Figure 2:

The top panel shows the distribution of the difference between the MASE estimates and true eigenvalues of a single graph. The graphs in the sample are distributed according to a two-block multilayer SBM with n = 300 vertices, block connection probability matrix B⁽ⁱ⁾ = 0.3I + 0.111^⊤. The distributions appear to be Gaussian, and as the sample size m increases, the mean of the distribution (red solid line) approaches to zero (blue dashed line). This phenomenon is also observed in the bottom panel, which shows that the estimation bias decreases with the sample size.