Figure 6

Simulation results on a correlated Gaussian measurement matrix (\(\pi =0.7\), \(n=300\), \(p=2000\), \(\sigma ^{2}=0.5\)): influence of the sparsity level \(s_0\) on relative errors (top left), true positives (top right), false positives (bottom left) with the inset being a semi-logarithmic plot, and probability of exact identification of nonzero coefficients (bottom right).