Figure 8

The empirical runtime of the simulator using different index types. The three-letter acronyms in the legend indicate the preference function type (‘l’ for linear), the sort key (‘s’ for subtree mass or ‘n’ for node mass), and the definition of priority (‘n’ for node mass or ‘u’ for uniform) for different variants of the treap structure. Twenty networks were generated for each configuration and size. Error bars, barely visible, indicate the 95% confidence interval.