Implement simulations to have plots of regret w.r.t a parameter

[Naereen]

• In this recent article ["SIC-MMAB: Synchronisation Involves Communication in Multiplayer Multi-Armed Bandits", by Etienne Boursier, Vianney Perchet, arXiv 1809.08151, 2018], we can see in Figure 3 a plot of $\log(R_T)$ w.r.t. $\log(1/\Delta)$.

• Similarly, in this recent article [“Nearly Optimal Adaptive Procedure for Piecewise-Stationary Bandit: a Change-Point Detection Approach”. Yang Cao, Zheng Wen, Branislav Kveton, Yao Xie. arXiv preprint arXiv:1802.03692, 2018] (see this page), we can see in Figure 1 and 2 a plot of $R_T / \sqrt{T}$ w.r.t. $K$ or $\Upsilon_T$ (for non stationary bandits)

I like this idea!

I want the same!

• TODO: think about this, and find a nice way of doing it, for basic stuff.