Figure 4

Discrepancies between input and average output degree distributions (left panels) and average transitivity values (right panels) for an ensemble of 15 Poisson (top panels), exponential (middle panels) and scale-free graphs (bottom panels) as generated by our algorithm and the algorithms presented in [30]and [20]. Each graph has N = 500 and mean degree, ⟨d⟩ = 5. In the left graphs, the input degree distribution is shown as a black circles; and output degree distributions are shown for the Newman (green dashed line) and the Volz (gray dashed line) algorithms. Output degree distributions are not shown for ClustRNet as the degree sequence always perfectly match the input. In the right graphs, the input is shown as black circles, and output transitivity values are shown for two runs: (1) using SV-transitivity (()) as the clustering measure in ClustRNet (blue line), and (2) ClustRNet [without a connectivity constraint] (orange line), the Newman algorithm (green dashed line) and the Volz algorithm (gray dashed line), all with transitivity (()) as the clustering measure.