Fig. 4From: Improved analysis of SP and CoSaMP under total perturbationsRelative approximation error vs. number of measurements in the SP and CoSaMP algorithms for Sparse matrixBack to article page