Figure 1 | Scientific Reports

Figure 1

From: An adaptive shortest-solution guided decimation approach to sparse high-dimensional linear regression

Figure 1

Estimated guidance vector \(\hat{\varvec{\gamma }}\) on an uncorrelated Gaussian measurement matrix with \(p = 1000\) and \(n = 200\). Each nonzero coefficient is uniform distributed in [0.5, 1]. Top: rank curves for the estimated guidance vector. Bottom: proportion q(r) of nonzero elements of \(\varvec{\beta }^0\) among the r top-ranked indices i. In the two subfigures on the left panel, \(\sigma ^2=1\) and \(s_0=10,20,30\); and in the two subfigures on the right panel, \(s_0=20\) and \(\sigma ^2=0,0.5,1\).

Back to article page