are shown.

21: % A new selection rule,

22: %called \textit{Relaxed}, is given and applied to each of the algorithms

23: %above. This small change also relaxes conditions (like the

24: %Restricted Isometry Constant and the number of measurements)

25: %demanded by other algorithms and improve the convergence rate for a

26: %given approximation/estimation error.

27: 

28: \end{abstract}

29: