Fig. 9.

Left and mid: Comparison of lower bounds of $\mathrm{\Delta }$- and $\mathrm{\nabla }$-strategy resp. in setting of Fig. 7 (right)—red: Theorem 2.15, black: Corollary 3.8 maximized over 1000 random choices of the $\mu$-parameters, blue: (Hautphenne and Latouche 2016, Theorem 3), $\mathrm{\Delta }$ and $\mathrm{\nabla }$: approximation via simulation. On the right: Comparison of upper bounds—black: improved norm bound from Remark 3.7 with respect to ${\Vert ·\Vert }_1$, blue: (Hautphenne and Latouche 2016, Theorem 2), $\mathrm{\Delta }$: approximation via simulation (color figure online)