Fig. 5

We consider \(m=100\), and we let \(\mathbf{G} ,\mathbf{H} \in \mathcal {M}^{10}_{100}\) (i.e., \(c=10\)) (left panel). We also implement the Naive Padding scheme, and we let \(\mathbf{G} \in \mathcal {M}^{10}_{100}\), \(\mathbf{H} \in \mathcal {M}^{10}_{500}\) (right panel). Considering \(\rho\) to take two possible values: \(\rho =r\) for \(c_g\) channels, or \(\rho =-r\) for \(c_b=c-c_g\) channels, where r varies in \(\{0.1,0.2,0.3,0.4,0.5\}\), we plot the matching accuracy obtained by M-FAQ with 10 seeds averaged over 2000 (left panel) and 100 (right panel) Monte Carlo replicates (with 10 seeds) versus \(c_b\). In red (resp., olive, green, blue, purple) we plot the results for \(r=0.1\) (resp., \(r=0.2\), \(r=0.3\), \(r=0.4\), \(r=0.5\)). The partially transparent points visualize the accuracy distribution and correspond to individual Monte Carlo replicates